Future Organizational Design by Lars Groth - Creative Commons Edition 1.1

FUTURE
ORGANIZATIONAL
DESIGN
The Scope for the
IT-based Enterprise
LARS GROTH
Foreword by HENRY MINTZBERG
As originally published by John Wiley & Sons, Ltd.
Electronic Edition v. 1.1

ii
In order to view this electronic edition with correct left-right

placement of pages on-screen, please adjust your Acrobat reader
to “Two Page View” and “Show Cover Page in Two Page View”
in the Page Display submenu of the View menu.
iii
Future Organizational Design
by Lars Groth
is licensed under a Creative Commons

Attribution-NonCommercial-NoDerivs 3.0
Unported License.
This means that:

• You are free to share this work — to copy, distribute and transmit it,
• under the following conditions:
o Attribution — You must attribute this work to the author in a definitive and easily
recognizable way (but not in any way that suggests that he endorses you or your use of
the work).
o Noncommercial — You may not use this work for commercial purposes without a
written permission from the author.
o No Derivative Works — You may not alter, transform, or build upon this work.
For the full license, please go to http://creativecommons.org/licenses/by-nc-

nd/3.0/legalcode.
Waiver: You are expressly permitted to use this work for educational purposes
even when the educational activities have a commercial aspect, i.e. when a fee
is charged. However, you are not permitted to charge a fee specificially for this
work without a written permission from the author.
iv
This is an electronic edition of the book
“Future Organizational Design

The Scope for the IT-based Enterprise”
by Lars Groth
This work was originally published by John Wiley & Sons Ltd. in 1999
(ISBN:0-471-98893-6). The rights reverted to the author in 2010, after the
original print run was sold out. This electronic edition has been created on the
basis of the author’s original manuscript, and has been corrected to include
changes made during final proofreading for the printed edition. In addition, a
few typographical errors in the printed edition have been corrected.
The present electronic edition contains all the text in the printed edition,
including Foreword, Preface, Series Preface, Acknowledgments, Author Index
and Subject Index. The pagination is identical to the printed version.
Accordingly, any citation made from the electronic edition will be identical to
the corresponding citation from the printed edition.
The author can be contacted at [email protected].

v
“Man is not the sum of what he has already,

but rather the sum of what he does not yet
have, of what he could have.”
Jean-Paul Sartre, Temporalité, in Situations (1947-49)
vi
vii
Contents
FOREWORD XV
PREFACE XVII
SERIES PREFACE XXI
THE PRESENT VOLUME XXI
Editors xxiii
Advisory Board xxiii
ACKNOWLEDGEMENTS XXV
I A PLATFORM FOR THE INVESTIGATION 1

1 RECOURSE TO REASON 3
SUSTAINED SUCCESS GROWS FROM KNOWING WHY 3
A Quest for Practical Directions 3
Venturing Beyond Prescriptions 5
Organization—A Human Endeavor 9
The Dawn 9
Modern Times 10
The Contributions of Information Technology 12
Model-driven Organizations 15
Configuring Mintzberg for Computers 15
The Joystick Organization 16
The Flexible Bureaucracy 17
The Interactive Adhocracy 17
The Meta-Organization 18
The Organized Cloud 18
From Analysis to Action 19
A KEY TO THIS BOOK 21
Part I: A Platform for the Investigation 21
Part II: Individual Capacity and Organization before the Computer 22
Part III: IT and the Preconditions for Organizing 22
Part IV: Extending the Space of Constructible Organizations 23
Part V: The New Organizations 23
2 ORGANIZATION AND TOOLS—THE HUMAN ADVANTAGES 25
A CRUCIAL LINK 25
To Be Human Is to Be Organized ... 25
viii
... and to Use Tools 27

THE POINT OF LEVERAGE 27
Organizations Are Constructed 28
But They Are also Systems 30
And Contingencies Matter 32
The Space of Constructible Organizations 33
Defining the Boundaries of Constructible Space 34
The Scope of This Investigation 36
Our Biological Characteristics 36
Our Psychological Characteristics 37
Social and Cultural Factors 37
Tools and Methods 38
3 THE BASIC PRECONDITIONS FOR ORGANIZING 39

THE ESSENCE OF ORGANIZATION 39
When the Task Becomes Too Large for One 39
BASIC ELEMENTS IN ORGANIZATION STRUCTURING 42
The Division of Labor and Structuring of Work 42
Coordination 43
The Linchpin of Organization 45
A Taxonomy of Coordinating Mechanisms 47
Two Classes of Coordinating Mechanisms 50
Real-Time Mechanisms 50
Programmed Mechanisms 52
THE BASIC PRECONDITIONS FOR ORGANIZING 53
II INDIVIDUAL CAPACITY AND ORGANIZATION BEFORE THE

COMPUTER 57
4 CONFINED BY PHYSIOLOGY 59
ONE THING AT A TIME 59
MEMORY 61
INFORMATION PROCESSING 65
Elements in Problem Solving 66
Procuring Information 66
Understanding the Problem 67
Generating Hypotheses About Solutions 68
Testing and Evaluating the Solutions 69
From Maximizing to Satisficing: Accepting Simplification 69
Unconscious Processing and Intuition 71
The Delays of Deliberation 73
OUR COMMUNICATION BOTTLENECK 74
THE CONSTRAINTS OF SPACE AND TIME 76
WISHING, WANTING, AND FEELING 78
COPING WITH REALITY 82
Imitation 82
Mental Sets 83
The Constraints of Sets 85
5 THE DAWN OF ORGANIZATION 89
EVOLVING FROM THE PRIMATE STAGE 89
Present-Day Hunter/Gatherers 90
ix
Domestication 93
THEORY FOR SIMPLE ORGANIZATION 95
The Problems of Organization Building in Preliterate Society 99
The Organization Domains and Their Structuring 100
Circumventing the Barrier of Cognitive Capacity 101
The Feudal Type Organization 103
Military Organization 105
The Basic Principles of Preliterate Organization 106
6 THE POWER OF TECHNOLOGY 109

THE NATURE OF TOOLS 109
The Breakthroughs 114
THE EXTERNALIZATION OF MEMORY 114
The Struggle to Remember 114
The Art of Writing—an Administrative Technology 117
The Significance of a Shared Memory 118
The Importance of Numerals 119
TECHNOLOGY TAKES OFF 120
Printing and Mass Literacy 120
Organization of Records 122
Communications Revolution 123
Couriers and Mail Services 123
Railroads, Telegraphs, Telephones and More 125
THE TECHNOLOGY-AUGMENTED PRECONDITIONS 127
Memory 127
Information Processing 129
The Development of an Analytical, Literate Mind 132
Communication 136
The Iron Constraint on Information Exchange 139
Serial Mind, Parallel Action 141
Emotions 142
CONCLUSIONS 143
7 THE MODERN ORGANIZATION 147
INTO THE MODERN AGE 147
The Growth of Complexity 149
The Starting Point 149
Scaling Efforts 149
The Birth of the Machine Organization 152
The New Needs 152
The Transition to a New Organizational Form 154
The Limits of Monolithic Bureaucracy 157
A NEW CONCEPT FOR COORDINATION 162
The Bureaucratic Advantage 162
Explicit Design and The Emergence of the Conceptual Model 167
The Constraints of Standardization 170
CULTURE REVISITED 171
III IT AND THE PRECONDITIONS FOR ORGANIZING 177

8 INFORMATION TECHNOLOGY CHARACTERISTICS 181
NEEDED: A REALISTIC ASSESSMENT 181
x
Three Basic Characteristics 184

THE EXTERNALIZATION OF PROCESSING 184
Trends in Processing Power 187
The Future of Software 190
STORAGE 191
Trends in Storage 193
Pillars of the Memory Revolution 194
COMMUNICATION 197
From Artifacts to Waves and Currents 197
Basic Input and Output 198
Electronic Mail 201
Telephones and Videophones 202
System-to-System Communication 203
AN ESCAPE FROM PAPER? 206
Information Presentation 206
Hypermedia 207
Multimedia and Animation 209
Structuring Information 210
The Functional Approach 211
The Object-Oriented Approach 212
9 THE IT-BASED PRECONDITIONS 215

MEMORY 215
PROCESSING AND CAPACITY FOR WORK 218
The Quantitative Revolution 220
Automation 221
COMMUNICATION 223
Our Very Own I/O Bottlenecks 223
Verbal and Pictorial Information 224
Numerical Information 225
The New Channels 226
Electronic Mail and Conferencing 226
Telephones and Videophones 229
Better Hoses, Same Nozzles 231
A SUMMARY OF THE MAIN IMPACTS 232
10 EMOTIONAL BARRIERS AND DEFENSES 235
ORGANIZATIONAL EFFECTS ON EMOTIONS 236
Emotions and Organizational Constraints 238
IT AND EMOTIONS 239
Abstraction and Complexity 240
From Action-Centered to Intellective Skills 240
Responsibility and Role Conflicts 241
Rigidity and Relentlessness 242
The Significance of Design 244
Social Isolation 245
Emotional Barriers to Virtual Organizations 246
A New Gender Gap? 248
Information Technology as an Emotional Booster 248
A General Caveat 249
IV EXTENDING THE SPACE OF CONSTRUCTIBLE ORGANIZATIONS 251

xi
11 THE INDIVIDUAL AND THE GROUP 255

THE INDIVIDUAL LEVEL 255
Support Tools 255
Cell Automation 256
Increasing the Span of Competence 257
Artificial Memory 258
Artificial Intelligence 258
Embedded Knowledge 259
Response Assistance 260
But Personal Productivity Is Not the Key 260
EXTENSIONS TO THE CONSTRUCTIBLE SPACE 261
Elimination of Routine Jobs 261
De-Specialization and Knowledge Support 262
Self-Service 264
Conclusions 265
GROUPS AND TEAMS 265
Meeting Support 267
Meeting Support Systems 267
Electronic Meetings 268
Work Support 269
Infrastructural Support 270
12 ROUTINES AND AUTOMATION 277
AUTOMATION—THE CORNERSTONE OF COMPUTING 277
Routine Automation 277
What and How Far Can We Automate? 280
The Potential of Evolving Automation—An Example 283
LIMITS TO AUTOMATION—REAL OR IMAGINARY? 286
The Debate on Office Automation 286
Task Elimination: An Example 288
Banking: A Possible Next Step 290
Circumventing the Maginot Line 293
Shrinking the Organization 294
Organizational Truncation 295
Hyperautomation 298
Consequences for Society 300
13 COORDINATION BY DEFAULT 303
The Structured Database 303
That Significant Record 303
Reach 304
Capacity 305
Speed 305
Multiple Databases and System-to-System Communication 306
The Single Organization 309
Banks, Automobiles, and Airplanes 309
Bigger, Better, and Brisker 312
Decentralization 313
Implicit Coordination as an Expression of Mutual Adjustment 314
Coupled Organizations 315
xii
On the Fringes of Organization 316

14 COMPREHENSION AND CONTROL 319
COMPREHENDING THE COMPLEX 319
Getting to Know 319
Availability of Information 319
Information Concentration 320
Causal Relationships 322
Informating Work 322
Possibilities for Centralization 326
Centralizing by Informating 328
Centralization by Hyperautomation and Elimination 331
Centralization by Remote Control 332
Possibilities for Decentralization 336
Decentralization by Information Availability 337
Decentralization by De-Specialization 337
Decentralization by Increasing the Depth of Control 337
The Migration of Power 342
Control: The More Sinister Aspects 344
V THE NEW ORGANIZATIONS 349

15 TOWARD THE MODEL-DRIVEN ORGANIZATION 351
ORGANIZATIONS: PATTERNS OF ACTION, PATTERNS OF LOGIC 351
THE ASCENDANCE OF THE ACTIVE MODEL 354
From Passive to Active Models 354
Early Examples 357
A Typology of Models 359
The Regulating Model 360
The Mediating Model 362
The Assisting Model 364
SOME REQUIREMENTS FOR MODEL-DRIVEN ORGANIZATIONS 366
Model Precision 366
Skill and Effort 367
16 THE NEW CONFIGURATIONS 369
EMPOWERING THE SIMPLE STRUCTURE 370
Extending Direct Control 371
Emergence of the Joystick Organization 374
PERFECTING THE MACHINE BUREAUCRACY 375
Staying within Tradition 376
Truncation 377
The Rise of Flexible Bureaucracy 378
The Achilles Heel of Machine Bureaucracy 378
Will the Answer Be Small and Nimble or Big and Flexible? 379
Big Will Still Be Better 382
THE ENDURING PROFESSIONAL BUREAUCRACY 384
Some Ruffled Feathers 387
A New Line of Conflict 389
REINTEGRATING THE DIVISIONALIZED FORM 390
Differentiated Centralization 392
xiii
TRANSFORMING ADHOCRACY 395

The Communication Bottleneck 397
Ascendance of the Interactive Adhocracy 398
NEW: THE META-ORGANIZATION 400
Supplier Clusters 406
Equal Partners 407
NEW: THE ORGANIZED CLOUD 407
RELATING MODELS AND CONFIGURATIONS 410
17 CONCLUDING CONSIDERATIONS 413
LONG ON CONSTRAINTS, SHORT ON POSSIBILITIES? 413
Practical Theory 414
SOME SUGGESTIONS FOR THE PRACTICAL USE OF THIS BOOK 417
Products and Services 419
Processes 420
Structure 420
LIMITS TO FLEXIBILITY—BUT NOT TO COSTS? 421
THE KNOWLEDGE FACTOR 425
REFERENCES 427
AUTHOR INDEX 435
SUBJECT INDEX 439
xiv
xv
Foreword
I have always been irked by the writings on the effect of information technology
on organization. They reminded me of that joke about the consultant who sits on
the edge of the bed each night and tells his wife how good it's going to be. Some
years ago, I attended a conference on the subject at the Harvard Business
School. Eventually 1 reached my limit: You guys have had thirty years to
investigate the issue. What exactly have you done since that Levitt and Whistler
article back in the 1950s? Then Lars Groth came along. I never would have
looked at his material: it was on this same damnable subject, and a couple of
hundred pages longer than what you see here. But he enticed me by sending a
brief excerpt, which seemed unusual. So I called him. He sent me more and we
arranged to meet at a stopover I had in Oslo airport. I read the material as the
plane landed and was fascinated.
There he informed me that this labour of love was written over the previous
eight years as a doctoral thesis to be submitted to the Sociology Department at
the University of Oslo. He did this doctorate in classic European fashion: you
meet your supervisor at the outset, disappear for years, and then show up and
dump this huge document on his desk. Trouble was the man passed away, and
Lars was having trouble getting anyone else to read it.
The rules in Norway are that another university can get involved, if it so
chooses. Lars had made a contact at the Norwegian School of Economics and
Business Administration in Bergen, and I reinforced this with a call to Torger
Reve who was there at the time. Eventually they invited Lars to defend it there.
Only one problem: they asked me to be the outside examiner. How could I
refuse?
The deeper I got into this, and I must repeat that there was more material here
originally, the more fascinated I became. This was not your usual thesis, not in
its history, not in its style, not in its substance,
xvi
not in its approach. But it was damn good—the first thing I can remember
reading that really brings some insight to the slippery question of the impact of
information technology on organization. Lars defended this unusual document
on a memorable day in Bergen.
In my comments there, I explained why this thesis should be unacceptable-
the subject is too broad, the topic is too vacuous, no systematic empirical work
was done, the document is too long. All of which is to say that there is no
formula for writing a thesis, any more than for applying IT Don't trust the
professors when it comes to these things. In fact, this is an extraordinary piece
of work on all fronts: depth, creativity, language structure, historical
perspective. It is a testimonial to scholarship without socialization: few of the
highly indoctrinated doctoral students do this well. I disagree with the author in
places, but adore the way his labour of love glows from beginning to end. This
is what scholarship should be about. So Lars received his doctorate, a tribute to
the School in Bergen.
I urged Lars to get it published. It was already as much a book as a thesis
save being too long and belabored in places. So, being dragged kicking and
screaming by the likes of me and the publisher, Lars whittled it down to the
document you see before you.
I hate those endorsement blurbs on the back of books, and usually refuse to
do them or allow them to be done on my books. (Greatest thing since sliced
bread, etc.) I don't like to do Forewords either: for one thing, you should really
read the book, for another, you should often be saying no, even to nice people
and good friends. But when Lars asked me (remember, I had already read it all),
how could I refuse?
Henry Mintzberg
1998
xvii
Preface
"Sometimes a scream is better than a thesis."

Emerson, Journals, 1836
This book began its life as a research project and doctoral dissertation.
Normally, then, it should not have ended up in your hands, as doctoral
dissertations belong to that peculiar class of texts whose meaning lies in the
writing, not in the reading. As I started out, however, the thought of spending
years of my life writing for an examining committee and the library shelves
became too much to bear, and I decided to write a book rather than a thesis. I
also wanted to write a text that was accessible and interesting for both
organization people and computer people. This combination turned into a
greater challenge than anticipated, and kept me busy for more years than I like
to think of. Eventually, however, I had the good fortune of being accepted both
at the Norwegian School of Economics and Business Administration in Bergen,
Norway (for the degree of dr. oecon) and at John Wiley & Sons (for
publication), and I can now finally say that it was a worthwhile effort! I hope
you will enjoy reading it, and I welcome feedback. You will find me at
www.lars.groth.com.
Changes have been made for the book edition though. Some of the chapters
have been rather heavily edited, long quotations and detailed explanations
needed for purely academic purposes have been shortened or omitted, the
number of associative detours originally made as a personal therapy during the
long years of writing have been pared down. However, I have not eradicated
them completely. It is those small sparks springing from the associative cortex
that makes the process of writing endurable, and I could not bring myself to
carry out a complete
xviii
purge – since doing so, I might even deprive the reader of some associations,
insights, and new ideas. So please bear with me if you think I am straying
somewhat from the subject: I will soon return. Writing a cross-disciplinary
book, I have also run into the problems associated with addressing quite
different professional and scientific communities- In addition to the danger of
breaching professional community tenets, explanations included to serve one
group may tax another's patience.
Clearly there is no panacea that can resolve this dilemma, and I have had to
make a number of choices. The central decision has been to stick to my
background as an organization sociologist and make the organization
perspective the dominating one. It has nevertheless been impossible to avoid a
certain dose of computerese, and I have tried to explain as I go along. If you are
a complete stranger to computers, and find the explanations insufficient, you are
welcome to visit my website and download a summary of the history of
computing originally included in the dissertation as an appendix. There are also
a number of books on the subject available Bit by Bit by Stan Augarten is
unfortunately out of print now, but you might try A History of Modern
Computing: 1945-1995 by Paul E. Ceruzzi or A History of the Information
Machine by Martin Campbell- Kelly (details in the list of references in the back
of the book).
Similarly, for those who feel they need some background on organization a
brief overview of the many-faceted field of organization theory (the second
appendix to the dissertation) can also be found on my website. For a more
elaborate exposition, I recommend the book Organizations: Rational, Natural
and Open Systems by Richard W. Scott (details in the list of references).
. Throughout the book, I have tried to use examples to illustrate the properties of
computers and information technology in general. Many of them are actual
systems that are or have been in operation; but because my main objective is to
say something about the kind of arrangements that should be possible, not just
what has been done already, I cannot stop there To illustrate and explain what I
see as the potential of information technology and its fundamental strengths and
weaknesses, I have also used imagined examples or thought models of systems
that are possible but not yet realized.
This immediately raises an important question about which level of
technology those models should assume. To allow only existing products as
bases for reasoning would be unduly restrictive when the pace of development
is as fast as it demonstrably is in the IT industry (this text alone has resided on
four generations of computer systems and has been edited with the help of three
different word processing programs in a total of seven versions). Any
conclusion would then be overtaken by new
xix
developments before the document left the printer. On the other side,
speculations based on potential technological capabilities fifty years from now
would not be very interesting either, since a) we do not have the foggiest idea of
what that technology will look like and what its capabilities will be and b) it
would be of no use for those who would like to do something about their
organizations today or in the coming decade, since the capabilities assumed
might not be available within the span of their entire careers.
I have tried to hit the middle of the road in this matter, by only assuming
capabilities that computer-based systems already possess or are likely to attain
in the near future. The trends in the development of the most fundamental parts
of computer systems, such as microprocessors, memory, and mass storage have
shown great stability in the pace of development for several decades; the present
level of chip complexity was in fact predicted fairly accurately by Gordon
Moore in 1964 (Noyce 1977). He overshot the target by less than a factor of 10,
which is not bad at all when you bear in mind that the number of components
per chip today (1999) is several million times higher than it was in 1964.
We therefore have every reason to believe that the established trends will
continue for a substantial number of years. Moreover, we also know that most
mainstream products today were at the laboratory or prototype stage ten years
ago, and it is not unreasonable to assume that most of the mainstream products
that will be available in the next decade can already be seen in today's
laboratories. There are, of course, always surprises, but as an industry matures
the number of surprises and completely new product classes tends to diminish.
During the ten years that have elapsed since I started to write this text, I have
followed the development fairly closely, and no dramatic and unexpected new
systems capabilities have surfaced. I therefore believe that we are on pretty safe
ground if we assume that the basic capabilities we can expect from computers in
the next couple of decades have already been demonstrated, and that the
improvements in their capacity can be predicted with sufficient accuracy.
Windows may be ousted as the dominating operating system and Microsoft may
go bust (market share for actual products is impossible to forecast), but the
fundamental capabilities of computers will prevail.
xx
xxi
Series Preface
The information systems community has grown considerably since 1984, when
we began publishing the Wiley Series in Information Systems. We are pleased
to be a part of the growth of the field, and believe that this series of books is
playing an important role in the intellectual development of the discipline. The
primary objective of the series is to publish scholarly works which reflect the
best of research in the information systems community.
The Present Volume

As the information systems field continues to mature, there is an increased need
to carry the results of its growing body of research into practice. The series
desires to publish research results that speak to important needs in the
development and management of information systems, and our editorial mission
recognizes explicitly the need for research to inform the practice and
management of information systems. Lars Groth's book Future Organizational
Design: The Scope for the IT-Based Enterprise serves such a purpose
wonderfully. The author has provided an intriguing interpretation of how
information technology could both inform and transform organizations. To this
end, Groth sees organizations as 'constructible' in their own right and he traces
how organizations have been constrained because of human limitations. To
overcome these limitations, mankind has developed a variety of tools and
techniques, typically in the form of alternative organizational structures. In his
analysis, Groth draws heavily on work of Henry Mintzberg who has extensively
explored the various forms of organizational structuring. Extending Mintzberg's
view of organizations, Groth proposes the possibility of new forms of
organizations because of the advancements in information technology. He
postulates the existence of new organizational opportunities that
xxii
heretofore were unimaginable. New structural configurations, made possible

because of IT, will offer great promise to industry and government. This book
provides some of the most innovative thinking to hit the field in a long time. It
will be of interest to anyone who has even remotely considered what IT could
do for organizations.
Rudy Hirschheim
xxiii
Wiley Series in Information

Systems
Editors
RICHARD BOLAND Department of Management Information and Decision

Systems, Weatherhead School of Management, Case Western Reserve
University, 10900 Euclid Avenue, Cleveland, Ohio 44106-7235, USA
RUDY HIRSCHHEIM Department of Decision and Information Systems,
College of Business Administration, University of Houston, Houston, Texas
77202-6283, USA
Advisory Board
NIELS BJORN-ANDERSEN Copenhagen Business School, Denmark

D. Ross JEFFERY University of New South Wales, Australia
HEINZ K. KLEIN State University of New York, USA
ROB KLING Indiana University, USA
TIM J. LINCOLN IBM UK Limited, UK
BENN R. KONSYNSKI Emory University, Atlanta, USA
FRANK F. LAND London School of Economics, UK
ENID MUMFORD Manchester Business School, UK
MIKE NEWMAN University of Manchester, UK
DANIEL ROBEY Georgia State University, USA
E. BURTON SWANSON University of California, USA
ROBERT TRICKER Warwick Business School, UK
ROBERT W. ZMUD University of Oklahoma, USA
xxiv
xxv
Acknowledgements
"Gratitude is the poor man's payment.”

English proverb
The impetus behind this project was my desire to be able to say something
sensible about the interrelationship of information technology and organization.
However, without the funding generously supplied by a number of
organizations, I would not have been able to probe the question in real depth. I
am therefore very grateful to the Research Council of Norway for their early
and bold support, which enabled me to attract support also from the County of
Akershus, Elkem Aluminium, the Ministry of Government Administration, the
Norwegian National Bank, Norsk Data (later taken over by Siemens), and Norsk
Hydro. Norsk Data and the County of Akershus supported the participation of
Akershus Central Hospital. In each of these organizations, there are many
people who have helped with this project, and I feel grateful to them all.
Without their support, this effort could never have succeeded. I would also like
to thank my employer during the first half of the project, Avenir, for gracious
understanding when the project started to slip behind schedule. Finally, I would
like to express my gratitude to my partners in Pharos, who generously provided
me with the necessary overdraft facilities when my income dwindled during the
intensive last year and a half of writing.
When I started to explore the possibilities for this project, I received crucial
support from four persons. First of all I must thank Marie Haavardtun, then
managing director of Avenir, who strongly encouraged me to go on and was
very helpful in providing contacts with possible sponsors. Tron Espeli, who was
secretary of the governing committee
xxvi
for the Research Council's program "Man, Computer, and Work Environment,"
went out of his way to help me structure the project to meet the Research
Council's requirements. Professor Sverre Lysgaard at the Department of
Sociology at the University of Oslo volunteered without hesitation to review
my work—as he did more than a decade earlier when I wrote my master's thesis.
Finally, my colleague Peter Hidas both urged me on and volunteered to act as
my mentor toward the Research Council.
Sadly, both Sverre Lysgaard and Marie Haavardtun died before the
dissertation was finished, and before I could present them with the final results
of their generous support. Their premature deaths were a blow to all of us who
knew them and regarded them as friends.
Also, Professor Erling S. Andersen (the Norwegian School of Management,
Oslo), Professor Per Morten Schiefloe (The Norwegian University of Science
and Technology, Trondheim), Pal Sørgaard (then Associate Professor at the
University of Oslo, now at the Directorate of Public Management) and
Professor Ivar Lie (University of Oslo), Professor Kjell Grønhaug, and
Professor Leif B. Methlie (both of the Norwegian School of Economics and
Business Administration, Bergen), Age Borg Andersen, and Otto Stabenfeldt
(both old colleagues from Avenir), Eivind Jahren (Ministry of Trade and
Industry), Jan Heim (then at the Norwegian Computing Center), and Kamar
Singh (GE Aircraft Engines) gave of their valuable time to read and comment.
Professor Tjerk Huppes (University of Groningen) found time to receive me and
offer advice, and Assistant Professor Jan Brage Gundersen (University of Oslo)
helped me with some of my philosophical excursions. My colleagues in Avenir
and in Pharos have also been both helpful and supportive, prodding me on with
their interest. I would especially like to thank Dag Solberg for his interest and
suggestions. Dag is certainly one of the most experienced practitioners in the
field of modeling in Norway, and he is also theoretically better versed in the
subject than many academic specialists. His comments have been very useful.
However, during the writing process, two people have rendered more help
and support than others, and without any formal obligation to do so.
First of all, I would like to thank Lee Gremillion for all his support and
encouragement over the last eight years. Lee and I first met when I called on
him in Boston early in 1990 following an article in Datamation on rapid
prototyping, where a project that Lee managed was highlighted. Together with
two colleagues, I contacted him to hear more about his experiences, and Lee, in
his characteristically forthcoming and friendly way, freely shared his hard-won
knowledge with the strangers from a small country far away. Later that year, he
came over to Norway as the main speaker at a conference that Avenir organized
in Oslo on the same subject. When
xxvii
he heard about my doctoral work, he expressed interest and offered to read my

drafts and comment on them. Since then he has been my main reviewer, and
whenever I sent something over, his comments returned with a promptness
worthy of a rather more profitable client. With his doctorate from Harvard
University, his background from academic appointments at Harvard, Indiana
University, and Boston University School of Business, and his experience as a
partner in PricewaterhouseCoopers in the U.S., his advice and criticism have
been invaluable to me. He has also been an inexhaustible source of
encouragement, which has helped greatly in pulling me through the deep
troughs that invariably occur in such projects.
The second person I would like to single out is Associate Professor Gunnar
Christensen at the Norwegian School of Economics and Business
Administration in Bergen. We met during work on the Norwegian government's
1992-95 plan for developing the use of information technology in Norwegian
industry, and afterward on one of the projects under that plan. I immediately
seized upon the chance of recruiting Gunnar as an informal reviewer, and, by
and by, he quietly accepted the role as sounding board. Patiently, he responded
to my questions, offered suggestions, and listened to my occasional tales of
frustration. During the final year, he also read and commented on the complete
text, and thus effectively assumed the role Professor Sverre Lysgaard had before
his death. As one of the few researchers in Norway who is equally well versed
in organization theory and computer-based systems, Gunnar has been of great
help. Of special importance was his assistance during and after my decision to
stand for the doctorate in Bergen rather than at my alma mater in Oslo. His help
with the formalities as well as with access to the other people there who had to
look at my work was vital for the final success of my efforts.
During this last phase, when the final draft was out there begging for the final
comments, I also had the good fortune of attracting the attention of Henry
Mintzberg, whose organization models constitute the platform for reasoning
about computer-based organization in this book. His instant enthusiasm
rekindled my dwindling fires of inspiration and provided the fresh energy sorely
needed to finally complete the project. When he accepted a request from the
Bergen school to serve on my examining committee, it was another morale
booster, and his kind support during my hunt for a publisher was invaluable.
Thank you for bothering to look under the small stones along the road. Henry.
You are living proof that success need not lead to aloofness.
Lastly, I want to thank my family for enduring the hardships with me. I have
read many such statements of gratitude toward a family through the years, and
until a few years ago I viewed them as perhaps little more than
xxviii
a social reflex. Now I know better. To have one of the parents strained by
dissertation work year after year, often working both evenings and weekends, is
an experience most families could well do without. I am very grateful that you
put up with me, supporting me even through the nth delay and then again
through the editing of the book manuscript. I hope I shall never test your love
and tolerance in this way again.
I A Platform for the
Investigation
In this part, my purpose is to build the foundation for the main analytic thrust of
the book. In Chapter 1, Recourse to Reason, I delineate the project’s point of
departure and the approach chosen for the analysis: to use the basic human
preconditions for organizing as a starting point, and investigate how they are
enhanced by technology—first by pre-computer technology and then by
information technology itself. The chapter summarizes the main findings,
outlines the other chapters and provides a short note on some central terms.
In Chapter 2, Organization and Tools—the Human Advantages, I set out to
establish the (in my view) crucial link between organization and technology and
explain the concept space of constructible organizations. The chapter ends with
a delineation of the scope of the analysis.
In Chapter 3, The Basic Preconditions for Organizing, I discuss the subject
of organization, especially how organizations are defined and what their basic
elements of structuring are. The structural configurations of Henry Mintzberg
are adopted as the main framework for the analysis. The discussion concludes
that coordination is the linchpin of all organization, and a taxonomy of
coordinating mechanisms (based on Mintzberg’s definitions) is proposed. The
chapter ends with the definition of what I see as the basic human preconditions
for organizing, which will serve as the foundation for my main analysis.
2
1 Recourse to Reason 3
1 Recourse to Reason
“A man’s behavior is the index of the man, and his discourse is

the index of his understanding.”
Ali Ibn-Abi-Talib, Sentences, 7th century.
Sustained Success Grows from Knowing Why

This is not a quick-fix book. It will not teach you simple prescriptions for
turning stagnating enterprises into dynamic winners with the help of the latest in
information technology. Neither will it inundate you with computerese—
although it will introduce a number of novel concepts and hopefully coin a few
new words. It is guaranteed not to admonish you to turn your company into a
knowledge-based learning organization, virtualized into multidisciplinary,
networked teams, assembled on the go for each new challenge, meeting and
working over the Internet and delivering their products in digital form directly
to the prosumer. Indeed, you will not even find a separate chapter on the
Internet.
This book is written on the presumptions that knowledge is better than
slogans and comprehension is better than prescription—that the real key to a
sustained, profitable exploitation of information technology is a thorough
knowledge of the technology, of organizations, of people, and—above all—a
well-founded understanding of how they interact and can be combined.
A Quest for Practical Directions

In course of the 1980s, the interest in the relationship between information
technology and organization was picking up. One of the most intriguing
statements a speaker could make at the time was that a widespread adoption of
advanced computer systems would make it possible to build new
4 I A Platform for the Investigation
organization structures, more efficient and flexible than the ones we were used
to. This was always a sure hit, especially with the more well-groomed
professional audiences in the chip-chip-hurrah community. From time to time,
however, uninitiated participants would have the temerity to ask “how” or
“what kind of structures,” instantly creating that special kind of embarrassed
silence experienced in close-knit congregations when newcomers ask “stupid”
questions about central dogmas.
That also happened to this speaker a few times, and time and again I had to
fall back on the well-worn examples of American Hospital Supply Corporation
and American Airlines (and, fortunately, a couple of credible local cases).
However, they did not quite seem to fit the bill. The companies in question had
undoubtedly changed some aspect of the way they did business, and with
notable success, but the systems’ organizational impacts were questionable,
apart from eliminating a number of positions associated with the old routines.
Often I ended up saying that the ways and means here were not quite clear
yet, as we were all in the forefront of a development that was just taking off,
and that, consequently, the new structures and ways of working had yet to
emerge. It was hardly a satisfying answer for the audience, and definitely not a
satisfying experience for me. I was in dire needed of a qualified answer, and my
quest for this answer escalated into the research effort behind this book.
My viewpoint was and is a practical one. I have been working as a consultant
since 1980, and my clients always expect practical advice that will produce
concrete improvements in their organizations. That is what they pay to get, and
that is what I strive to provide. The basic goal for my research has therefore
been to offer my clients better advice and perhaps also help others who needed
to understand how their organizations could really come to grips with this new
and exciting technology. The basic questions I wanted to answer were no more
and no less than those I had encountered during my talks:
• What will the organizations look like that really take advantage of the full
power of information technology?
• How should they be structured?
• How will they function?
• What will be their benefits and drawbacks?
• Are the opportunities the same across the board, or do they vary among
organizations of different kinds?
If I could answer these questions, I felt I would also be in a much better

position to help my clients both to take advantage of contemporary systems and
to stake out the road ahead—since I would then be able to
tell if their particular organizations could benefit from intensive use of

information technology, what they would need to do in order to exploit it, and
(just as important) how they could not exploit it.
Most of the literature in this field is prescriptive and refer to concrete
examples. Prescriptions may be helpful for organizations very similar to those
described, and good examples can be inspiring, but organizations always differ
more than one should believe from a cursory look. Blind attempts to replicate
other organizations’ achievements may easily turn into disaster, since successful
adaptation always requires more than mimicry. Without real understanding of
the strengths and limitations of information technology, and how and why it
enables certain organizational designs and not others, it is very difficult to
determine where the greatest potentials are—and equally hard to stake out the
most advantageous path into the future. This book is an attempt to establish
such an understanding.
Venturing Beyond Prescriptions

It may seem brash to imply that this is a new approach—but the literature on the
organizational effects of information technology largely ignore the substantial,
established body of knowledge about organizations. It is as if the advent of the
computer has at one and the same time obliterated history and changed human
nature beyond recognition. However, IT is just a technology (albeit powerful), it
is not some kind of magic; it is not impenetrable to reason. It does not throw
everything into a turmoil or invalidate all previous knowledge about technology,
organizations and how humans use technology. Our natural abilities and
dispositions have hardly changed at all in historic times; our basic social habits
and the way we prefer to pattern interpersonal relationships are also remarkably
stable. We therefore have every reason to believe that major parts of existing
organizational and psychological theory are valid also in the age of information
technology.
As a machine, the computer is also something we can comprehend. We can
analyze its actual and possible contributions in depth, just as we have done with
previous organization technology—for the computer is certainly not the first
tool we have created to improve on our natural capabilities for organizing.
Tracing the history of human civilization, we will notice the impacts of
innovations like networks of posting stations, roads, sailing ships, railroads,
cars, airplanes, telegraph, telephone and radio—not to mention the art of
writing, the most momentous innovation of them all. Throughout history, we
have time and again used these tools creatively to expand the space of
constructible organizations—the sum of all the variations in organization
allowed by our biology, our tools and
our mores. We have literally millennia of experience with organizations, with

organizational tools and their interrelations, and for at least a century we have
been studying these matters scientifically in the modern sense of the word. To
discard such established knowledge can never be wise, even if it may sometimes
be fashionable.
In this book, the main link to established organization theory will be the
organizational configurations Henry Mintzberg first presented in his book The
Structuring of Organizations in 1979. They are particularly well suited to the
purpose, since they in an excellent way sum up the work of numerous others,
provide very useful concepts for the analyses, and are sufficiently well-known
to serve as a frame of reference for a large part of today’s managers.
My main arguments and findings are summarized in Tables 1-1 to 1-3 on
pages 6-8. The basic notion is that our use of technology—any technology—has
its roots in our desire to overcome limitations in our natural, physiologically
defined capabilities, and that this also applies to the construction of
organizations. To gain an understanding of how we might exploit information
technology in organizations, I therefore first had to define our most important
limitations with respect to organization-
Table 1-1: Overview: The basic human abilities and constraints and the
resulting organizational configurations.
Basic Human Abilities Basic Human Constraints
Versatile and creative in work. Serial: Only one task at a time.
Memory with great capacity and flexibility. Short term (working) memory extremely
limited, long term memory fickle and
unsuited for precise administrative
information.
Flexible information processing capacity, Limits in the working memory severely
good mechanisms for integration and restricts human ability to tackle complexity.
simplification.
Versatile communication abilities, great Verbal communication slow and serial.
capacity for visual processing.
Communication range well adapted to Severe limitations in range, communication
simple, local communication. over distance depends on messengers.
Emotions always important—we are less … and conflict.
rational than we like to believe. Emotions
are the source of both cohesion …
Archaic Configurations: Emergent Organizations
Simple Structure Feudal Form Adhocracy
Table 1-2: Overview: The basic contribution of pre-computer tools, the

remaining constraints and the resulting organizational
configurations.
Contributions of Pre-Computer Tools Remaining Constraints
Writing provides unlimited information Laborious search and retrieval, implicit
storage without loss of content and implicit coordination limited by need for physical
coordination of work. file access.
By augmenting the working memory, Processing still bound to the human mind
writing greatly improves the human ability and thus limited by the capacity of the
for complex processing. It also allows living humans.
monitoring of complex events, extensive
distribution of tasks and automation
Physical transport revolutionized, Fast, large volume point-to-point
information exchange doubly so (telegraph, communication expensive, basic human
telephone, radio). communication capacity unchanged.
Some improvements in the speed with which Basic human input/output limitations still
we can absorb and disseminate information, an iron constraint.
significant improvements in information
accessibility.
Some improvements in the management of Emotions still just as important for both
emotions. cohesion and conflict.
Main new organizational opportunities with Pre-Computer Tools
- Revolutionary increases in productivity and quality through automation.
- Revolutionary development of mechanical energy.
- Ability to tackle much more complex undertakings.
- Explicit design of organizations.
- Ability to organize and support really large organizations.
Modern Configurations: Explicitly Designed Organizations
Machine Bureaucracy Professional Bureaucracy Divisionalized Form
building and how they constrain us in establishing and maintaining

organizations (Table 1-1). I then explored the range of organizations built on
these basic capabilities alone (the archaic configurations at the bottom of the
table).
To isolate the possible contributions of information technology, I first looked
into the most important of the pre-computer technologies and how they helped
us overcome some of our basic limitations (Table 1-2). The most important new
opportunities these tools provided are also listed in the table. The art of writing
towers over all other inventions, as it liberated us from the crushing constraint
of having to remember every single piece of information we needed to retain.
Table 1-3: Overview: The basic contribution of information technology, the

remaining constraints and the resulting organizational
configurations.
Contributions of Information Technology Remaining Constraints
Revolutionary im-provements in information Use of information still limited by human
storage and implicit coordination reading capacity.
Information processing outside the mind, Human internal processing capacity is still
vastly extended scope for automation unchanged.
Vastly increased bandwidth for electronic Basic human communication capacity
communication, dramatically lower prices. unchanged.
Much improved comprehension of complex Basic human input/ output limitations still
information. an iron constraint.
No improvements in the management of Increasingly abstract work with systems
emotions. that can also be relentless and pacing may
induce strain.
Main new organizational opportunities with Information Technology
- Extensive elimination of work and increased flexibility through hyperautomation and
implicit coordination, possibilities for much larger and more responsive organizations.
- Support for extremely large, organized entities that are not organizations in the classic
sense.
- Close coupling of separate organizations in extended value chains, wholly or partly
automated.
- Extensive centralization of power through informating.
- Increased decentralization through improved information availability, despecialization
and increased depth of control.
- Elimination of routine jobs through increases in personal productivity.
- Improved group cooperation over distance, improved cohesion in teams and groups
who cannot otherwise meet.
Computer-based Configurations: Model-driven Organizations
Joystick Flexible Interactive Meta- Organized Cloud
Organization Bureaucracy Adhocracy Organization
Some of the potential created by this expanding inventory of tools was exploited
fairly early; some remained dormant. In fact, it was not until the advent of the
Industrial Revolution that the potential was explored to any depth, and it was
not until the twentieth century that the three modern organizational
configurations were developed to a point where we finally came up against the
limits of pre-computer tools. In contrast with the archaic organizations, which
more or less emerged from tradition and
custom and remained remarkably stable over time, these modern configurations
were to a substantial degree consciously designed, and thus open to continuous
improvement. The requirements were analyzed, the positions described and the
work of each employee detailed, often to a substantial degree. The main
organizational achievement of this age was the archetypal Modern
Organization—the Machine Bureaucracy—which proved to be the
configuration required to harness the new production and transport technologies
and to create the wealth of modern society.
Against this background, then, the contributions of computer-based systems
could be isolated. To ensure that all important aspects of the technology were
covered, I both analyzed the properties of the technology as it stands today and
looked into the main trends of development for the next ten years as a basis for
analyzing information technology's potential for organizational innovation. The
technology's four most significant contributions according to this analysis are
outlined in Table 1-3, together with the realization that our emotions are still
with us, and may be in serious conflict with some of the properties of computer-
based systems and the way we use them. In the table you will also find the main
new organizational opportunities afforded by these contributions, together with
the five proposed computer-based configurations.
Organization—A Human Endeavor

The Dawn
In prehistoric times, before we had developed significant tools, the level of
organization was modest in scale and scope. Archeology and anthropology
indicate that there were two basic organizational configurations, the Adhocracy
and the Simple Structure, building on the two primeval coordinating
mechanisms: mutual adjustment and direct supervision. These are typical for the
hunting band and the patriarchal family, respectively.
These two configurations are very different when it comes to scalability—the
ability to support larger structures. Adhocracies do not scale well; the volume of
communication needed to support mutual adjustment increases geometrically
with the number of people involved. The closest we can get is a representative
system of some kind, as pioneered in the Greek city states. The Simple
Structure, however, can easily be scaled up by encapsulation and delegation,
preferably with geography and lineage as structuring elements: the chief simply
delegates power to subordinate lords to rule separate parts of the kingdom.
They, in turn, may delegate power to rule even smaller parts, until one reaches
the bottom unit, which is always small enough to be managed by one person.
The beauty of such a system from a cognitive point of view lies in the
extreme economy it offers in information processing, communication, and
memorizing. Based on land rights and family lineage, it contains its own
structuring information; that information is moreover constantly enacted in
everyday life, and thereby reinforced in everyone’s memory. By delegating total
authority to his vassals over their fiefdoms and the people who live there, the
king effectively encapsulates the information required to run the fiefdoms and
shields himself from it. He now has to worry only about the loyalty of his
vassals (often no small worry, unfortunately), the size of their tributes, and their
contributions to his army. The number of people with whom he must deal is
drastically reduced.
The feudal type of organization can in fact be viewed as a forerunner of the
Divisionalized Form, which set separate objectives for the divisions, duplicated
operating functions across them, eliminated the need for inter-divisional
coordination, and thereby strongly limited the need for information flows across
division boundaries. A divisionalized organization thus provides a wide span of
control for top management, which was also exactly the goal of the feudal
system—developing out of the need to build large organizational structures with
minimal requirements for memorizing, information processing, and
communication.
The basic organizational forms of preliterate societies, then, relied either on
the rule of one or on a form of rule by consensus or council. For larger
structures, where one ruler or one council could not manage the complexity, the
iron-clad constraints on human memory, communication, and information
processing capabilities forced a reliance on two principles: the delegation of
authority and the encapsulation of information.
Modern Times
The art of writing was the first breakthrough technology for organizational
purposes. By making it possible to externalize memory, it lifted many of the
constraints placed upon us by our limited recall. The ability to write down
intermediate results and collect written information also made it possible to
process much more complex problems individually and to time-slice (work on
many problems more or less in parallel) much more easily. But the real
revolution was found in the way writing let us distribute large tasks among a
vast number of persons, synchronize and coordinate their activities, and
communicate intermediate results between them. This allowed a literate society
to routinely tackle tasks so massive that they would completely overwhelm any
illiterate society.
Of great importance was also the coordinative effects of the active file of
written records— once you can base your work on the information in a file,
which is constantly updated by the results of what you did, your work is
automatically coordinated with the work of all the others who use that particular
file in the same way. Such a file will provide an implicit coordination of all
those who work with it; and because of their common base of information, they
will be able to act with considerable consistency without ever talking to each
other. The fact that this information will not disappear with the user also means
that both private and public administration can survive even sudden and
unexpected changes inpersonnel. This was the first technology-dependent
coordinating mechanism, and it marked a watershed in organization history.
Once the memory barrier was lifted, communication quickly became the
bottleneck for organization building. Even though we know of regular courier
services as early as 2000 BC, communication technology capable of serving as
an infrastructure for mainstream organization building had to wait for the
industrial revolution—for the steamship, the railroad, regular postal services,
the telegraph, and finally the telephone. Mass distribution and mass media also
became important tools for erecting and sustaining large organizations. Finally,
mechanical automation has helped us overcome our limited ability to carry out
physical operations in parallel, since automation can be viewed as “canned”
action—as the enactment of previous design.
It took a long time to discover and exploit the new organizational possibilities
opened up by the evolving technology. Long after the invention of writing, the
state remained the chief domain for organization, and the feudal structure the
main organization type. Administrative technologies provided by literacy were
used only to support already existing organization practices, and literacy itself
was limited to a select few. The best explanation we can give for this is that
there were no pressing needs for new organizational forms. However, when
industrial development started in earnest, the needs arose.
The first steps toward industrial production and modern organizational forms
consisted of a more intensive exploitation of traditional approaches within the
Simple Structure model. However, traditional organizational practices could not
be scaled up to accommodate the fine-grained specialization that was now
developing. The new specialization and mechanization required much more
emphasis on coordination and control than the craft shop approach and its
derivatives. The key is that it entailed splitting all the problems of design and
production—an integral part of the craftsman’s work—away from the worker.
Those tasks now had to be carried out by specialists in design and planning, and
the workers were required only to carry out their ordained tasks. Specialization
in production therefore called for a much more sophisticated approach to
planning and administration, and therefore to information processing,
communication, and, consequently, to organization. The
division of tasks, hierarchical supervision, detailed rules, precision, clarity and

reliability that was required could not be fully realized within the framework of
a traditional Simple Structure. The configuration that Mintzberg (1979) calls the
Machine Bureaucracy was born.
Its development depended not only on technology but on the emergence of a
new concept for coordination. March and Simon (1958) have pointed to a new
and vastly more efficient method than the direct orders of traditional
organizations, namely, coordination by plan—which is roughly equivalent to
Mintzberg’s standardization of work process. By pre-programming
coordination, an explicit routine is formulated, and the need for communication-
intensive coordination by feedback, on which both direct supervision and
mutual adjustment depend, is dramatically reduced. The principle of pre-
programmed work is, of course, carried to its logical conclusion in automation.
The penalty is the cost of its conception and also of change—the efforts needed
to analyze the requirements and construct and modify the programs are
considerable, not to mention the construction of new machines.
In the modern organization, then, work is no longer organized in accordance
with custom and tradition, but according to a conscious design based on an
explicit analysis of the desired outcome and the available means. In my view,
this represents a decisive break with the past and marks the transition to a new
paradigm for the organization of human work. While traditional organizations
more or less emerged from the social context, like a natural order of things, the
new paradigm built on conscious analysis and explicit design, and focused on
the coordination of interdependent, specialized tasks. The creators of the new
organizations would almost gleefully break with tradition, if that was
instrumental to improvements in effectiveness and efficiency.
By explicating analyses and design and committing them to paper, the new
organizers also created (unknowingly) the first explicit conceptual models of
organizations. By lifting the models out of the subconscious world of tacit
knowledge, and literally spelling them out, they also opened them up for
conscious inspection and improvement. This is the foundation for the modern
organization.
The Contributions of Information Technology

As compared to earlier technology, information technology's single most
significant contribution is the way it allows us—for the first time in history—to
process information outside the human mind. This is a contribution fully on par
with the memory revolution brought about by the art of writing, even if the kind
of processing power provided by the computer
is much more narrow than the capabilities of the human mind. However, the
computer's outstanding (and rapidly improving) capabilities in quantitative data
processing and logical operations definitely provide revolutionary new
possibilities within the fields of computation (from finite element analysis to
weather forecasting), of automation, and of monitoring extremely large amounts
of information and exceedingly complex chains of events. Rather than marking
the end of straightforward automation, as a number of people argue, the
computer in my view inaugurates the age of hyperautomation—both in material
production and in the provision of immaterial products and services.
The second most important contribution is the way computer-based systems
improve our information storage capabilities. Even if writing has allowed us to
store almost unlimited amounts of information (the Earth is large, and there is
always a clean sheet of paper available), the accessibility of that information has
been very limited. The larger the files, the more labyrinthine they become, and
the establishment and maintenance of anything but the simplest keys are
extremely time-consuming. Moreover, to access a paper-based file, you have to
walk up to it or have somebody do it for you.
The database, on the other hand, allows us to store very large volumes of
information in a form that allows very efficient search and retrieval, and access
can be had from anywhere as long as there is an electronic link available. The
most revolutionary aspect of this is not the free-form information base,
however—it is the structured database, where information is stored in
predetermined fields. In a structured database, information can be accessed not
only as individual tidbits, but selections and aggregations can be made to
produce reports, statistical methods can be applied, and very sophisticated
analyses can be performed. The most awesome aspect of the structured
database, however, is its coordinative power. Its precursor, the paper-based file,
was also a powerful instrument for coordination, but the number of persons who
can work with the same paper-based file is quite limited, access to information
requires physical access to it, and the flexibility of the file is also poor—it has
generally only one key (usually name, address or date), and it is close to
impossible to scan it for other information items.
The structured database, on the other hand, has a vast (and rapidly increasing)
capacity, it can have multiple keys (all fields can be keys if your system is
powerful enough), and it can be accessed from almost anywhere in the world.
The largest database systems today are probably the airline reservation systems,
where the largest ones are approaching 200 000 terminals with real-time access.
All those users—mostly travel agents and airline personnel (and even
customers)—are completely coordinated by the database without ever having to
talk to each other. While
you are on the telephone in the late afternoon with your travel agent in
Cincinnati, mulling over if you should book that last available seat on the first
flight from London to Milan on Christmas Day, it may be snatched by an early
riser in Oslo, Norway—and you will know it and suffer the consequences in the
same instant (give or take a few seconds, depending on the current response
time of the system and the alertness of your agent). The sheer volume of the
coordination that takes place every day through these databases is—even in
theory—impossible to carry out without computers, and this great coordinative
power of the structured database is the engine behind the lion's share of business
applications. It is also the secret behind most of the famous examples of
Business Process Reengineering. Together with budding hyperautomation, it is
driving the increases in efficiency and the downsizing, reorganization and
delayering going on in large corporations today. It also promises even greater
revolutions ahead.
Only after these two primary qualities of computer-based systems comes
communication—today perhaps computing’s most touted feature. The most
important aspects of this (when we are talking about organization) are not
personal communication and networking, but rather remote access to
(structured) databases and direct machine-to-machine communication. This is
not to say that personal communication and networking are unimportant, but
they have limited potential compared to other uses of computer-based systems.
The reason is that there are a number of iron constraints that apply to human
communication, no matter how advanced the channel is: we cannot speak,
listen, write, read or comprehend information in verbal form faster than our
forbears, and the limiting nozzles sit in our own heads. No matter how wide the
hoses you lead up to those nozzles, their basic throughput will remain more or
less constant. Any communication across the network will therefore reduce the
time available for other communication activities, and even if computer-based
systems make verbal communication both faster, more convenient, and
somewhat quicker to effect, the total increase in communication capacity will be
quite modest—especially when compared to the quantum leaps we can achieve
in other areas of computer use.
The conclusions to this analysis runs counter to a number of the claims that
have been made over the last few years about the impact of information
technology—that networking is the main impact of information technology, that
hierarchy is being supplanted by networked teams, that classic automation is
outdated and that the large firms of today are doomed in the competition with
the agile, virtual organizations of tomorrow.
I claim instead that many of those assertions are based on superficial analyses
and a lack of understanding both of the basics of organization, of
human cognition and of the distinctive properties of information technology. In

this book, I propose a different framework for understanding the interrelations
of information technology and organization, a framework I believe is more
complete, and therefore provides a better basis for a deep understanding of what
the unique contributions of information technology are and how we can best
exploit them without abusing ourselves in the process.
Model-driven Organizations
Configuring Mintzberg for Computers
Each of the contributions of information technology briefly outlined above does
of course open up possibilities for organizational innovations. However, the
greatest opportunities will come from combining them through the use of ever
more comprehensive and integrated systems that are woven into the very fabric
of work, providing a new basis for automation, task elimination and system-
mediated coordination in organizations. By transforming the way we produce,
the way we work and the way organizations are coordinated, computer-based
systems expand the space of constructible organizations, opening up new
territory for organizational innovations.
Organizations that use information technology extensively will be very
complex entities. On the surface, they will not necessarily look more complex
than organizations used to do—they may even look simpler, with their lean
complexion and seemingly effortless production and service fulfillment.
However, the introduction of computer-based systems creates a new level of
sophistication, of complexity, and of abstraction—actions are increasingly tied
to symbols, and the formalization of computer programs permeates much of the
dialogues and strongly influences the work situation of those who use them.
While most modern organizations were designed with the help of a fairly
simple explicit model reflected in organization charts and verbal descriptions of
positions, tasks and workflows, the new computer-based organizations will
require a more sophisticated approach. We will in fact need to use some of the
tools that are mandatory when designing and developing large computer
programs, to ensure that organization and systems are fully integrated and
harmonized. A main tool here is the conceptual model, which contains a quite
formal and precise description of the main objects and events in an
organization—whether they are to be incorporated in a computer program or
handled by organization members. This modeling process is considerably more
refined and precise than the modeling efforts inherent in a traditional
organizational design,
and computer-based systems and the future organizations relying on them are
therefore much more clear-cut representations of conceptual models than
previous organization structures.
A computer-based system will in fact incorporate its own model, while at the
same time representing that model’s expression—or at least a part of it, as there
will usually be human actors involved in the dialogue as well. Even that
dialogue will be strongly constrained by the system’s inherent model, which can
only allow actions (accept input) that are defined within it. In addition to being
descriptive, therefore, the model inherent in a computer-based system is also
active, in the sense that it becomes a directive part in the ongoing patterns of
actions constituting the organization.
When systems multiply, their fields of operation will increasingly meet or
overlap, resulting in a need to integrate their operations. In turn, this will
necessitate a more comprehensive conceptual model of the organization’s
problem domain. If this web of systems becomes sufficiently comprehensive,
we will reach a situation where the major part of the operative actions (the
interactions that are directly relevant to the organization’s purposes) constituting
the organization will be directed toward and through the computer-based
systems, and not directly toward other humans. Somewhere around that point
we will cross a threshold: the main constituting part of the organization will be
the integrated computer-based systems, their programmed patterns of action,
and, implicitly, the conceptual model they are based on. The coordination of the
organization members will be mediated mainly by the systems, and thereby
(logically) by the model, not by direct human communication. Such an
organization will not only be model-based; it will be model-driven. The model,
integrating several of the computer-dependent coordinating mechanisms, will
constitute a supermechanism for coordinating the organization.
My analysis suggests the possibility of five such model-driven
configurations, three of them modifications of configurations originally
proposed by Mintzberg (1979), and two new ones. They are listed at the bottom
of Table 1-3. They will be based on models of different basic designs, utilizing
different combinations of computer-based systems.
The Joystick Organization

The Joystick Organization is the entrepreneur’s dream: an organization where
computers provide top management with detailed, real-time information about
all vital activities (like sales), and where the employees’ activities are highly
circumscribed by the systems they work with—systems that can be continually
modified through parameterization directed by the top management team or
even the president him-/herself. Developments in
this direction can already be seen, especially in large retailing operations. It will
rely on a regulating model, where the controlling and automating aspects of
information technology is exploited to the maximum.
The Flexible Bureaucracy

The Flexible Bureaucracy represents the natural evolutionary path for the
Machine Bureaucracy. The key to the transformation of this classic modern
organization is the IT-based transition from inflexible to flexible standardization
combined with much more efficient internal coordination. Hyperautomation,
programmed routines and often implicit coordination will combine to make
these organizations dramatically more agile than before. The lean (but often
very large) Flexible Bureaucracy in my view seems a much more likely
successor to this century’s dominant organizational form than the small, more
craft-oriented flexible specialization firms proposed by Piore and Sabel (1984)
and others. That also applies to the largest of them, which are now in a period of
contraction due to process reengineering and increasing automation. After a
period of consolidation, these behemoths of the corporate world may well start
to grow again. After all, information technology is a technology for automation
and coordination—it should allow us to construct larger organizations, not only
smaller ones.
The Flexible Bureaucracy will also rely on a regulating model, but it will be
less comprehensive (relative to the total task structure of the organization) than
the model powering the Joystick Organization. The larger the proportion of
professionals in the organization, the more likely it is that there will also be
strong elements of an assisting model present, focused on enhancing the
professional qualities of the work.
The Interactive Adhocracy

The Interactive Adhocracy is perhaps the most intriguing of the first three
configurations. Adhocracy is an organization where coordination mainly takes
place through mutual adjustment. It is typically team-oriented and creative,
producing unique solutions to unique problems. Adhocracies are extremely
communication intensive, and spend a large percentage of their energy and
resources on coordination. Therefore, they cannot compete with more efficient
organizations in turning out standardized products or services.
The most promising way of transforming these organizations is not to use
groupware or similar solutions to make communication among organization
members smoother and more efficient; we will still come up against the iron
constraints of the basic human communication
capacities outlined in the previous section. Rather, the solution is to use the
technology to eliminate the need for most of the communication in the first
place. If the organization’s main tasks can be modeled with sufficient precision
and incorporated into a computer system (or an integrated suite of such
systems), the members of an Interactive Adhocracy may be coordinated simply
through their work with the system—in the same way that all travel agents who
use the Amadeus or SABRE reservation systems are perfectly coordinated in
their seat reservations without ever having to talktogether. An early example
may be the organization that produced the structural design of Boeing’s new
777 jetliner. If we can thus eliminate most of the communication overhead in
the Adhocracy while retaining its creative strengths, the result could be a
formidable competitor in the territory between mass production and the one-of-
a-kind designer shop.
The Interactive Adhocracy will have to rely on a mediating model, which
constitutes both a tool for work and a medium for communication.
The Meta-Organization
The Meta-Organization is the first of the two new configurations suggested by
my analysis. It is an entity consisting of two or more closely coupled
organizations, coordinated by a common system or suite of systems. Early
examples are the clusters constituted by a number of modern car manufacturers
and their main subcontractors, located around the perimeter of the central
factory. Working to the drumbeat of the main factory’s production control
system, all these separate organizations are synchronized to the point that they
effectively function as one integrated organization, with the subcontractors
delivering parts directly to the pertinent positions on the main assembly line
only minutes before they are required.
There are two kinds of Meta-Organizations—the Supplier Cluster, organizing
subcontractors around a main producer, and the Equal Partnership, where no
single member has a dominating position. The Meta-Organization relies on a
regulating model.
The Organized Cloud

Finally there is the Organized Cloud—a phenomenon that is not an organization
in the classical sense, but that is nevertheless highly organized. We have already
mentioned two examples—the airline reservation systems Amadeus and
SABRE. Trading systems for stocks or currency are others. While the Meta-
Organization typically comprises a small number of organizations that are
tightly coordinated for most of their activities, the members of an Organized
Cloud are only coordinated
in a very narrow and specific area. Clouds are also typically much larger,
indeed, they can become exceedingly large. The high performance, on-line
database systems that form the core of Organized Clouds have up to now been
very expensive, and the formation of Clouds have been limited to high-yield,
time-critical activities. However, as costs continue to fall while performance
improves, the Cloud will be viable for a much wider range of purposes. Clouds
rely on highly specific mediating models that are restricted to a very narrow set
of tasks.
From Analysis to Action

It is possible to work for improvement in an organization on several levels, from
the discharge of a single task to the structuring of the total organization.
However, single tasks are normally not very interesting targets, and to obtain a
satisfactory level of insight and understanding in a larger part of the
organization, it is usually necessary to analyze the organization’s total field of
operation quite carefully in order to secure both a tenable technological solution
and a good implementation of it. It is also very important that the initial analysis
aim at getting behind existing work arrangements to capture the gist of the work
at hand—or, rather, its objectives, since the work itself (the present tasks) may
be superfluous within a new framework.
This is not a trivial requirement, and the best methodology to follow the
initial analysis is to devise a number of coaxing exercises to promote creativity.
To avoid being trapped by existing procedures, it is necessary to employ a top-
down approach in the initial analysis, starting with the primary objectives at the
highest organizational level relevant to the project: what is, quite simply, the
nature of the products and services we aim to provide? The goal should be to
describe the desired implementation of these objectives at the level of products,
services, customers or clients, and to chart the way they are related. I would
propose an object-oriented approach for this, which will force us to focus
precisely on the central objects and help us avoid function analysis—which all
too easily becomes bogged down in a detailed description of existing tasks and
routines.
The analysis should then proceed along one or more of three dimensions,
depending on what is appropriate for the project in question: product-related
possibilities, process-related possibilities, and structure-related possibilities.
Product innovation based on information technology is nothing new—indeed,
information technology has already become one of the main enablers both for
improvements in products and services and the development of totally new
kinds of products and services. Very often, such
advances hinge on one particular aspect of the technology. For instance, the
development of services so diverse as today’s extremely flexible and efficient
airline reservation, automatic teller machines and electronic toll fee stations has
been totally dependent on the existence of powerful databases with remote
access. A fuller understanding of the strong and weak points of the technology
will make it easier to pinpoint the most promising possibilities.
Process orientation has become the collective mantra of the business
community over the last few years. Practically all medium to large
organizations will have a number of processes that are central to their
operations, some of which will produce and deliver products and services to
outside customers/clients, and some of which will serve vital, administrative
needs. The key to these processes is coordination—coordination of the efforts of
all those who are part of the process, as well as coordination with customers,
suppliers and other groups and individuals both inside and outside the
organization. The comprehensive analysis of computer-dependent coordinating
mechanisms presented here should point to the most tenable approaches.
Compared to the situation a couple of decades ago, where systems were still
viewed mainly as specific tools for rather narrow functions in the organization,
today’s focus on processes is a really significant step forward. It is also a step
up onto a higher level of complexity, perhaps the highest level we presently can
handle with some confidence. However, as we gradually integrate our
processes, they will themselves become candidates for closer integration and
coordination. We then reach a level where the whole organization – and often a
part of its environment as well – must be described in the same model and
served by a set of integrated systems. This involves rising to yet a higher level
of complexity – today barely within reach, and only for organizations with a
well defined and fairly narrow problem domain. To tackle integration at the
organization level will require thorough understanding of the relationship
between work, technology and organization, and we will need advanced
methods for analysis, description and modeling. At this level, organization
structure becomes one of the paramount issues.
Organization structure is of course a subject of considerable interest already
at the process level, as key processes can involve large numbers of people and
many organization units. To achieve the best possible results, it is always
important to choose structures that match our objectives, the nature of the
required processes, and the systems central to those processes. Sometimes, one
will have a choice between process designs calling for different structures, and
it is important to know the strengths and weaknesses of those structures if the
desired results are to be achieved. However, structure first becomes a paramount
concern when
we approach not only a single process, but try to go one step further and
integrate processes, support functions, systems and system use across the total
organization. An understanding of the potential of conceptual modeling will
then become mandatory, paving the way for the use of such models as a basis
for really comprehensive computer-based systems—thus allowing organizations
to achieve new levels of integration and coordination. The analysis in this book
of the potentials for different types of organizations should also make us more
aware of the potentials of our own organization and the ways in which it could
(and could not!) be transformed.
A Key to this Book

As indicated in the Preface, I have tried to write a narrative that is easy to follow
and pleasant to read. However, even with the best intentions on my side, readers
may at places be confronted with leaps of thought that remain invisible to one
who has been steeped in this material for years. I will therefore say a few words
about how each part and each chapter fits into the scheme.
Part I: A Platform for the Investigation

In Part I, my purpose is to build the foundation for the analysis itself. In
Recourse to Reason, I have just delineated my approach and summed up the
main conclusions.
In Organization and Tools—the Human Advantages, I set out to establish
the crucial link between organization and technology, and explain the concept
space of constructible organizations, ending with a delineation of the scope of
my analysis.
In The Basic Preconditions for Organizing, I discuss the subject of
organization, especially how organizations are defined and what their basic
elements of structuring are. The goal is to identify a suitable framework from
the body of organization theory on which I could base my own analysis.
Mintzberg’s structural configurations are adopted as the main framework, and
the discussion concludes that coordination is the linchpin of organization. A
taxonomy of coordinating mechanisms, based on Mintzberg, is proposed. The
chapter ends with a definition of the basic, human preconditions for organizing,
which are to serve as the foundation for the analysis of technology use.
Part II: Individual Capacity and Organization before the Computer

In Part II, my purpose is to analyze the contributions of pre-computer
technology. Confined by Physiology begins by looking at the six basic human
preconditions in more detail. I also discuss important, traditional methods for
alleviating or circumventing some of these constraints.
In The Dawn of Organization, I explore the problems of organization-
building in societies without significant tools for organizational purposes, and
try to determine the extent of the space of constructible organizations in such
societies. The analysis focuses on the methods and techniques used to build and
maintain preliterate organizations. The analysis corroborates the conclusion that
coordination is the essence of organization, and ends with what I see as the
basic principles of preliterate organization.
In The Power of Technology, I discuss the nature of tools and the way the
most important pre-computer technologies have alleviated our original
constraints, gradually allowing for extensions of the space of constructible
organizations. The single, most important innovation was undoubtedly the art of
writing, and the great impact writing has had on our mental capacities is
explored. Next, I discuss the communications revolution of the nineteenth
century, and the chapter ends with some thoughts on complexity and the nature
of automation.
In The Modern Organization, I try to assess the relationship between the
development of pre-computer tools and the emergence of the modern
organization. I conclude that the new forms of organization, especially the
Machine Bureaucracy, were based on a new and vastly more efficient concept
of coordination: the transition from direct to indirect supervision through
standardization of work processes in the form of explicit routines and
automation. I also propose that the emergence of the modern organization
involved another breakthrough: the emergence of the explicit conceptual model
and the concomitant explicit design of organizations. The chapter ends with a
short discussion of the effect of culture on organizational forms.
Part III: IT and the Preconditions for Organizing

Now the platform for analyzing the impact of information technology is finally
in place, and I start out in Information Technology Characteristics by
assessing the state of the art of the technology and the likely achievements in
basic performance improvements during the next decade. In The IT-Based
Preconditions, I proceed to analyze how information technology can improve
the capabilities of the individual over and beyond the contributions of earlier
technology. This provides the foundation for the subsequent analysis of possible
new organization forms and practices.
While working on this part, I felt it was necessary to balance the fairly
technocentric analysis in chapter 9, and underline that human nature is not
exclusively defined by logic and reason. In chapter 10, Emotional Barriers and
Defenses, I therefore discuss how our emotional side may put a spoke in the best
technological wheel.
Part IV: Extending the Space of Constructible Organizations

In The Individual and the Group I then start on a prelude to the kernel by
analyzing the possibilities information technology provides on the individual
and group level. This is necessary both because they represent the primordial
elements of organization as well as the fundamental building blocks of larger
organizations, and because there are a number of application types (among them
some of the most hyped-up ones) that apply first and foremost to this level.
Then, I move on to the core of the matter: the larger organizational context
and the tools and potentials that apply to the organization as a whole. First I
look at Routines and Automation. Automation will in my view continue to
represent an extremely important contribution to the development of modern
societies, allowing enormous increases in productivity—something which will
also have a number of interesting side effects. Computer-based automation also
includes automatic routines at various levels, which is a very important
prerequisite for two later themes. One of them, Coordination by Default, is
about how databases can contribute to the age-old problem of coordinating
work, both improving on existing arrangements as well as providing new ones.
The second I have called Comprehension and Control; it is about how
information technology can improve our understanding and control of both our
work and our organizations by making information more accessible and even
procure information that was previously unavailable. This has clear implications
for organization structure and the way organizations can be run.
At the end of each of these three chapters, I discuss the possible extensions
information technology may offer to the space of constructible organizations.
Part V: The New Organizations

I then close in on the final target in Part V. First, in Toward the Model-driven
Organization, I discuss what it really means to build organizations with
information technology: that computer programs become ever more prominent
parts of the organizational fabric, and therefore also become part of the very
patterns of actions that constitute organizations. Next, I return to the conceptual
model: After computers and computer
programming were introduced, the model and modeling activities have become
very explicit, and are becoming extremely important within the computerate
paradigm. In my view, active models will make up the central element in most
organizations in the future.
In The New Configurations, I first discuss if and how the extensions to the
space of constructible organizations combine to modify Mintzberg’s
configurations, and find three significant new variants: the Joystick
Organization, an entrepreneur’s dream evolving from the Simple Structure, the
Flexible Bureaucracy, a formidable fighter growing forth from the Machine
Bureaucracy, and the Interactive Adhocracy—an Adhocracy where system-
mediated communication allows true mutual adjustment to work in much larger
settings than before. I then end by proposing two altogether new configurations:
the Meta-Organization, a closely coupled group of separate organizations, and
the Organized Cloud, which challenges our notions of what an organization
really is. Finally, in Concluding Considerations, I discuss the practical need for
theory, provide some ideas on how to use the book and discuss the relations
between flexibility, cost, and productivity and why computers increase the ante
in the race for improvements—and why superior IS and organization
professionals and really gifted Organization Design Managers may be the
ultimate competitive instrument.
25
2 Organization and Tools—the

Human Advantages
“Man is a tool-using animal . . . Without tools he is nothing, with

tools he is all.”
Thomas Carlyle, Sartor Resartus, 1833–34.
A Crucial Link
The basic notion behind this book is that technology has been a very important
factor in the emergence, development, and design of organizations throughout
history, and that changes in organization-relevant technology will spur changes
in the structure and functioning of organizations as well. Why do I believe this?
Apart from the commonsense assumption that telephones and computers
must matter, and convincing empirical evidence that railroads and the telegraph
did so in the past (Chandler 1977, Beniger 1986), there are also theoretically
well-founded reasons for believing so, and I would like to elaborate a little on
this theme before proceeding to identify which of our abilities and limitations
are most relevant for our organizing efforts.
To Be Human Is to Be Organized ...

Everything we know about ourselves tells us that organizing is a fundamental
part of human life—for as long as we know, humans have organized themselves
in order to accomplish tasks that are not within the reach of single individuals.
All that archeology and anthropology have discovered supports this; humans are
and have always been social
animals, and the isolated individual is an anomaly. Organization may well be

rudimentary, as in the small bands of hunter/gatherers believed to constitute the
primordial form of human society, but they nevertheless have a social structure
and a basal role diversification, and a number of the hunter/gatherers we know
of from historic (and present) times in fact have quite sophisticated social
structures. Some of the oldest texts known, such as the Epic of Gilgamesh
(believed to have been written down as early as in the beginning of the second
millennium BC), contain descriptions of elaborate social organization. In the
opening verse in the Epic of Gilgamesh alone, there are mentions of the king, of
nobles, of warriors, and the concept “shepherd of the people/city” (Sandars
1964).
The Bible even contains concrete directions—and reasons—for organizing,
as in Exodus 18:13–23, where Moses’ father in law saw that the task Moses had
taken upon himself—to pass judgement on all matters, large and small, for all of
his people—was too big for one man. To resolve the situation, he devised a
method in the form of an organizational structure to share the work among
many, leaving Moses only with the most important cases.
Implicit in these examples lies another, decisive factor: Not only are humans
born as organizers, they accumulate their experience and increase their
collective skill from generation to generation. For even if every innovation must
spring forth from an individual mind, it is (if successful) rapidly absorbed into
the collective consciousness of the inventor’s society, and may even spread
beyond to other societies if there is sufficient contact between them.
Even if the pinnacles of individual creativity throughout history are
impressive, then, it is as a collective phenomenon, as a meta-mind stretching
through time and space, spanning thousands of generations, that the human
intellect really shines. And the basis for this is our ability to organize—our
ability to congregate in groups, tribes, and societies where knowledge and skills
can accumulate and be transferred to new generations, who, in their turn, can
develop them further. Although individual contributions are recognizable, they
are indeed impossible to separate from the collective consciousness of
humanity. As Boulding says, discussing the levels of theoretical discourse for
systems theory (1956, p. 8), “. . . it is not easy to separate clearly the level of the
individual human organism from the next level, that of social organizations”.
To become masters of the Earth, we had to organize, and so we have done—
in modern civilizations, the inventory of organizations is large and extremely
diverse; they are a part of everyday life for nearly every human, and only the
hermit escapes daily contact with them.
2 Organization and Tools – the Human Advantages 27
... and to Use Tools

Humans are not the only organizers in the animal kingdom, however, although
undoubtedly they are the most accomplished. An even more distinguishing
characteristic is our ability to make tools, particularly the way in which we use
the process of collective accumulation of knowledge and experience to develop
ever better tools, tools with increasing power and complexity.
Our array of tools and methods have grown large and diverse, and we apply
them—or at least try to apply them—wherever we come up against challenges
that go beyond our bare physical and mental powers. It is hardly possible to
imagine that the realm of organizing should be exempted from this; the only
possible reason would be that our abilities for organizing had no bounds, and
that we never experienced any gap between them and our ambitions.
While this may hold true for our Lord, we—as mere humans—must rather
content ourselves with the fact that the unaided human is an animal with definite
physical and mental limitations, restricting the amount of work or the amount of
information any single individual can cope with. As March and Simon say
(1958, p. 11):
This, then, is the general picture of the human organism that we will use to analyze
organizational behavior. It is a picture of a choosing, decision-making, problem-
solving organism that can do only one or a few things at a time, and that can attend
to only a small part of the information recorded in its memory and presented by the
environment.
Not only do we have limitations that can only be (partly) overcome by

organization, but those very limitations even restrict the nature and size of the
organizations we are able to build.
To organize on the scale necessary for the conquest of the Earth, we had to
go beyond the social organization in the family, group and village, which built
directly on our innate abilities. We had to develop tools and methods for
building larger and more effective organizations, as we have done for so many
other ends. The digital computer is no more than the newest of these tools,
although it may prove to be the most powerful of them all so far.
The Point of Leverage

Earlier, I underlined how our success as a species derives from the powerful
interplay between individual creativity on the one side and collective actions
and accumulation of knowledge on the other. This field of force is evident in the
theories of organization as well, but there has been a
tendency to simplify the picture, by downplaying either the role of the

individual, the role of the environment, or both. I feel distinctly uncomfortable
with this—perhaps because my career as a consultant has awarded me with
practical experience with a fairly large number of organizations. In every one of
them, I think, I have seen the crucial role of both the individuals that make them
up and the environment they work in.
Organizations Are Constructed

Theories are by their very nature always simplifications of reality, not least
when the subject is human behavior in and around organizations—and the
reason is not only differences in contingencies such as history, cultural settings,
ages, and power structures. It is easy to lose sight of the simple fact that
organizations are not physical entities acting and behaving on their own, but
derived entities—wholes that are constituted through the actions of the
individual human beings that make them up. Those humans have their own
peculiar characteristics, dreams, objectives, and preferences, and the character
of the organization, its successes and failures, is the result of an interplay both
between those individuals and between them and other individuals outside the
organization itself.
My experience therefore supports most of the basic views of Silverman,
Weick, Berger and Luckmann, the social constructivists, and some (ontological)
postmodernists such as Clegg. Organizations are constituted through the daily
actions of their members and of the people they deal with in the environment,
directly or indirectly. Organizations do not act; it is the people constituting them
who act. Even single individuals can be of decisive importance in shaping the
fate of very large organizations. The spectacular rise of ITT, for instance, is
probably attributable to Harold Geneen (and, arguably, so is its fall), and the
growth of IBM up to around 1950 was no doubt to a large degree a result of the
vision, ruthlessness, and willpower of Thomas Watson, Sr. Further, IBM’s
phenomenal success as a computer company in the following four decades was
not the result of an inevitable development, but primarily a consequence of the
stubborn effort displayed by Tom Watson (son of Thomas). Keenly interested in
the new machines, he defied his father’s skepticism and, together with a small
team of corporate mavericks, managed to develop and produce first the 701,
IBM’s first digital computer, and then the remarkably successful 650—all
against strong, persistent opposition from IBM’s planning department, who
could not see any need for machines more powerful than the company’s existing
punched card equipment (Augarten 1984).
There is also no doubt that Henry Ford was the driving force behind the
ascendancy of Ford Motors, or that (in Norway) there would have been
no Norsk Hydro (or even Elkem) without Sam Eyde’s vision, tenacity, and
unsurpassed energy.
Even in the absence of such singular entrepreneurs, two organizations that are
formally very similar—for instance, two municipal administrations of the same
size and in the same part of a country—can be very different in how they work,
how efficient they are, what their main problems are, and how receptive they are
to change. Many of the dissimilarities can simply be traced back to the
differences between the actual persons working in the two organizations,
especially the differences between their most significant members (which
usually include not only managers and local trade union representatives, but
also strong personalities with informal influence over others). Even persons who
no longer work in the organization may cast long shadows, clearly visible in the
daily proceedings—both as symbols of unity (or discord) and through their
legacies in the form of policies and procedures.
When organization members act, however, they will of course be
influenced—often heavily—by the interpretations of meaning that their roles in
the organization imbue them with. They act within a set of frames 1 (Goffman
1974) that to a large degree incarnate the collective interpretations of both other
organization members and important persons and collectives outside the
organization. Persons with important roles in an organization are therefore often
perceived as acting for the organization, and the organization as “acting”
through those persons. Clegg’s concept of modes of rationality also fits neatly
into this framework. Acting within their local frames, agents will use the means
available and allowable to construct their organizations in a way that meets their
purposes. Since frames will be different in different parts of the world, and even
within different local regions in the same society, modes of rationality will also
differ, and no single organizational solution will achieve total domination—
although solutions that are successful in certain settings may inspire actors
elsewhere to adopt certain aspects of them that are compatible with
1 A frame is a scheme of interpretation that makes it possible for us to interpret, organize, and make
sense of particular events and actions. Frames are therefore also expressions of the generally
accepted norms in the social domains where they are valid. The same event may have very different
interpretations in different frames. For instance, crying in a funeral is generally positively regarded
and readily understood, while laughing would meet with strong resentment. In a meeting with old
friends, however, laughter would be the normal thing, and crying would be met with puzzlement
and concerned questions about the reasons. In a ceremony such as the opening of a new session in
parliament, both laughter and crying would be regarded as improper. We all recognize a very large
number of frames. Some are more or less universal, many are common to most people within a
particular society, and many are local.
the local conditions. Wholesale transplantations may also be tried, but are sure
to run into trouble.
Over time, there will of course be recurring patterns of action in
organizations, certain actions will acquire a commonly understood meaning, and
expectations about the durability of certain patterns of action will grow. It is
these recurring patterns of action that constitute and define the organization in
daily life; it is when we confront these patterns through interaction with
organization members that we sense and experience the organization itself.
The patterns themselves and the expectations they generate are normally
quite resilient; organizations can retain a remarkable degree of stability even
with a high turnover of people, and can endure great stress without breaking.
They can also continue to cling to life through year after year of unsuccessful
operation (Meyer and Zucker 1989). Even when an organization is
economically and legally dissolved, as after a bankruptcy, it happens—not
infrequently—that a number of the people who constituted that organization
will reconstitute themselves as a new organization with roughly the same
purpose and many of the habits of the old one. The ingrained resistance to any
change of routines and ways of working in almost any organization is another
manifestation of the strength of recurring patterns of action.
But They Are also Systems

However, organizations also clearly exhibit systemic properties. A defining
characteristic of a system is that it is more than the sum of its parts (Bertalanffy
1973). That is, if we study the individual cells that make up a fox, we cannot
deduce the full nature or the behavior of the fox. By dissecting the fox, we
destroy it and lose sight of its systemic properties. The same is true for social
systems: if we only study isolated individuals, we cannot understand
cooperation. By focussing solely on isolated, single acts carried out by the
members of an organization, we cannot comprehend its structure and
dynamics—and, just as important, neither can we understand the isolated acts,
since their meaning is largely derived from their organizational (systemic)
context—from the frames within which their are conceived. Actions in
organizations are also very often responses to actions by other organization
members. People thus receive feedback on their own actions and may modify
them according to their interpretations of this feedback. Their interaction
(interlocked behavior, in Weick’s [1979] terms)—and hence the organization
itself—therefore acquires a systemic character.
Even people who are not intended targets of a particular action may
choose to interpret it as something that concerns them, and act accordingly, with
ramifications not only for the original actor but for others in the organization as
well. The universe of actions that constitute an organization is therefore
dynamic, with patterns of actions and reactions reverberating through it, over
time creating the recurring patterns of action that constitue an organization. We
then have a system that exhibits both stability and dynamism. It shows stability
in the sense that it is a recognizable social entity with roughly defined roles and
a relatively predictable behavior. It is dynamic in the sense that its constituting
members will change and that they over time will come up with new actions and
establish new patterns of action.
Since organization members also have relations with people in the
organization’s environment (indeed its business transactions are built upon such
contact) organizations are also open systems. Changes in the outside relations,
in the problems and opportunities they represent, will provide major impetuses
for internal changes. Organizations where the members, especially the leading
members, are not able to interpret important changes in the environment in an
adequate way, or do not respond to them, will soon be in trouble, which testifies
to the fact that the stability of open systems is a precarious stability: it requires
continuous effort to maintain it. This is, by the way, in accord with Ashby’s law
of requisite variety (Ashby 1956), which says that to survive, a system must
contain within itself greater variety than the variety it is confronted with by its
environment 2.
If organizations are systems, then, it follows that they have characteristics
that arise from their systemic nature, and not from the actions of any single
individual. This is a salient point—and a point of controversy for at least some
action theorists, as Silverman (1970) notes. For if organizations are constituted
only through the actions of their individual members, there seems to be no room
for characteristics that are not traceable to one or a number of individuals.
I think this paradox is resolved when we take a closer look at the nature of
systems—both systems in general and the peculiar class of systems that we call
organizations. Systems are composed of parts, and real systems 3—such as foxes
and organizations—can be physically divided into their
2Humans are so successful as biological systems because they are extremely flexible with respect to
food, organization, and tool use. Insects and bacteria flourish because of their prodigious breeding
capacity, which allows a very rapid proliferation of successful genetic combinations or mutants.
Humans, insects and bacteria thus all possess a large capacity for rapid variation. Most large
animals do not: an example is the koala, whose numbers dwindle in step with the diminishment of
the Australian eucalyptus forests.
3There are also conceptual systems, such as logic and mathematics (Bertalanffy 1973).
constituent parts. However, the systems are not defined only as the sum of their
parts, but also (and indeed primarily) through the interrelations of these parts.
This means that the systems characteristics of organizations emanate first and
foremost from the interactions of their members. Of course, they have to be
manifested through concrete actions by concrete people, but since the
conception of every significant action in the organization is influenced (to some
degree) by the complex interactions and the established expectations within the
organization or toward people in its environment (often both), the systems
characteristics emerge as a quality of the individual actions themselves.
And Contingencies Matter

If organizations are systems, then, their systems characteristics become
important. What are the recurring patterns of action like? Are they different
from organization to organization, or are there similarities? Maybe some
internal and environmental conditions are important?
This is leading toward the questions posed by contingency theory, and I see
no reason to back away from them, since I think there are regularities in the
systems characteristics of organizations, even if no two organizations are
completely alike—just as most people have two arms and two legs despite their
individual differences, and most small children giggle when tickled, no matter
what culture or nationality they belong to. Galbraith (1977) addressed this
question (Is there a general theory relevant to a specific organization?) in the
introduction to his book Organization Design (pp. 7–9). His conclusion, based
on interpretations of empirical studies, was that 50% to 75% of the variance in
organizations could be accounted for by general theoretical propositions,
leaving the rest to specific factors peculiar to the individual organizations. This
means that no organization can be understood apart from its history, its
particular setting, and the particular individuals who dominate it, but neither is
any organization isolated from more general relationships.
One can probably differ in opinion on how much of the variance can be
explained by general propositions—that proportion will vary, among other
things, according to the cultural homogeneity of the sample—but there is, in
my view, ample evidence that there is indeed a mix. Therefore, we have no
choice but to approach the analysis of organizations on several levels: that of the
individual actor, that of the single system, and that of the system in its
environment. Important insights can be gained on either level; they are all
significant for organization design and organization change, and all are relevant
for my present purpose.
The Space of Constructible Organizations

The starting point of my research efforts was the apparently innocent question
of what organizations based on innovative use of information technology would
look like. An effort to answer that question cannot build only on empirical
evidence of past and present achievements, since we are only beginning to learn
how to use computers. Rather, we must try to map out the space of possible
organizational arrangements after the introduction of computers. We can now
outline this problem more precisely.
As acknowledged above, organizations are shaped by a great number of
factors. Not only the traditional contingencies apply. Clegg (1990), for example,
drawing on a large, cross-cultural selection of studies, presents a convincing
case for the way cultural, technical, economic, and other contingency factors
can influence people to assemble their organizations in innumerable ways—and
still operate them successfully, both domestically and when exposed to
international competition. In addition comes the fact that organizations are
constructed by real humans, exhibiting great variation in their dispositions and
goals.
Building on this, we can say that the space of possible organizational
solutions in any particular situation is determined by the local mix of relevant
contingency factors at that point in space and time, including the normal range
of individual characteristics of the members of that society, as well as their local
social norms and arrangements. We can call this space the local space of
possible organizations. Acknowledging our inborn limitations as humans, we
can also safely conclude that there must be some absolute limits to what we can
achieve, defined jointly by our biological nature and the capability of the
available tools. These limits define what we might call an absolute space of all
possible organizational solutions.
However, this absolute space is not constant, nor is it entirely defined by
nature. I will therefore avoid the term “absolute space” and replace it with three
concepts that reflect the mixed nature of humans: on the one side, we are
biological creatures with a set of biologically defined characteristics, like other
animals; on the other side, we have an extreme social and cultural plasticity and
creativity, providing a potential for development that is utterly different and
superior to that for any other creature on Earth; but even so, there are things we
do not do, even if we could actually accomplish them.
The first of these concepts I will call the primal space of organizations. It is
defined by man’s biological characteristics and basic psychological and social
needs—the possible organizations of the human animal, if you like. The second
one I will call the space of constructible organizations
(“constructible space” for short). It represents the expanded organizational

space made possible by the tools, methods, and social practices developed by
humans to relieve the limiting constraints of the primal space. The constructible
space encompasses all existing local spaces. In fact, it is equal to the sum of all
local spaces.
While the primal space is by and large constant (changed only by biological
evolution), the constructible space is changing all the time. Generally, it also
expands, since new technologies and the evolution of methods and customary
practice tend to increase the number of alternatives. Of course, developments in
social and moral attitudes will over time render some customary arrangements
unacceptable (such as slavery and serfdom in our parts of the world), but such
curtailments have so far not remotely matched the increases in variation. The
arrow has thus pointed largely in one direction, and only a major loss of
technological, methodological, or social capability through a global catastrophe
could possibly lead to a significant contraction of the constructible space.
The third concept is the technical space of organizations, which is defined
solely by what should be possible given humanity’s physiological capabilities
extended by the existing technology and methods, thus excluding psychological,
social, and cultural constraints. It will, of course, be larger than the constructible
space, since it is always possible to imagine physically feasible organizational
solutions that will not be psychologically feasible or socially acceptable in any
actual situation, and thus impossible to realize within any local space.
My prime interest is the constructible space and how it is extended by
information technology. However, as a step on the road, it is necessary also to
try to outline the primal space, and the way the constructible space has grown
with the development of tools and methods. Since many people in the computer
business also show tendencies to equate the technical space with the
constructible space, in blithe ignorance of the social and psychological needs
and preferences of normal humans, it will also be necessary at various points to
discuss some of the main differences between the two.
Defining the Boundaries of Constructible Space

What, then, defines the boundary of the constructible space? Obviously, the
number of factors is large, and an exhaustive analysis is probably impossible.
Analytically, at least, we can discern four broad classes of factors, hinted at
above, which are the most significant:
• Biological characteristics: Obviously, we cannot build organizations that

presuppose telepathy, unlimited human memory, or the ability to run at the
speed of five hundred miles per hour.
• Psychological characteristics: As humans, we have a psychological

makeup that limits what we can accomplish and tolerate in daily life. The
limits here vary strongly with the circumstances. For instance, people are
prepared to endure much higher strain in a crisis (such as a war or disaster)
than in a normal work situation. Opinions also change over time—as for
instance views on what kind of job conditions that are either harmful enough
to impair an organization, or too harmful to be acceptable to employees.
Indeed, to reduce the acceptable range of harmful working conditions has
been a central purpose for the labor unions since their inception.
• Social and cultural factors: This class comprises social organization
(including family systems), culture (including knowledge, norms, and laws),
and social institutions in the fields of religion, economy, and politics. It is
obviously of great importance to the definition of the constructible space. A
special case here is the sphere of illegal organizations, like the Mafia and
other agencies of organized crime, and organizations that defy commonly
accepted norms, such as organizations of various kinds of dropouts. Insofar
as they are fairly stable, de facto elements of most societies, they must
clearly be viewed as being inside the constructible space: in fact, it is easy to
show how they comply with the local mix of contingency factors in their
particular sub-culture, and thereby define a viable local space. Borderline
cases will be organizations such as the extermination camps of Hitler’s
Germany, the gulags of Stalin’s Soviet Union, and the concentration camps
of other, lesser perpetrators, which may be defined either as temporary, freak
outgrowths of the constructible space or as a genuine part of the
constructible space to which access is (fortunately!) normally hindered.
• Available tools and methods: Analytically, tools and methods can be
viewed as two different classes of factors. However, they are so often
intertwined that it seems more appropriate to group them together: writing is
not possible without writing materials, nor is accounting; lateral filing
requires special equipment; and so on. Sometimes it is also difficult to draw
the line between a method and a social practice in association with a tool:
the clock is certainly a tool for timekeeping, but is the practice of reliably
reporting to work at a particular time every day (so necessary for certain
organizational forms) a method or a social practice? Perhaps it is both—a
basic instance of a method for coordination that has become an ingrained
part of the social fabric of industrialized societies.
The example of the concentration camp underlines an interesting feature of

constructible space: As one moves toward its boundaries, it becomes more and
more difficult to actually construct and maintain
an intended organization—sometimes because it pushes the limits of generally

accepted norms, sometimes because it verges on exceeding the tolerances of
human psychology or physiology, and sometimes because it stretches
technology toward its limits.
If we were modern social physicists, then, we might endeavor to create field
equations for the constructible space—and discover that there is a Great
Attractor at its center, perhaps in the form of the rationalized myths of our
societies (Meyer and Rowan 1977), which shape the organizations we build
unless we consciously (and at the price of a considerable effort) go in another
direction. Now, social physics is an antiquated approach, and field equations are
hardly viable tools in the social sciences—but as a metaphor, they can be useful.
All local spaces will have their own set of attractors in the form of traditionally
preferred organizational forms, and it will always take vision, boldness, and
energy to go against tradition and construct something new.
For an organizational form to fall within the constructible space, it is, of
course, not a requirement that every particular instance of that form survives. It
is, for instance, perfectly possible to set up a new organization in a market
where the competition is too tough for most newcomers to make it; indeed, this
happens regularly in open, market-oriented economies, as entrepreneurs often
overestimate their chances and end up in bankruptcy. The point is that it is
possible to set up that kind of organization at all. Environmental factors such as
the number of existing firms in a market or the profitability of a certain line of
business is therefore not a factor in defining the constructible space.
The Scope of this Investigation

The object of this investigation is to understand how and why information
technology extends the boundaries of the space of constructible organizations.
Obviously, it is impossible to discuss all possible permutations; to make a
significant contribution without overextending my undertaking, I will have to
concentrate on what I believe are the most important factors involved.
Our Biological Characteristics

As indicated above, I consider the first class of factors, our physiological
characteristics, to be the most basic determinants of the size and shape of the
constructible space. These constraints can also be assumed to be constant on the
timescale of interest to us here. Proper training can improve individual
performance, but our basic capabilities have probably remained fundamentally
unchanged for thousands of years and represent
the iron constraints that put absolute limits on human achievements. It is also
these constraints that we have most eagerly attacked with the help of technology
and methods—not only because they are so unyielding, but also because they
are the ones most open to amendment by such means. I will therefore discuss
these constraints in considerable detail, and base most of my analysis on the
way they are alleviated by technology.
Our Psychological Characteristics

The basic situation is probably much the same for the second class of factors,
our psychological characteristics. That is, they have been constant for a very
long period of time. However, they are a lot more pliable than our biological
capabilities, at least in the way they manifest themselves. The limits for the
amount of aggression or compassion that can be expressed, for instance, or the
amount of psychological stress a person can endure, are to a considerable degree
determined by social and cultural factors. We do not find this kind of variation
in talking speed or in the number of items that can be held simultaneously in
short-term memory.
Since our psychological attributes seem to be both less limiting to our
organizing abilities and less amendable by technology, they have received scant
attention in the development of computer-based systems. The exceptions are, of
course, the user interface, which has received much notice over the last ten to
fifteen years, as well as methods and strategies for introducing computer-based
systems in organizations.
Both of these factors as well as the size and vigorousness of the market for
computer games and gimmicks such as cartoon-like screen savers, “eyes” that
follow the cursor around, and so on, suggest that emotions and other nonrational
parts of our psyche may be more important in our interactions with technology
than normally recognized by computer professionals. Although the subject is
somewhat elusive, I have found it difficult to leave it out altogether, especially
since I believe emotions play a very important part in deciding the viability of a
number of computer applications, not least in the “groupware” category. I have
therefore included discussions on the role of emotions in several of my analyses.
Social and Cultural Factors

Social and cultural factors are, of course, even more malleable than our
psychological characteristics, as we can see from the great variation found
among the societies in the world today and in historic times. Their influence on
the size and shape of the local constructible space is great but so varied that it
constitutes a vast field of study in itself, challenging not only the discipline of
sociology but almost all of the social sciences. This makes it impossible for me
to incorporate social
and cultural factors into my analyses—other than as examples to throw light on

particular problems—with one very important exception: the tools and methods
we use and have used to construct our organizations.
Of course, this does not mean that the present analysis is culturally neutral.
Although I have strived toward a dissociation from my particular social and
cultural background, I readily acknowledge the fact that a completely neutral
stance is simply not possible for any human. My analysis will therefore
obviously be most appropriate for the Western industrialized sphere. However, I
also believe that humans have enough basic traits in common to make the
analysis valid, to a large extent, even in other cultural settings.
Tools and Methods

Tools and methods are both creations of the human mind, and as such they are
wholly constructed expressions of both knowledge and social values. A
computer is indeed a cultural manifestation, and a mighty one at that.
Tools and methods also represent the most pliant class of factors with bearing
on the space of constructible organizations, and is by far the class subject to the
most rapid development. For the last several hundred years at least, the
development of tools and methods have arguably outpaced all other patterns of
change in human society, and the speed is not getting slower.
As I noted above, tools and methods can be roughly separated analytically,
but in practice they are intimately connected and often intertwined, and will
more often than not have to be discussed as combined phenomena.
This brings us back to an implication that was raised in the discussion on
biological characteristics: it is, of course, a central question just how new tools
and methods make their undisputed contributions to changes in social and
cultural conditions (which include organizations). However, if we believe that
the basis for the construction of the social fabric (and of organizations) is the
individual actions of the members of that society, it follows from our
conclusions above that the direct influence of tools and methods must come
from the way they change and enlarge the realm of possibilities for individual
actions. And such changes must spring mainly from enhancements of our basic,
physiologically defined capabilities.
The main analytical thrust in this book will therefore remain the
augmentation of our natural, biological constraints by technological means
(including both tools and methods), and what kinds of extensions to the space of
constructible organizations these augmentations will allow. To get started, then,
we must first decide which of our abilities (or constraints) are most important to
our ability to organize.
39
3 The Basic Preconditions for

Organizing
“I am certainly convinced that it is one of the greatest impulses of

mankind to arrive at something higher than a natural state.”
James Baldwin, “The Male Prison,” in Nobody Knows My
Name, 1961
The Essence of Organization

I have stated that organizations, in my view, are constructed; like any other kind
of social system, they are constituted through the actions and interactions of
their members, both between themselves and with people in an organization’s
environment. I also said that the systems characteristics of organizations
emanate precisely from these interactions and manifest themselves as a quality
of the individual actions themselves. By enhancing the abilities and capacities
of the individual organizational members, then, the use of tools will alter the
systems characteristics of the organization as well.
Further, the most important systems effects must be those that arise from the
types of recurring patterns of actions most common to organizations: the actions
that aim to carry out their basic functions, and thus constitute what we may call
their structure. To discover the main enabling qualities of new tools, we must
therefore identify the human abilities and constraints that are most important to
those basic functions.
When the Task Becomes Too Large for One

What, then, constitutes the essence of organization? What are the basic features
and functions of organizations? When sifting through the
literature on organization theory, one is struck by the fact that although nearly
everyone complains about how difficult it is to define “organization,” they still
tend to end up stressing largely the same central features. Jay R. Galbraith
(1977), contributes an intuitive definition, when he says that (p. 2): “ . . .
organization is that ‘something’ which distinguishes any collection of 50
individuals in Kennedy International Airport from the 50 individuals comprising
a football team in the National Football League.” Getting more formal, he
stresses the need for a shared purpose, for division of labor and “information-
based decision processes” (1977, p. 3).
Henry Mintzberg, on his part, starts his book The Structuring of
Organizations (1979) with an illustrative story about the potter Ms. Raku, who
started out making pottery in her basement. She did everything herself, just like
any ancient craftswoman—wedged the clay, formed the pots, tooled them,
prepared and applied the glaze, and fired the pots in the kiln. She then marketed
and sold the pots to craft shops. Everything went smoothly; there were no
problems—except that demand outstripped supply.
Ms. Raku then hired an assistant who was eager to learn pottery, and
everything still went without hassle, even though Ms. Raku now had to divide
the work. The assistant wedged the clay and prepared the glazes, and Ms. Raku
did the rest—since the shops wanted pottery made by her. It required some
coordination, but since they were only two, and worked in a small studio, this
posed no problem.
Before long, however, Ms. Raku was again outselling the production
capacity. More assistants were needed and, even with three new people,
coordination could be conducted informally. But as still more assistants were
added, Ms. Raku faced more serious problems. There were simply too many
people to coordinate everything informally and without plans, and, besides, Ms.
Raku was now mostly away from the studio, spending time with her growing
number of customers. The time had come for the first assistant to become studio
manager and full-time supervisor.
Ms. Raku’s ambitions were limitless, and the company continued to grow,
branching out into new product lines (even clay bricks) and new customer
groups. Eventually, she was the proud president and owner of the large,
divisionalized company Ceramico, with her office located on the 55th story of
Pottery Tower. Ms. Raku and her company had traversed the history of human
organization in a couple of feverish decades.
Mintzberg concludes this introduction (p. 2, italics and bold in the original)
by making the following observation:
Every organized human activity—from the making of pots to the placing of a man on
the moon—gives rise to two fundamental and opposing requirements: the division of
labor into various tasks to be performed and the coordination of these tasks to
accomplish the activity. The structure of an
3 The Basic Preconditions for Organizing 41
organization can be defined simply as the sum total of the ways in which it divides
its labor into distinct tasks and then achieves coordination among them.
To be recognizable, an organization must have a domain for its activity, and it

must have a set of objectives and goals. There is great variation as to how
consistent and explicit domains, objectives, and goals are, but it is difficult to
envisage an organization completely without any kind of consciousness about a
common domain and some common objectives; we would then really be
confronting something that was more akin to Galbraith’s accidental collection
of 50 people in Kennedy International Airport. Next, an organization must have
what Galbraith (1977) calls an organizing mode—a way of decomposing work
into subtasks and a way of coordinating them for the completion of the whole
task. This gives rise to a need for significant amount of communication and
information processing, a need that grows with the level of task uncertainty and
the pace of change in the environment (Galbraith 1977). To Galbraith, this
strongly indicates that the required level of information processing will have a
decisive influence on organization structure: The amount of uncertainty will
dictate the amount of control and coordinative activities that will be needed,
which again will require certain levels of information processing. He therefore
ends up with an information processing model of organization as the basis for
his design framework.
James G. March and Herbert A. Simon, in their Organizations (1958) also
stress the high degree of coordination in organization behavior. In his foreword
to the third edition of Administrative Behavior (1976, p. xvii, italics in original),
Simon says that: “. . . the term organization refers to the complex pattern of
communication and relationships in a group of human beings.” To Simon, the
organization is mainly a decision-making system—coordination and control for
him therefore become synonymous with the communication of decision
premises and decisions. He further emphasizes the importance of
communication as “essential to the more complex forms of cooperative
behavior” (1976, p. 106), and he is very clear about the need for stable,
predictable cooperative patterns and communication channels if an organization
is to operate efficiently.
Such definitions are what Scott (1987) would term rational systems
definitions of organizations: they do not include either the informal
communication and conflicting interests of the natural systems definition, or the
relation to the environment of the open systems tradition—both of which are
also vital parts of action and constructivist perspectives. However, the authors
quoted here are aware of both aspects and discuss them in their books.
The views referred to above are also in close accordance with general
systems theory, which claims to be a general theory of organization in
physical, biological and social systems, and which has had substantial influence
on organization theory for the last three to four decades. Both W. Ross Ashby
(1962), Kenneth Boulding (1956), Walter Buckley (1967), and Ludwig von
Bertalanffy (1973) define organization mainly in terms of communication.
Basic Elements in Organization Structuring

Most writers seem to agree on the basic features of organizations: the division of
labor and the concomitant need for coordination. The coordination efforts in
turn require both information processing and communication. Necessary
exchanges with the environment and organizational adaptations because of
environmental changes require additional processing and communication.
Variations in the nature of these basic features should therefore have
considerable influence on organizational structure.
The Division of Labor and Structuring of Work

The division of a greater, common task into smaller ones that are suitable for
single persons is the defining feature of purposeful organizations. In principle,
there are two ways of dividing work: Everyone can do the same in parallel, as
when 50 persons go together to clean up a beach and all collect litter in their
own plastic bags, or the total task can be divided into specialized subtasks.
Practically all purposeful organizations belong to the latter category, simply
because there are extremely few of them that have tasks so simple that it is
possible for every organization member to do exactly the same thing. Once the
overall task is divided into more or less specialized jobs, it becomes a challenge
to structure those jobs by grouping them (and thereby the people who execute
them) in a way that ensures both that the organization’s mission is accomplished
and that the efficiency of the operation is sufficient to ensure the survival of the
organization.
A basic determinant for organizational performance is the grouping of tasks.
Grouping is necessary to establish a system of coordination and supervision, of
resource sharing, and of performance measurement (Mintzberg 1979, Nadler
and Tushman 1988). The basis for grouping can be either by activity, output, or
customer. These three categories can be further decomposed—activity into
function or skill, for instance, and customer into market segment or
geographical region. Most often, different bases for grouping are used at
different levels in the organization. For instance, top management may be
grouped according to function (marketing, finance, production, etc.), the middle
level according to
market or product (or both), and production according to process or function.

The reason is, of course, that different criteria for grouping may apply at the
various levels.
The criteria used for grouping normally reflect the interdependencies that are
seen as most important. Mintzberg (1979) counts four such interdependencies:
work-flow interdependencies embraces all kinds of interdependencies between
separate tasks or stages in functionally specialized organizations, process
interdependencies refers to interdependencies within the separate stages
themselves (note that “process” here does not mean the same as the “process” in
BPR), scale interdependencies are simply about economies of scale, and social
interdependencies denote social interaction and social needs
When organizations are drawn between conflicting criteria, as they often are,
they must either choose the one they deem most important or, if the conflict is
too pronounced, try to accommodate it by creating various kinds of matrix
organizations. The most common conflict is between product/market and
function/process. A common solution is then to have one array of managers
with responsibility for one set of considerations, and a second array for the other
set. There are even examples of three-dimensional matrixes, with functional,
product-, and market-oriented axes (Mintzberg 1979).
Coordination
Grouping is in itself the primary instrument for coordination. Usually, we (quite
intuitively) try to group together those functions that seem to have the most
immediate interdependencies. The reason is, of course, that physical proximity
allows richer communication, and, generally, the richer the communication, the
closer, swifter, and more flexible the coordination. The primary group, where
coordination is effected through informal communication and where feedback is
immediate, is the building block of all organizations.
However, while informal communication may be quite sufficient for the
coordination of individual activities within the primary group, it cannot support
the necessary coordination within and between the larger units in the
organization. To accomplish this, the organization has to communicate and
process large amounts of information across groups and units (Mintzberg 1979):
The information part of the work flow (such as work documents and time sheets
and even oral instructions), control information and decisions, which constitutes
the main bulk of information associated with coordination (performance and
problems upward; directions and plans downward), and finally staff information
(operating data toward the staff, professional advice or plans and programming
toward the line).
All of these information flows have to be taken care of in order to make an

organization function properly, and this is where we hit the main challenge of
organized work. Indeed, all the authors cited here view communication and
information processing as the main bottlenecks for organized activities,
although perhaps Galbraith expresses it most clearly. To him, information
processing is the very focus of design; as organizations will seek to reduce the
need for it as much as possible. They will group tasks with great care, they will
preplan as far as the environment allows, and they will only maintain the
flexibility that is needed to cope with the variations the environment forces
upon them. If they can, they will in fact try to influence their environment to
make it more stable, and if their competitive situation allows it, they may create
slack resources in order to tolerate lower internal performance.
If this is not sufficient, an organization will have to increase its information
processing capacity in order to cope. Two main alternatives are open (Galbraith
1977): Improving vertical information processing capacity (this is where
Galbraith early on saw an important role for computers), or creating lateral
relations. The information processing capacity of a hierarchy is bound to be
quite limited—if it attempts to coordinate the activities of different units by
communicating through the formal structure, the organizational hierarchy is
easily overloaded. The development of lateral relations is seen as the main
remedy. Both Mintzberg (1979) and Nadler and Tushman (1988) follow
Galbraith closely here, and the prescription is what Mintzberg calls liaison
devices and Nadler and Tushman call structural linking. They divide it into four
basic types, ascending from liaison individuals (persons with a special
responsibility to inspire cross-unit coordination by informing about certain
aspects of their units’ activities), via cross-unit committees or task forces (with
the same purpose, but with more comprehensive participation), integrating
managers or departments (similar to liaison individuals and cross-unit groups,
but with a stronger mandate and more clearly defined responsibility), to the full
matrix organization, where there are two (sometimes even three) intersecting
chains of command. Mintzberg also argues that the organization’s planning and
control systems represent lateral linkages, especially action plans (Mintzberg
1979).
Another aspect of the struggle to coordinate is the question of centralization
versus decentralization. If our capacity to communicate and process information
were limitless, organizations could be totally centralized, with all decisions
made by one brain, knowing all and directing everyone. Since this is impossible,
no organization can be said to be totally centralized; however despotic the top
manager, the majority of decisions will nevertheless be made by his or her
underlings, although they will obviously try to cater to their master's tastes.
Decentralization, then, is a means to reduce the information processing

requirements in the organizational hierarchy by spreading the processing
throughout the organization—to engage more brains, so to speak. That will also
reduce the amount of information that has to be communicated up and down the
hierarchy. It is important to note that physical dispersal of services or facilities
alone does not necessarily imply real decentralization (of decision-making
power). It is quite conceivable, for instance, to run a bank with a network of
branch offices and still have all loan applications sent to a central office for
processing; indeed, that was what most banks did before the advent of
computers and remote terminals.
Every organization also exists in a social, political, and technical context, and
will always have a multitude of formal and informal relationships with both
organizations and persons in its environment. These relationships may span the
gamut from customers and suppliers to the Internal Revenue Service to the
families of employees. To survive, the organization must maintain the necessary
exchanges across its borders and be able to adapt to or resist changes in the
environment to a sufficient degree to keep resources flowing. Both the daily
exchanges as well as the necessary adaptations make heavy demands on the
coordinative capacity of any organization, and the nature of the environment is
therefore a very important parameter for organization design—to Mintzberg, it
is even the most important determinant for organization structure. He describes
(Mintzberg 1979) four basic types of environments (complex-stable, complex-
dynamic, simple-stable and simple-dynamic), each of which correlates with one
of his four basic coordinating mechanisms and thus also with one of his
structural configurations.
The Linchpin of Organization

As a one-sentence conclusion of the discussion so far, I think a passage from
Mintzberg’s definition of organization quoted earlier in this chapter is perfectly
suited (1979, p. 2): “Every organized human activity—from the making of pots
to the placing of a man on the moon—gives rise to two fundamental and
opposing requirements: the division of labor into various tasks to be performed
and the coordination of these tasks to accomplish the activity.”
First of all, then, organizations above a certain small size (a few individuals)
have a division of labor, since any single individual has a definite ceiling on his
or her work capacity, be it physical or mental work. Single tasks must be made
small enough to be fit for single persons. When you thus divide work among
several people, it means that coordination between those tasks can no longer be
effected within one brain, as when one person does everything. This sounds like
a rather trivial conclusion
(indeed a tautology), but the implications are far-reaching, because coordination

of more than a handful of people involves most of what we know as
organization.
Coordination among several individuals in turn implies communication,
which may be either routine (proceduralized) or ad hoc. Without information
flowing about what everybody should do and how they progress, coordination is
impossible. Next, information about what is happening both inside the
organization and in its environment must be collected and distributed, and the
organization will also need to process information of this kind.
Communication and information processing also imply that the organization
will need to establish some sort of an organizational memory—which in
modern times usually means files and archives of different kinds, in addition to
the vital information organization members carry around in their heads. (In
preliterate times, people had to carry all the information around in their heads—
a very constraining demand, as we shall see later.)
Further, there must be accepted mechanisms for reaching decisions on all
levels, which also means that there will be a power structure of some kind
(even in a small, egalitarian group, some people will usually have more
influence on decisions than others). Finally, to secure permanence, the
organization must fulfill a number of other conditions, which we have not
discussed so far. If it is a normal public or private organization, it will have to
reward its members in some ways, and provide the necessary tools, premises,
and amenities for their work and well-being.
In my view, these propositions are as objective and well-founded as any
social science proposition can be. Despite the great latitude of today’s
organization theories, and the enormous breadth of variation in actual
organizations, it seems evident that organizations have some basic problems in
common, and that coordination emerges as the crucial factor.
An analysis of the organizational contributions of any tool, including
information technology, should therefore preferably be based on an
organizational model built around the concept of coordination. To my
knowledge, the best framework available that answers to this criterion is the
structural configurations of Henry Mintzberg. The model presented in his The
Structuring of Organizations (1979) also represents, in my view, the best
attempt so far to meld the main structuring elements of organizations into one
model. It has the added advantage of being well known both in academia and
among professionals and managers. I will therefore use it as a workbench for
my own analyses in this book.
Before we go on, however, I think it is necessary to take a second look at the
subject of coordinating mechanisms, and I also think it will be
useful to establish a taxonomy (Figure 3-1, p. 51), which we can refer to during
subsequent discussions.
A Taxonomy of Coordinating Mechanisms

Even if Mintzberg’s configurations and coordinating mechanisms are well
known, I also hope I will have readers from professions and fields of research
that are not acquainted with them. If you feel you need more background
knowledge, I can only recommend you to go to the source itself (most complete
in Mintzberg 1979, shorter in 1983 or 1989). There are no better expositions
available. For the benefit of those of you who know them in principle but suffer
from the same sneaking amnesia that I do, I will first provide a short summary
of the main points before I discuss the coordinating mechanisms in more detail.
If a group is sufficiently small, says Mintzberg, both the division of work and
coordination can occur naturally through informal communication between the
group members. That is the secret of the small team’s flexibility—everyone is
constantly updated on the activities and intentions of all the others. Mintzberg
(1979) terms this basic coordinating mechanism mutual adjustment. The
corresponding structural configuration is the Adhocracy, a creative, project
oriented organization living in a complex and dynamic environment.
However, mutual adjustment demands a high volume of communication.
Ideally, every member of the group must communicate with every other
member—or, at least, all the members must listen in on the group’s shared
dialogue, and contribute the relevant information about their own actions and
needs. Obviously, when the group expands, the required volume of
communication rapidly saturates human communication capacity, and
coordination through mutual adjustment breaks down—just as Ms. Raku in
Mintzberg’s example discovered when plant hangers started to take the wrong
color and people began tripping over pots stacked on the floor.
At that point, someone must take the lead (either through appointment or
common consent) and start planning and directing the work of the others. At its
simplest, this takes the form of direct supervision. That is what Ms. Raku
resorted to when mutual adjustment ceased to work. She named one of her
assistants studio manager, and divided her own time among planning,
supervision, and customer relations. The corresponding configuration is the
Simple Structure, a centralized organization (often a startup) in a simple,
dynamic environment, with a strong leader who keeps the organization simple
and informal, distrusts professionals and likes to interfere everywhere (because
he or she “knows best”).
Even direct supervision in its basic form breaks down fairly rapidly,
however; since here is a definite limit to how many workers one person can
continuously direct and coordinate. The actual number depends on the nature of
the work, but even with the simplest and most undifferentiated work there
cannot be more than a few tens at the most. However, direct supervision can be
extended through delegation: the original leader can appoint a number of people
to supervise a group each, and concentrate on supervising these subordinate
leaders. In turn, they may again appoint leaders of still more groups, and so on.
Theoretically, there is no limit to the size of such a hierarchy, but it is evident
that lateral coordination must be achieved chiefly by channeling information
and decisions up and down the hierarchy: to reach a decision involving two
groups at the bottom, the matter must be brought up to the first common leader.
The strain on individual communication capacities therefore increases toward
the top, which is the only point where the communication lines to and from all
the groups meet.
To avoid communication saturation, the need for coordination must therefore
be reduced. Two avenues for action are open: the lower layers of the
organization may be regrouped to create self-contained organizational entities
that require a minimum of coordination, or the activities may be standardized in
some way to reduce the need for supervision in the first place.
The first solution is what we normally understand as divisionalization, or the
Divisionalized Form in Mintzberg’s terminology. The complexity of day-to-day
business is encapsulated within the division, corporate management will
normally not meddle; and communication with headquarters can be limited to
results, plans, and budget proposals upward, and appraisals, budgets, and
planning directions downward. Such a company is often old and very large, and
the divisions are most often organized as Machine Bureaucracies.
Standardization can take three forms: Standardization of work,
standardization of outputs, and standardization of skills. Actually,
divisionalization usually implies standardization of output: As noted above,
corporate headquarters is primarily interested in a division’s profit, not its
internal workings. Even the work of individuals can be supervised in this way,
as when factories employ piecework rates combined with standards for product
quality, or as when Ms. Raku tells the clay wedger only to deliver the clay in
four-pound lumps, not how to wedge. However, as I will argue in a minute,
standardization of output is really not a proper coordinating mechanism.
Standardization of work involves a direct specification of how work is to be
done. It usually presupposes fairly extensive planning, to ensure that
interdependent tasks are carried out in such a way that the necessary
coordination between them is automatically ensured. On the other hand, it

eliminates a considerable volume of supervision, and allow for a much higher
ratio of productive to supervisory work in an organization. Ms. Raku resorted to
standardization of work when she hired a work study analyst to organize her
first production lines. Mintzberg’s standardization of work corresponds to
Galbraith’s preplanning, and is the defining characteristic of a bureaucratic
organization, or Machine Bureaucracy as Mintzberg calls it. The Machine
Bureaucracy is often large and old, it operates in a simple and stable
environment and excels in mass production of products and services of a high
and uniform quality.
Another approach is to build on standardization of skills. This is what
vocational and professional training is all about. On all levels, from bricklayer
to brain surgeon, the educational process equips craftspeople and professionals
with a professionally certified set of solutions to common tasks, ready to be
activated. When Ms. Raku expanded, she hired assistants from the local pottery
school, and they could immediately go about their work without further
instructions. Mintzberg calls the organizations based on this coordinating
mechanism Professional Bureaucracies. On the surface they may resemble
Machine Bureaucracies, but they are normally dominated by one or more
professions, and the rules that govern them are typically decided inside
professional associations and educational institutions, not within the
organizations themselves. Their environments tend to be stable but more
complex than the Machine Bureaucracies’. A hospital is the proverbial example.
Usually, an organization does not depend on only one coordinating
mechanism; several or all will be used at some point and on some level. But
often the organization has a defining mechanism that will serve as a basis for
the main part of the work and contribute heavily to the organization’s character.
Mintzberg later (1989) added another two configurations, 1 the Missionary
Organization and the Political Organization, which are not so much complete
configurations as images of organizations kept together by strong common
norms or pulled apart by strong conflict, respectively. He also acknowledges
that most organizations today are mixtures of configurations. He therefore now
also represents them as forces acting on organizations. Circumstances decide
which one of the forces will exert the strongest pull and shape organizational
structure. For our discussions later, both the pure types and the concept of pulls
will be useful.
1He has also changed the name of two of the configurations: the Simple Structure has been renamed
the Entrepreneurial Organization, and the Adhocracy is now called the Innovative Organization. I
use the original names here, since they are the most widely known.
Two Classes of Coordinating Mechanisms

As we noted, Mintzberg defines three main coordinating mechanisms, one of
which has three subforms:
• Mutual adjustment
• Direct supervision
• Standardization:
- of work
- of skills
- of output
Mintzberg also argues that these mechanisms create a continuum as one moves
from simple to more complex work. Mutual adjustment gives way to direct
supervision, which in turn must yield to standardization of some kind. However,
if the level of complexity (especially combined with a high rate of change) rises
even further, it will saturate the adaptive capacity of standardized coordination.
Mintzberg then postulates a return to mutual adjustment as the only
coordination mechanism that can handle really complex work with a lot of
problem solving. However, it is clear (although he does not point it out
specifically) that this is mutual adjustment on another level than in the small
group working in Ms. Raku’s workshop. His example is NASA’s Apollo
project, which was extremely large and complicated, and where the adjustments
had to take place in a “representative” system based on professional
competence. He also sees a similar role for direct supervision, which may
become the answer when an organization (or indeed a nation) in crisis appoints
a manager with more or less dictatorial power.
If we take a closer look at these coordinating mechanisms, we see that they
fall into two main classes: the real-time ones (“coordination by feedback,” a
term borrowed from March and Simon 1958), where coordination is continually
adjusted as people observe the effects of their own and other people’s actions,
and the programmed ones (“coordination by program”), where coordination is
effected through instructions or plans (“programs”) generated beforehand. Both
main types have two main variants (see Figure 3-1).
Real-Time Mechanisms
The real-time mechanisms are mutual adjustment and direct supervision, which
are the same ones that Mintzberg defined.
With mutual adjustment, coordination is achieved by a continuous exchange
of information among those who participate in the work. It is
Coordination of Work
Coordination Coordination
by Feedback by Program
Mutual Direct Standardization Standardization

Adjustment Supervision of Work of Skills
Tacit Explicit
Skills Skills
Figure 3-1: A taxonomy of coordinating mechanisms.
the coordinating mechanism best suited for complex problem-solving work with
little standardization. In its pure form, there is no single person who continually
directs or supervises the work, and it is therefore necessary for all organization
members to have a sufficient understanding of their goal, the overall design of
the work and how their different tasks fit together. They must also be
sufficiently motivated to do their part voluntarily. Mutual adjustment is an
inherently egalitarian coordinating mechanism, best suited for settings where
people are on a fairly equal footing. Without compromises, it is impossible to
extend it to organizations larger than the small group.
The reason for this is, of course, because mutual adjustment in its pure form
requires everyone to communicate with everyone else. As the number of
participants increases, the number of possible information links multiplies: with
five people, there are 10 links; with 10 people, 45 links; and with 20 people, 190
links. 2 To use terms from network theory (Lincoln 1982): to employ mutual
adjustment as the prime coordinating mechanism, a network must be very
dense—and, since our communication abilities are limited, that means they will
also have to be small. In large organizations, real mutual adjustment can take
place only inside organizational units small enough to allow all-to-all
communication, or between similarly small groups of managers or group
representatives, acting on behalf of their departments or groups. Through an
elaborate project hierarchy, it is thus possible to achieve a kind of layered
mutual adjustment, but only with strong elements of hierarchy and bureaucratic
control.
2The formula is n⋅(n-1)/2, where n is the number of members in the group.

Direct supervision is quite different from mutual adjustment, since it

presupposes that someone directs the others, tells them what to do (even how to
do it), and monitors their actions during execution. While mutual adjustment
requires all participants to know (and accept) the goals and task designs, direct
supervision in principle requires only one person to know the goals, the overall
design, and how tasks are meant to fit together. It is also inherently hierarchic,
and therefore easily extendible through delegation of authority.
Programmed Mechanisms
Mintzberg lists three coordinating mechanisms based on standardization:
standardization of work, standardization of skills, and standardization of output.
In my view, only the first two of these are proper coordinating mechanisms, in
the sense that they are used to coordinate the work of organization members in
order to achieve particular patterns of action in an organization. Standardization
of output does not involve any coordination of people or work at all, only a
prescription for a certain result—usually in the terms of form of profit, although
even total sales, or tons, or pieces of whatever one produces may be used. It is
therefore of little interest for discussing the potential organizational
ramifications of computer-based systems, even if it may represent a useful
method for pacing work or for controlling the profitability of large and far-flung
corporations.
I will therefore include only the two basic programmed coordinating
mechanisms, standardization of work and standardization of skills.
With standardization of work, as Mintzberg describes it, coordination is
achieved by specifying beforehand and in some (often considerable) detail how
work is to be done. It is best suited for fairly simple work where tasks do not
change very often, and can then be very efficient. Most large organizations use
standardization of work extensively, especially Machine Bureaucracies.
While standardization of work may be said to represent the special program,
developed for a specific collection of tasks in a specific organization, standardi-
zation of skills represents the general program—an education designed to enable
one to tackle cooperation of a specific kind, unrelated to a particular organi-
zational setting. I propose here that there are two kinds of standardized skills.
For the set tacit skills I borrow part of the name from Polanyi’s tacit knowledge;
it is also related to Argyris’ (1980) theories-in-use. It comprises the kind of
internalized skills that are seldom or never made explicit, and which we may not
even be aware of as distinct skills. The prototype of such skill sets is the
standard social skills everyone learns during childhood and adolescence, which
makes it possible to function as a normal member of society and perform the

expected roles in everyday interaction.
The other set, explicit skills, includes the skills that are taught in schools or
during apprenticeship, and which serve not only to teach the candidate a set of
concrete skills, but usually also a code of professional conduct and a notion of
the accepted level of quality. In larger organizations, this coordinating
mechanism is best suited to fairly routinized but complex work (as in the
medical professions).
Not only does this kind of education serve to standardize work processes on a
professional basis, it also contains elements that are designed to enable
coordination both within the profession and with colleagues from other, relevant
professions. Mintzberg’s example (1979) is the medical professions: the
cooperation between the various specialists both among doctors and nurses
during an operation is largely regulated by procedures learned during their
education, and are usually not specific to a particular hospital. In fact, my own
contacts with health personnel strongly indicate that the main stumbling block
for organizational development in large Norwegian hospitals is the combination
of standardized professional procedures and high staff turnover, which makes it
excessively demanding for a hospital to introduce and maintain procedures
different from those at other hospitals.
The Basic Preconditions for Organizing

With the discussion in this chapter in mind, which of our abilities have
significant bearing on our capability for organizing ourselves? Clearly, our
capacity for physical action is pivotal; our limitations here is the reason we need
to organize in the first place (apart from strictly social purposes). Further, as
noted above, all cooperation presupposes communication, and our bottlenecks in
this area are, of course, extremely important. Then there is our ability to
accumulate and retrieve information and our capacity for information
processing—what our brain can actually accomplish with the information it is
fed. Finally, to carry on cooperation through time, the reliability of organization
members also becomes an important issue.
Translating these five concepts into actual human faculties or properties,
there seem to be six areas where we quickly run into limits restricting
organization building:
1. Capacity for work: Obviously, both our need for organizations and their
nature are strongly dependent on the nature and amount of work that we can
carry out—on how much a single individual can accomplish. Although our
capacity for physical work is of obvious
importance, our special interest is in the limitations we have with respect to

mental work.
2. Memory performance: The basis for any intellectual activity, and crucial
for accumulation of knowledge and for management of complex
relationships. Both the storage capacity and the retrieval capabilities of our
long-term memory is of vital importance. So are the limitations of our short-
term working memory.
3. Information processing capability: Closely related to the question of work
capacity, our ability for reasoning, problem solving, and decision making is
directly related to the amount of complexity we can handle.
4. Communication bandwidth: This is the first of communication’s two
aspects. The amount of information we can absorb and disseminate per unit
of time is of obvious importance.
5. Communication range: This is the second aspect of communication. How
far and how fast we can communicate is also central, as are the possibilities
of communicating not only over distance, but through time.
6. Emotions: The five properties above are derived from the rational activities
in organizations. However, we are not entirely—maybe not even
principally—rational beings. As the action, constructivist, and postmodern
approaches to organization (among others) point out, emotions play a
decisive part in our daily lives both within and outside organizations. We all
have our secret aspirations, phobias, likes, and dislikes, and we all have to
live with our basic, primate instincts and psychological makeup.
Some may find it strange to include emotions in the small number of basic
human properties that are most important for organizing, especially when my
expressed purpose is to analyze the interplay between organization and
information technology. However, my purpose at this stage is not to single out
the human faculties that will be most influenced by the use of information
technology; rather, it is to decide which ones are the most important for the
construction and maintenance of organizations—and I believe that emotions are
extremely important, for the spirit that can develop in organizations, for the
conflicts they harbor, and for their reliability as logical “machines.” I also
believe that emotions (used here as a collective term for the nonrational part of
the human mind) and the social relations they foster have extremely important
impacts on the use of technology, and often determine if a specific application
will be successful or not—quite independent of its “rational” merits. My
discussion of emotions will therefore differ from the way I treat the five other
faculties listed above: the discussion will focus not so much on how emotions
are “enhanced” or “improved” by information technology, but on how emotions

will impact its possible use and thus influence the general development both of
the technology itself and of IT-based organization.
It is difficult to ascertain which of these abilities or properties are most
important, both in man’s almost mythical “natural state” and in contemporary,
industrialized society. It may even be meaningless to rank them, since they are
so intertwined in real life. Most of the limitations we encounter can fortunately
be ameliorated by tools (but to a varying degree), and through our history as a
species we have amended our shortcomings in gradually more advanced and
powerful ways. Both by material technology and with the help of techniques
and methods of various kinds, we have considerably increased our organizing
abilities. Those shortcomings that respond most readily to amplification by
material means easily attract the most interest, of course, especially in a
technologically mesmerized society like our own.
Information technology represents nothing more and nothing less than a new
chapter in this history. It promises, however, to become an extremely important
chapter, which we will be busy writing for a long time to come. In keeping with
contingency theory, we may say that IT modifies and extends the technology
contingency factor. This may seem innocent enough, but, in my opinion,
computer-based systems modify this factor substantially—and in the process
extend it from a matter primarily of the operating core to an important
contingency factor not only for the rest of the organization, but for its exchanges
with the environment as well. Through this, it may open up possibilities for new
structural configurations and provide the basis for significant shifts in the fit
between common configurations and the different kinds of business or task
structures they can efficiently support.
57
II Individual Capacity and

Organization Before
the Computer
In this part, my purpose is to establish a platform for the analysis of the possible
contributions of information technology to the space of constructible
organizations. I begin Chapter 4, Confined by Physiology, by looking at the six
basic human preconditions or constraints (as listed at the end of Chapter 3) in
more detail. I also discuss two of the most important methods we have always
used to alleviate or circumvent some of these constraints (in addition to
simplification), namely, imitation and the creation of mental sets.
In Chapter 5, The Dawn of Organization, I explore the problems of
organization building in societies without significant tools for organizational
purposes and try to determine the extent of the space of constructible
organizations in such societies. The analysis is based on historical records and
anthropological evidence from primitive societies, and focuses on the methods
and techniques used to build and maintain organizations. The analysis
corroborates the conclusion from Chapter 3 that coordination is the essence of
organization, and it concludes with the basic principles of preliterate
organization.
In Chapter 6, The Power of Technology, I discuss the nature of tools and the
way the most important pre-computer technologies have alleviated our original
constraints (preconditions for organizing), gradually allowing for extensions of
the space of constructible organizations. The single most important innovation
was undoubtedly the art of writing.
Finally, in Chapter 7, The Modern Organization, I try to assess the
relationship between the development of these tools and the emergence of the
modern organization. I conclude that the new forms of organization, especially
the Machine Bureaucracy, were based on a new concept of coordination: the
standardization of work processes in the form of explicit
58 II Individual Capacity and Organization Before the Computer
routines and automation. This allows a transition from direct to indirect

supervision, which is vastly more efficient. However, I also propose that the
emergence of the modern organization involved another breakthrough: the
emergence of the explicit conceptual model and the concomitant explicit design
of essential parts of the patterns of action that constitute organizations. This
opened the door for conscious improvements and a rational approach to
organization, as opposed to the traditional approach of oral societies and
societies with a weak literate foundation. I end the chapter with a short
discussion of the effect of culture on organizational forms, and the possibility of
claiming any significant common ground in organization structure.
59
4 Confined by Physiology
“Man is a mind betrayed, not served, by his organs.”

Edmond & Jules de Goncourt, Journal, 1861
The six areas listed at the end of Chapter 3 are all about restrictions ultimately
rooted in our physiology. The limits to our capacity for physical work are the
most obvious, but even the others reflect our biological capabilities at varying
levels. Our capacity for communication, for instance, relies both on the physical
bandwidth of our senses (chiefly the eyes and the ears), the strength and nature
of our voice (relying on sound waves), the physical characteristics of our mouth
and vocal chords, and the brain’s capability for information processing.
Processing, in turn, is dependent on extremely complex processes in the brain,
of which we presently have very limited understanding. The same is true with
our emotions, desires, and drives, and the irrationality they often give rise to. To
understand the nature of organizations and the way tools empower us, we must
therefore have a basic understanding of our fundamental capabilities and—most
important—our limitations as biological creatures.
One Thing at a Time

As any traffic authority can confirm, distractions while driving—from
conversations to stereos—increase the likelihood of accidents. The reason is that
we have to split our attention between driving and talking or pushing buttons.
Even if driving is a more or less automatic activity for most of us, the simplest
additional activities attenuates our attention sufficiently to slow down our
responses to unexpected events on the road.
The fact is that we are quite single-minded animals, both with regard to
physical action and intellectual activity. There are very definite limits to what
we can consciously perceive and do in parallel. To achieve good control of any
complex physical activity, for instance, we need to practice
again and again until the control of movements is automatized (Ellis and Hunt
1989) and our conscious processing capability is relieved from the burden of
coordination. That is why we need so much practice as children to walk and run
with reasonable control. As grown-ups, we confront the same process again
when learning to drive a car or to master new sports. Our conscious system is
simply not able to handle the coordination of many muscle groups in real time.
To achieve perfection in sports that require complex movements combined with
high precision (like gymnastics), extreme amounts of drill are necessary—many
hours every day, years on end.
It is fairly easy to understand some of the limits for parallel physical
activity—in the end, we only have two arms and two legs. However, our
possibilities for actual coordination of muscles seem to be restricted by the same
basic mechanisms that limit our mental activities. Generally, activities (both
physical and mental) that require our full attention preclude the possibility of
doing something else simultaneously (Ellis and Hunt 1989, Barber 1988): we
cannot carry on two conversations at once, or read a book and ponder a
complicated problem at the same time.
On the other hand, it is possible to be engaged in one conversation and still
keep “an ear” on a conversation close by. You can split your attention if you
reduce concentration on the main task (Ellis and Hunt 1989). It is a common
experience in the proverbial cocktail party: if you are engaged in a trivial
conversation with someone, you will tend to notice the contents of the
conversations around you, scanning for something more interesting. However, if
the conversation you are engaged in is sufficiently absorbing, you will not
notice anything other people say, except (possibly) if someone mentions your
name.
Serious work, then, requires almost all of our attention and blocks other
activities. If we are interrupted or attention drifts (shifts to a daydream or
pondering a piece of news read in the morning paper), activity comes to a halt.
The only exceptions are activities that are automatized to such a degree that we
do not need to allocate much attention to them. If less common situations or
problems occur, however, even normally automatized activities will absorb our
full attention until a normal situation is restored.
Since there is no way of getting around the attention barrier (except for tasks
that lend themselves to automatization), our work acquires a serial nature: we
have to attend to one task at a time. We often do not complete it at once, but
break off, do something else, and then return. In fact, this is the normal mode of
office work. It may sometimes look like we do many things at once, but a closer
look reveals that we are only switching back and forth—allocating slices of time
to each task, as Mintzberg found managers do all the time (Mintzberg 1989).
4 Confined by Physiology 61
If the number of such parallel “sessions” increase, we quickly approach

information overload. Air traffic controllers in busy areas probably come close
to the limit for human conscious control. Even when they are highly
experienced, and have a level of automatization as high as this kind of mental
work can allow, they are occasionally simply overloaded—sometimes with
disastrous effects. Barber (1988) cites the evidence of an actual accident in the
Zagreb area where two planes collided in mid-flight in September 1976 (p.
101):
It was nearly two minutes before the DC-9 first communicated with the upper-
sector controller, following the instruction to change to the upper-sector radio
frequency. Meanwhile, the controller for that sector had been working without his
assistant, having in effect been responsible for two jobs for some minutes.
Moreover there were eleven aircraft in his sector, he was in radio communication
with four other overflying aircraft, and he took part in a telephone conversation
with Belgrade regarding two others. In that short interval he transmitted eight
messages and received eleven. The task facing him seems to a lay observer to have
been an unenviable one, and it is apparent from the working practices for air-traffic
control (cf. Sperandio 1978 1) that this is not a mistaken impression. Indeed the
inquiry board were clear in their view that he had been subject to overloading. (He
was subsequently prosecuted, held partly responsible for the accident, and was
jailed.)
Our capacity for conscious action is thus limited by the serial nature of our
mind. As additional work is piled upon us, we cannot compress our workload in
time by doing several tasks in parallel; we have to increase the time available
for work each day, thereby taking time away from eating, sleeping, family life,
and socializing. This is a merciless fact, learned the hard way every day by
millions of people in rich and poor countries alike. Taking our fairly modest
physical strength into account as well, it comes as no surprise that most human
endeavors require cooperation, and the amount of work or level of complexity
does not have to be large before organization is necessary.
Memory
Memory lies at the very base of our nature as intelligent beings. Without
memory, without any retained experiences or patterns to which we could
compare sensory signals, we could not live. This is aptly reflected in archaic
Greek mythology: Mnemosyne, the goddess of Memory, was no less than the
child of Earth and Heaven (Gaea and Uranus), and the
1Reference in original. The article is: Sperandio, J.C. (1978): “The Regulation of Working Methods
as a Function of Workload Among Air Traffic Controllers, ” Ergonomics 21:193–202.
mother of the nine muses. Our memory is both wonderful, fascinating, and
frustrating. It is also a very complex phenomenon. As Mishkin and Appenzeller
put it in the introduction to their article “The Anatomy of Memory” (1987):
Within the small volume of the human brain there is a system of memory powerful
enough to capture the image of a face in a single encounter, ample enough to store
the experiences of a lifetime and so versatile that the memory of a scene can
summon associated recollections of sights, sounds, smells, tastes, tactile sensations
and emotions.
Memory is indeed many-faceted; several areas of the brain are involved, and
there is obviously a fair amount of specialization between them (Mishkin and
Appenzeller 1987, Geschwind 1979). There are also various theories about how
memory is structured logically, and whether different parts of memory functions
according to different principles (Ellis and Hunt 1989). For our purpose, the
distinctions here are not so important, and we can leave that discussion alone.
The main divisions from a “user” standpoint are between the sensory registers,
short-term memory, and long-term memory.
Sensory registers are simply buffer memories that store the raw data pouring
in from our sensing organs for the very brief time (a few tenths of a second) it
takes for our attention to select them for further processing and to transfer the
result of the pattern recognition process to short-term memory for interpretation.
Information not selected (by far the vast majority of it) is lost when the
information in the sensory registers decays or is “overwritten” by new data.
Even if we often find it irritating that information is lost in this way, it is in
reality (as Weick remarks) a good thing (Weick 1979, p. 208). It is precisely the
process of selection and interpretation that allows us to sense and function at all;
otherwise, we would be permanently overwhelmed by unstructured information
both from our senses and from our memory.
The short-term memory is the workbench of our active consciousness. There
has been much theorizing about the similarities and differences between short-
term and long-term memory, but the current view is that short-term memory is
not so much a separate memory system as a workspace for information selected
for transfer from the sensory registers, as well as for information retrieved from
long-term memory when we want to use it in an active thinking process. For this
reason, it is often termed working memory (Anderson 1990, Ellis and Hunt
1989). It can retain both sensory impressions (such as sights and sounds),
numbers, words, concepts, and ideas. This working memory has a very limited
capacity; laboratory experiments suggest that the normal range for humans is
between five and nine elements or “chunks,” with seven the average. It
is thus probably not a coincidence that the number seven is sacred or prominent
in religious conceptions in many cultures.
The “chunks” can be of any kind, size, and complexity—from single letters
or numbers, to complex concepts (like “democracy”) or objects (like “passenger
airplane”). The very fact that we perceive them as chunks implies that we see
them as organized entities, conceptually or physically, with certain main,
defining properties that blend into a single representation in working memory.
The contents of working memory must be constantly rehearsed to be
maintained for more than a few seconds. (Experiments indicate that, on the
average, it takes us only 18 seconds to forget 90% of the contents in working
memory if rehearsal is prevented.) However, with rehearsal, our access to its
contents is fast and reliable, due to its high level of activation. We are able to
compare, juggle and manipulate the items maintained in working memory.
Long-term memory, on the other hand, retains information for an indefinite
period of time, once it is encoded. Indeed, most current theories assume that
encoded memories do not decay, and that forgetting is just failure to retrieve
(Ellis and Hunt 1989). There are a number of different explanations for why
retrieval may fail. It has been claimed that we can recall even trivial childhood
incidents under hypnosis, and experiments with electrical stimulation of points
in the brain’s temporal lobes have elicited forgotten childhood memories
(Penfield, referred in Anderson 1990). It is, however, difficult to verify the
correctness of such “provoked” memories, and they are generally not accepted
as proof that we retain all memories.
There are likewise a number of theories about recall, building on different
views of the organization of memory. We know, however, that we can
remember things directly, recall the wanted item with the help of a cue, or
slowly work our way toward the right information following chains of
memories or associative paths in memory, uncovering new cues as we go along.
Sometimes, we may even have to leave the conscious process alone for a while,
to allow the wanted item to “drift” to the surface. Memories below a certain
threshold seem to be impossible to recall under normal conditions, but
experiments indicate that they are still present—number/noun pairs once learnt
but apparently forgotten seem to be easier to learn and remember later than
totally new pairs (Anderson 1990).
Much to our chagrin, then, we seem to have limited control over what we can
recall. Not only is it very often difficult or even impossible to remember the
items we need, but what little we do recall is quite likely to be incomplete or
even distorted, especially when it comes to details. Facts are not only selected,
interpreted and accorded meaning during recording, but are subject to later
selections, interpretations, and changes
in a subtle interplay with other memories, as well as with preconceptions,

desires, and hopes. It is not only the future that is unpredictable and open to
conjecture; so indeed is the past. Anyone doubting this is referred to the daily
affairs in our courthouses, where the dominating activity is the painstaking
review of human explanations and physical evidence in order to establish
plausible descriptions of past events. It is also precisely the fickleness of human
memory that leads us to grant greater credibility to documents and pictures than
to human explanations, even when there is no reason to suspect intentional
misrepresentations. As Cohen (1980, p. 85) remarks describing the archaic
Greek view of memory, “The hidden things of the past, no less than the future,
have to be wrested from the gods. Recollection, therefore, becomes a species of
retrospective prophecy. ”
In addition to the fuzziness that often mars our memory and makes us
doubtful as to what actually took place or what the details of a conversation
really were, our memory may also let us down in a more deceptive way: there
may be errors in what we think is a clear and unequivocal memory. Sometimes
we may even “remember” events that never happened, or remember as our own
experience something told us by another. When it comes to early childhood
experiences, for instance, it is often difficult to separate what we actually
remember from what we have been told by parents, older siblings, or others.
Our brain has no error detection and correction mechanisms that can alert us to
such erroneous memories, once they are established.
There are thus clear limits to the amount of knowledge that any one person
can absorb, remember, and, most important, recall and use, and with only our
unaided memory at hand, we are therefore severely restricted when it comes to
organization building; our memory deficiencies restrict the size of organization
that can be run. It is difficult to keep tabs on large numbers of people, and the
accumulation, transfer, and dissemination of knowledge is cumbersome.
Moreover, information is continually subject to deterioration both in individual
memories and during communication. The lack of permanent records precludes
the accumulation of knowledge above a certain, rather basic level; social
relationships spanning great distances are difficult to maintain, and trade is
generally restricted to low-volume, direct barter.
No wonder, then, that organization in nonliterate societies tends to be fairly
small scale (by our standards) and mainly tied to family relations—by far the
strongest, most important, and stable social framework of societies past and
present. Even personal connections outside the family line tend to be couched in
family terms: for the aborigines of the Cape York Peninsula in Australia, for
instance, trading partners in neighboring tribes were classified as ritual brothers
(Sharp 1952). The relationship was considered so close to a real brotherhood
that one of the men was
always defined as elder and one as younger, with the “elder” brother having a
perpetual advantage over the “younger” in the trading relationship, because a
younger brother by custom had to show deference to an older one. Similar
practices can also be found elsewhere in the world, for instance among the
Lapps of Northern Scandinavia and seems to be a significant characteristic of
more complex nonindustrial societies (Pehrson 1964).
It is also noteworthy that some of the most successful nonliterate,
expansionist peoples have been those with the most extensive family systems.
The Bantus of Africa, for instance, had to their advantage a notion about a
common forefather, implying that all Bantus are related. They also kept tabs on
their lineage several generations back. If threatened, a Bantu could summon the
assistance of his entire family line, down to the point where it merged with his
opponent’s. If the opponent was not Bantu, then he could call upon all Bantus.
This enabled the Bantu tribes to amass superior forces in all instances when
opposed by more fragmented tribes. In just a few hundred years, they swept
southward from the North of Africa, eradicating weaker tribes in their way.
Today, the large majority of the African population south of the Arab territories
belongs to the Bantu group.
Information Processing
Although we cannot handle more than one serious information processing task
at a time; at least that process can be multidimensional—but only within narrow
bounds, as our working memory quickly becomes saturated. There is a definite
limit to how many variables or aspects of a particular subject we are able to
juggle at the same time. To picture the relationship between cost, sales volume,
price, and profit is barely possible, as long as the factors are stable and the
relationship between them linear. If cost is production cost, however, with a
nonlinear relationship to volume and time from order to delivery, and if the cost
and effect of marketing and sales must be taken into account, as well as the size
of different orders and the consequences of rebate schemes, then the conscious
mind quickly bows out; the number of variables exceeds the amount we can
keep in our working memory; they are simply not simultaneously available for
processing.
The consequence is that we are not able to “see” all the different relationships
and the effects of their mutual dependencies in one “picture.” To work around
our mental limits, we have to use time, ponder parts of the problem separately,
structure it in subsets that can be treated as single elements, and so on.
Sometimes, such long-time “submersion” in
a complex problem alone allows us to organize it sufficiently to get an overall

grasp on it and see a solution, but if the problem is complex enough, we need to
commit intermediate thoughts and analyses to paper, use other tools (like
computers!), or resolve the matter by dividing the task among several people.
Elements in Problem Solving

However, that is only part of the problem. In addition to the limits to our sheer
processing capacity, there are many other constraints involved in problem
solving as well. To examine them a little closer, we can use a three-stage model
for problem solving proposed by Ellis and Hunt (1989, p. 219): understanding
the problem, generating hypotheses about solutions, and testing and evaluating
solutions. Since Ellis and Hunt refer mainly to laboratory experiments, they
omit one stage that is very important in real-life situations, especially in
organizations: the procurement of sufficient information. For real-life problem
solving, we can therefore propose the following stages (stages 2 to 4 are adapted
from Ellis and Hunt 1989, p. 219):
1. Procuring information
2. Understanding the problem
3. Generating hypotheses about solutions and selecting among the alternative
hypotheses
4. Testing and evaluating the solutions
The logic of this ordered list of stages notwithstanding, as Ellis and Hunt
emphasize, most problem solving is an iterative activity where we cycle through
the different stages and even jump back and forth between them.
Procuring Information
While simple in the laboratory setting, where the experimenter furnishes you
with the setup, the procurement of information is much more complicated in
real life. Often, we do not even know exactly what kind of information we need,
and if and when we find out, it too often turns out that we cannot obtain much
of it. Procuring information is indeed a main organizational activity; there are
entire departments devoted to it. The accounting department, for instance, has as
its sole purpose to keep track of the economical performance of the
organization, market analysts are occupied with collecting information about the
outside world, production
planners try to provide information about expected production schedules for

both sales and management, and so on.
In addition, we almost always have a simplified (and sometimes quite wrong)
perception of the problem itself and the causal relationships involved in it. We
therefore often procure the wrong information set, and we also generally tend to
believe that the information we have is more complete than it actually is.
Elementary cognitive errors, such as misinterpretation of other people’s
behavior (including oral and written communication) may compound the
problem.
Understanding the Problem

As Ellis and Hunt emphasize, “Before a problem can be solved, it must first be
understood.” Before we can start to seek out a solution, we must have a
sufficiently clear picture of the problem. Research shows that an adequate
mental representation of the problem is very important to finding a good
solution, or even finding a solution at all (Ellis and Hunt 1989, Anderson
1990).
Understanding the nature of a problem may also involve understanding its
causes. They are not always obvious, they are not necessarily objective in the
sense that they seem the same regardless of perspective, and they are certainly
not always unitary. When management rationalists Kepner and Tregoe (1965, p.
17) maintain,
Here it should be pointed out that every problem has only one real cause. It may be a
single event that produces the unwanted effect, or it may be a combination of events and
conditions operating as if they constituted a single event.
they display an attitude that may be valid for simple engineering problems, but
is grossly inadequate for the complexities of human life. Consider, for example,
the editorial by Garrett Hardin in Science referred to by Weick (1979, p. 68),
with the title “Nobody Ever Dies of Overpopulation.” It treats the catastrophe
that occurred when East Bengal (now Bangladesh) was hit by a cyclone in
November 1970, and 500 000 people living on the low islands in the river delta
were killed. A similar catastrophe hit in April 1991, with about 150 000 dead.
Now, we may ask: What caused the death of these unfortunate people? Was it
the cyclone, an unpredictable, natural disaster? Was it the lack of dikes that
could keep the water out? Or the lack of cyclone-safe shelters? Or was it
perhaps the fact (as Hardin argues) that overpopulation has forced people to live
in places where even an ordinary storm constitutes a grave danger? And is it
possible to devise a single best solution to ensure that such a catastrophe does
not repeat itself in the future?
By choosing perspective, or problem representation, then, we also decide

which information is relevant and what kind of causes will be allowed for
consideration. By choosing evaluation criteria, we decide what kind of solutions
will be considered “best.”
Generating Hypotheses About Solutions

When we have acquired at least a tentative understanding of the problem space
and have constructed a preliminary problem representation, we start the hunt for
a solution, employing one of two main strategies. In some instances, especially
when solving technical problems (involved in, for instance, the construction or
computer programs), we will use an algorithmic strategy, which consists of a set
of rules or procedures that ensures a solution. In everyday life, however, most
problems do not have algorithmic solutions, and we have to use a heuristic
strategy—a “commonsense,” “rule-of-thumb” approach, a problem solving
method that works and is used in practice, regardless of whether we know why
it works; indeed, we may not care to know at all.
To illustrate the difference, Ellis and Hunt (1989, p. 220) use the example of
locating a friend with the name J. Smith in a large city, where there are 41 J.
Smiths listed in the telephone directory. An algorithmic strategy for solving this
problem is to start at the top and call all J. Smiths in consecutive order until the
right one is found. Normally, such a strategy will not appeal to us, however—it
is much more likely that we will use our knowledge about his occupation and
make an educated guess about where in the city our man is likely to live. This
way, we can considerably reduce the number of people we need to call (if our
guesses are about right, which they often are in everyday life). Everyday
solutions, of course, often include both algorithmic and heuristic elements (Ellis
and Hunt 1989, Anderson 1990).
Regardless of the strategy, the solutions we generate also depend on
background knowledge, past experience (a problem that is unsolvable for most
people may belong to the routine repertoire of an experienced specialist).
Values are always of substantial importance, both during problem definition and
in searching for a solution. A liberal left-winger and a Christian fundamentalist
would almost certainly have very different problem definitions if their teenage
daughters became pregnant without being married, and would quite likely arrive
at different solutions—such as an abortion for the liberal, and a hurried marriage
for the fundamentalist. Another very significant determinator for the generation
and selection of solutions is what we know or believe about their consequences,
including risk assessment.
Testing and Evaluating the Solutions

That brings us to the last stage in problem solving, which involves choosing the
actual solution to be executed—a simple step as long as the problem and the
possible solutions are well understood, and the criteria for judgment are clear. If
price is the only criterion, for instance, it is simple to choose among the
alternatives for a transatlantic flight presented by a travel agent. Unfortunately,
however, the situation is seldom so easy except for simple and fairly
inconsequential decisions. Important decisions tend to be complex and may
involve both ambiguous problem definitions and solutions that are difficult to
compare. They may also have consequences and benefits that are contested, or
the criteria for judging the possible solutions may be unclear or have unclear
priorities (or even both).
From Maximizing to Satisficing: Accepting Simplification

The picture that shines through this short discussion is not exactly that of a
supremely rational being, analyzing all relevant facts, choosing the best among
all possible solutions, and carrying it out flawlessly. And there are even many
more cognitive pitfalls in the various stages of problem solving, especially when
we operate under uncertainty (see, for instance, Kahneman, Slovic, and Tversky
1982). It is indeed simply impossible for humans to find the one best solution
(given that such a solution exists in theory, which it often does not) for anything
but the very simplest problems.
With what we now know, it seems preposterous even to suggest it. However,
we need only go back 50 years to find this illusion widely accepted as fact,
upheld both by the economists’ image of “economic man” and by the “rational
manager” of the management theorists. These concepts did not receive any real
dents until the publication of Herbert A. Simon’s Administrative Behavior in
1947, and even if they are now academically discredited, they linger on in the
simpler parts of the management realm and flavor many an offering from the
more archaic breeds of management consultants.
Simon’s great, commonsense realization was that man operates with limited
information and wits in an exceedingly complex world, and perforce has to
simplify, to operate with a bounded rationality, to satisfice—not maximize
(1976, p. xxviii, italics in original): “Administrative theory is peculiarly the
theory of intended and bounded rationality—of the behavior of human beings
who satisfice because they have not the wits to maximize.” As a contrast to
economic man, Simon defines administrative man, who “. . . makes his choices
using a simple picture of the
situation that takes into account just a few of the factors that he regards as the
most relevant and crucial” (1976, pp. xxix–xxx). Simplification is indeed our
basic weapon against complexity, and is used also when tools and
organizational measures are brought into service. We simplify our models of the
world quite simply because rich models are too complex to handle—and we
then convince ourselves (and others) that there are actually just one or two or
three factors that “really count.”
Research on judgment under uncertainty is indeed rife with examples of how
we make our judgments on the basis of information that is superficial, the most
readily available, or easiest to think of, and how we are deceived by our
intuitive interpretations of our immediate impressions of reality, even when we
have solid theoretical and factual knowledge to guide us toward more correct
solutions (Kahneman, Slovic, and Tversky 1982; Rachlin 1989). We also have a
strong tendency to let concrete, immediate urges and experiences displace or
overrule abstract knowledge about later, possible consequences—such as
continuing to smoke even while acknowledging that it may lead to cancer at a
later date, or driving our car very fast when late for an appointment, even
though we know it increases the likelihood of a dangerous accident, with
consequences that would be way out of proportion to the importance of the
appointment in question.
Simplification is usually a combination of a conscious process (consciously
choosing some variables over others) and an unconscious or intuitive one (just
regarding some variables as “naturally” important or unimportant). The criteria
in both cases can be questionable; for instance, it is not uncommon to have an
over-representation of computable variables in organizational decision proces-
ses. Computable variables are convenient to handle and have the added advan-
tage of appearing to be very objective and accurate, even when they are not—
follies like market forecasts five years ahead reported to the tenth of a percent
are routinely presented to credulous audiences in the most serious companies.
To be able choose the right variables in real life, those that are really
important, we need considerable experience with the problem domain. If we
enter a new field of work or knowledge, or suddenly have to live in a country
with a very different culture, we need a period of adjustment, until we
internalize the essentials of the new setting.
As noted above, Simon’s view of administrative man as satisficing rather
than maximizing delivered the first real blow to the glossy picture of the
manager as the supremely rational being presiding over tidy, rational
organizations. Another strong blow was delivered by Henry Mintzberg (1973),
who found that the old picture of the manager as a reflective, systematic planner
was utterly false. He discovered that managers are interrupt-driven, strongly
oriented to action, and dislike reflective activities. They prefer oral to written
communication, and
work chiefly through formal and informal meetings and telephone

conversations. In addition to the simplification such informal work habits imply,
another interesting way of thinking is suggested, which came to occupy more
and more of Mintzberg’s attention: unconscious processing and intuition.
Unconscious Processing and Intuition

In pursuing this line of thought, Mintzberg involved himself in a strong (but
friendly) controversy with Simon, who held that even if man’s rationality is
bounded, we still talk about conscious rationality, and not about obscure sub-
conscious processes. Simon also contended that human thinking is made up of
programmed sequences, sufficiently similar to computer programs that
computers can be used to describe and simulate them (Simon 1977).
The notion of unconscious processing and intuition is indeed interesting but
still controversial. It may amount to the next step in the “derationalization” of
decision-making that Simon started, and certainly deserves a discussion at this
point.
As noted above, our conscious mind becomes bogged down fairly quickly as
variables are added or as the relationships between them are made more
complex. Our working memory does not suffice. Experience suggests, however,
that subconscious thought processes can integrate and weigh a larger number of
variables. A number of sayings and proverbs allude to this—we talk about
“sleeping on a problem,” about “problem gestation,” about “digesting”
information or dramatic experiences, and so on. Most of us have, for instance,
probably experienced the anguish of having to face pivotal decisions about our
lives or careers, and we know that we do not rely entirely on rational analysis in
such circumstances. We “ruminate” on the decision until we have an answer
that feels right “in our stomachs.”
Incubation effects (more rapid problem solving after a period of thinking of
something else) have been demonstrated experimentally (Anderson 1990)—
especially for problems requiring sudden insight to solve (such as the problems
presented in books or magazine columns on recreational problems and riddles).
Subjects who had a break for some hours after studying the problem for a little
while, and then resumed, had a higher percentage of success than those who did
not have a break. Ellis and Hunt (1989) recommend such breaks to evoke the
incubation effect as practical advice for problem solving, in spite of the fact that
we do not thus far have a satisfactory explanation of why it works.
Many scientists have reported similar personal experiences when working
with and suddenly solving particularly intriguing and hard problems—
problems they may have struggled with for long periods of time (Goldberg
1989). Kekule von Stradonitz’ discovery of the molecular structure of benzene
(which had puzzled chemists for many years) is perhaps the most famous
example: He was dozing off, half-dreaming about strings of dancing carbon
atoms. Suddenly one of the strings snaked back on itself, forming a ring—at
which point he woke up with a jolt, the benzene ring clear in his head (Asimov
1975).
In his annual hour with the Nobel Laureates on Swedish TV, the host, Bengt
Feldreich, always ended the program by asking the participants if they believed
in intuition—and the majority of them invariably did, citing their own
experiences with that sudden flash of insight. They all agreed that it feeds upon
years and years of experience and knowledge accumulation, and that the
problem-solving process is only partly conscious—the solution comes as a
sudden culmination of a combination of conscious work and an inscrutable,
subconscious process beyond inspection. This is in harmony with the views of
most of the people who write about intuition, whether they think, like Simon
(1989), that intuition is essentially a conscious recognition of established
patterns, or, like Rowan (1989), that it is knowledge gained without rational
thought: they all agree that intuition does not come totally out of the blue.
Rather, it requires a solid foundation of factual knowledge and experience.
As a process not controlled by the conscious self, subconscious processing is
of course subject to influence from all kind of facts, judgments, conjectures,
desires, and other emotions that our mind harbors. Nevertheless, the intuitive
insights that arise time and again from the depths of our mind are often worth
more than weeks and months of “rational,” conscious analyses. Indeed, as
Simon (1989) notes, analysis is often an activity that experts carry out only to
check the validity of solutions found almost instantly through intuition. In his
now famous article “Planning on the Left Side, Managing on the Right” (1976),
Mintzberg describes how such processes and their outcomes in fact seem more
important for managers than conscious, rational analyses—even though
managers and management consultants usually hold forth the banner of
rationality both as an ideal and as a description of their way of working.
The more complicated the problem is, the more likely it is that the solution
will not be found through rational analysis, but through “weighing,”
“digestion,” and “sleeping on it”—mediated thorough informal discussions with
people who also have “thick” information on the subject. The interest for
intuition as a management device has since then spread, and is now definitely
on the increase (Agor 1989), and new evidence is steadily surfacing (Weiss
1990).
Nowhere in the organizational world is the dependence on subconscious

processes so apparent as in choosing main strategies for an organization. A
common explanation for AT&T’s miserable failure in attempting to enter the
computer business in the last half of the 1980s, for instance, was that no one in
the management had a “feel” for the special characteristics of the computer
market. One may argue that they did not know the market, and the answer is
both yes and no—they no doubt had the information available in reports,
memos, and presentations. But, they had not internalized that knowledge—they
had not been able to digest and integrate it, as only a submersion over quite a
long time can bring about. Therefore, they did not have any gut feeling about
the matter and could not come up with a vision or sense of direction. 2 In short,
they lacked precisely what is generally acknowledged to lie behind Microsoft’s
phenomenal success: The longstanding, total submersion in the computer
industry and the deep understanding of the technology and that characterize Bill
Gates, Steve Ballmer and the other key figures in that company. Strategy needs
vision, and vision is not obtained through calculations—not even in business.
Of course, intuition may be wrong, just like rational analysis. If vital facts
pass unknown, no amount of subconscious integration can make up for it.
Combined with the process of groupthink it can keep a set of beliefs about the
world alive in a group of people long after it has ceased to be true, blocking
“unpleasant” facts and lead to decisions out of touch with the real world.
Intuition is moreover easily influenced by our own feelings, hopes, and wishes.
Its roots in the subconscious, its integrating powers, are thus both the source of
its strengths and of its weaknesses. While it can no doubt be powerful for
finding solutions to complicated problems, it is therefore not a viable tool for
reliable execution and coordination of the everyday chores that also fill our
lives. There, we have to rely on our limited, but still powerful conscious mind,
and rather simplify, decompose, and distribute where the complexity of a task
exceeds our powers.
The Delays of Deliberation

The time we need to reach a decision can vary enormously. If it is a question of
a minor, recurring problem, we usually have the answer on hand, and a decision
can be made in fractions of a second. If the problem at hand is totally new, of
major importance, and involving a lot of parameters, we will often need to cycle
repeatedly through the different stages of
2Note how all these common life expressions about problem solving and direction finding allude not
to the conscious mind, but to instincts and the autonomous nervous system (which, among other
things, runs our bowels for us).
the problem-solving process, and it can take a long time to reach a decision. A
further reason for this is that we often like to obtain the “gut feeling” described
earlier and convince ourselves that it points in the same direction as our
rationally derived answer. In other words, we feel a need to bring our
unconscious, integrating abilities into play.
These limitations clearly restrict the number and magnitude of the decisions
that any one individual can handle, thereby constraining our freedom of
organizing. They are of course compounded by our limited ability to absorb
information, and are especially important as limitations for the centralization of
control: In essence, you can only have first-hand control where you can decide
yourself—decisions delegated mean control surrendered, even if you try to
uphold control by orders, rules, regulations, or law. This is a basic dilemma
known to every entrepreneur—expansion means loosing the total control you
have as owner/manager of a tiny start-up.
Our Communication Bottleneck

Human beings have many senses registering information about both the world
around us as well as our own states. Ordinarily, we count five, but there are
many more: We sense temperatures, air pressures, acceleration, and the
positions of our limbs, to mention a few. Our senses are capable of receiving
and processing an astounding amount of information. In computing terms, our
visual system processes raw data at the rate of hundreds of megabytes per
second. Moreover, our additional ability to quickly scan the picture the cortex
presents us with, and pick out and classify its important features in real time, is
nothing short of an information-processing miracle. Simultaneously, our brain
can also receive and digest information about the states of the muscles in the
body, and coordinate their movements in real time with immense precision. It is
only when we try to build walking, self-guiding robots that the prodigious
information-processing capacities of the brain are really driven home to us. The
day when a two-legged humanoid robot can compete with human downhill
skiers in the Hahnenkam competition in Kitzbühl’s notorious “Die Streif” track
is indeed far off (but not entirely unthinkable—such a day may come).
However, the communication that builds organizations is first and foremost
verbal. It is the spoken and written word. But our rate of acquiring and
disseminating verbal information is painstakingly slow compared to our
processing of pictures. The raw processing power of the aural system is only a
fraction of the visual, and when we count only the factual information contained
in speech, it is only a small fraction of that again. Of
course, if we take into account the information contained in voice volume and
inflection (which is frequently as important as the actual words), the amount of
processed information in digital terms is much greater and perhaps similar to the
interpretation of music.
Normal speech (and thereby listening) happens at a net rate of about 150
words per minute, not counting pauses for thinking, groping for words, etc.
Very fast talk approaches 250 words per minute, but by then both speaker and
listener will begin to experience problems. Sustained speaking for longer
periods of time (for instance, a lecture) probably averages around 100 words per
minute.
Reading is faster, but not so much—at least for factual prose. Most of us
level off around 300 to 400 words per minute even when really concentrated
and absorbed by what we read (like when devouring a really exciting novel).
Taking into account our normal lapses in concentration when reading factual
prose (with no plot or drama to capture our primate minds), it is difficult to
average much more than 100 words per minute for longer periods of time
(several hours). Assuming that the average word has 6 characters, that amounts
to 10 characters or bytes per second (or 0.00001 megabytes). Even
accomplished speed readers cannot go much beyond 1000–1500 words per
minute (about 100–150 bytes per second, or 0.00010–0.00015 megabytes), and
research indicates moreover that such reading is not very effective—it
resembles most of all “skimming,” giving an overview of the text without a
concern for detail (Barber 1988).
Writing is the slowest means of verbal communication. Quite apart from the
process of formulating the text, the physical writing process itself is a plodding
activity. Until voice-recognition technology has developed to the point where a
machine can reliably take rapid dictation and render it into text, the typewriter
and the computer keyboard are the fastest devices we have available. An
experienced (but autodidactic) 3–5 finger typist like myself typically enters text
at around 25–30 words per minute, not counting error correction. An extremely
fast touch typist can exceed 125, about the rate of normal speech.
The immense difference in speed between picture processing and verbal
communication is the main reason for the efficiency of graphical presentation of
data. Presenting data as pictures and graphs taps the enormous bandwidth of our
visual system, and makes it possible to absorb both quantitative information and
the interrelations between variables much faster and more accurately than
through text and numbers. There is much research going on along these lines,
not least for military applications. For fighter pilots being guided toward enemy
aircraft or around enemy defenses, it will, for instance, be much easier to have
the changing direction and altitude merged graphically into a curving tunnel on
a screen (and
then proceed to fly “through the tunnel”) than to be presented with numerical
data and compass bearings.
The extremely narrow bandwidth of verbal communication places serious
limits to the achievable levels of organizational coordination and control. Every
minute, large amounts of information are created or received in any
organization, and to be acted upon much of it needs to be aggregated, processed,
communicated, and presented. Much must also be stored for future reference.
As the rates above indicate, such work is very labor-intensive—and the result is
that only a fraction of the received information is ever processed, only part of
the processed information reaches the people who need it, and they again will
only acquaint themselves with a selection of what they actually receive.
The results are familiar for all who work in organizations: Decisions are
made on shaky foundations, changes in the environment go unnoticed or are
acted upon too late, coordination is often inadequate, there is much duplication
of efforts, and different parts of the same organization may even be working
against each other without realizing it.
As Mintzberg notes in his introduction to The Structuring of Organizations
(1979), coordination is effortless only as long as the number of people that must
coordinate their actions remains well below ten, and it is handled through
continuous and informal communication. At that level, coordination is hardly
noticed as a separate task—it just comes naturally. As soon as the number of
people climbs into double digits and beyond, coordination and control becomes
the most pressing operational problem, and a wide array of tools and techniques
are brought to bear on it—schemes for division of labor, organizational
structures, delegation of authority, coordination meetings, reporting, accounting,
and so on.
Almost all of the planning, supervisory, and administrative work carried out
in an organization is an expression of the continuous fight to stay in control of
events and coordinate the various parts of the organization and its interaction
with the environment. These coordination problems, rising from the limitations
of our basic communication abilities, constitute one of the most iron-clad
constraints on operational effectiveness and efficiency in all organizations
above the minimum size.
The Constraints of Space and Time

Parents who have tried in vain to call in children playing outside have no
problem appreciating the fact that the unaided human voice has its limits. A
conversation is difficult to keep going if the distance exceeds several meters,
and even a primal scream does not go far on a day with normal
wind and background noise. People living in mountainous terrain, like the Swiss
and a few others, have devised rudimentary “languages” or code systems
consisting of patterns of high-pitched tones or whistles that bear from hill to hill
or across ravines, but even under exceptional conditions, their range is limited
to a few kilometers. Our basic communication abilities thus allow for local
communication only, mostly with one or a few persons at a time. Without
special surroundings, such as an amphitheater (which has very favorable
acoustic properties 3), even a Stentor or a British sergeant major would have
trouble addressing more than a few hundred people at a time.
Our vision does make it possible for us to receive information over great
distances when there are no obstacles—after all, the naked eye can see stars
trillions of kilometers away—but our means of replying are not on the same
level. Some of the earliest techniques for communication tried to remedy this,
by using visual aids—such as the smoke signals of the American Indians and
the beacons of the Vikings—that can be seen from great distance. In the century
before the telegraph was invented, several European nations built national
systems of semaphore lines. Their aural counterparts are the “talking drums” of
certain African tribes.
Time is an even more merciless enemy of communication than space. Writers
sometimes contend that one of their characters “left his/her words hanging in
the air,” but apart from this strictly literary storage mechanism, all unaided
human communication must take place in real time. Once a word is uttered or a
gesture performed it is also a thing of the past, and it may be remembered,
distorted, disputed, or completely forgotten.
The fact that unaided human information exchange can only take place
locally and in real time puts severe constraints on the possibilities for building
and sustaining large organizations. The only means of communication over
distance is then the dispatch of messengers, and the messengers have to rely on
their memory to ensure that the message reaches its destination uncorrupted.
The use of messengers also brings in the question of authenticity: When you
speak to someone in person, you know immediately that the message is
authentic. When you have to rely on messengers, you never know if the
messenger intentionally or unintentionally is misrepresenting the words of his
master. As Eriksen (1987) shows, history is rife with examples of messengers
having decisive influence on historic events. The delay or liquidation of
messengers can also have profound consequences. And when the messenger has
to spend
3The amphitheater is actually a very advanced acoustic device. The ancient Greeks built many, and
even in the largest, which lies in Epidaurus and can seat 22 000 people, the actors on the stage can
be heard by everyone in the audience without any artificial amplification. The theater is still in use.
not only hours or days, but maybe weeks and months on the road to reach his
destination, events unknown to him could already have changed everything by
the time the information is presented to the receiver.
We must remember at this point that physical travel has been very slow all
the way up to the middle of the nineteenth century, especially over land. In the
year 1800, for instance, it took six weeks to go from New York to Chicago
(Chandler 1977). Until the advent of large, swift sailing vessels, transport of
cargo was also expensive and time consuming. Land transport remained slow
and costly until the railroad revolution in the middle of the nineteenth century.
Throughout most of our recorded history, therefore, long-distance trade has
concentrated on high-value items, such as metals, spices, fur, and silk.
The strains on the communication system were prominent in every major
empire in history, and large works were undertaken to speed the passage of
messages. The Roman roads are well-known (total length exceeded 300,000
kilometers and main roads were paved with stone), the Incas built roads as well,
and the Mongols under Genghis Khan built a vast system of posting stations,
where the Khan’s express messengers could change horses on their breakneck
journeys to and from the Khan’s command posts. Communication technology
has thus for a long time been a key factor in our ability to extend our
organizations beyond the local community.
A striking example of the importance of enhancing our natural
communication apparatus was demonstrated by the attack on Iraq on 17 January
1991. It opened with a massive air strike against radar installations, command
and control centers, and communication lines. The rationale was that, if
successful, orders for counterattack could not be given, information about the
allied attacks and their effects could not be collected, and even consultation
between different Iraqi command centers would be impossible. With only the
real-time, local communication capabilities of the unaided human available,
modern armies and air forces are instantly reduced to a fraction of their
theoretical strength, even without any other material or human loss. The
subsequent development of the war and the total collapse of the Iraqi army
illustrate this point well.
Wishing, Wanting, and Feeling

However sophisticated we have become, however much we hide behind our
machines, our natural sciences, and our rational facades, we are still beings of
flesh and blood, with complex minds, full of instincts, ambitions, hopes, fears,
and desires. Some of our emotions are raw and basic, others refined and even
noble, but the mixture is volatile and always prone to
produce unpredictable effects. Our secret inner lives can be pretty wild and
untamed. As the zoologist Desmond Morris noted in the introduction to his
widely popular book The Naked Ape (1967, p. 9),
. . . in becoming so erudite, Homo Sapiens has remained a naked ape nevertheless;

in acquiring lofty new motives, he has lost none of the earthy old ones. This is
frequently a cause of some embarrassment to him, but his old impulses have been
with him for millions of years, his new ones only a few thousand at the most—and
there is no hope of quickly shrugging off the accumulated genetic legacy of his
whole evolutionary past. 4
Although each one of us (presuming a minimum of honesty) can confirm this

through simple introspection, the importance—or even the mere existence—of
emotions has to a large degree been ignored in the literature on organization,
surfacing mostly in discussions of motivation, work satisfaction and stress (as
noted by Hochschild in the preface to Fineman 1993). When you first notice
this, it is a bit puzzling—when you reflect upon it, it starts to look like a very
serious defect in organization theory and an embarrassment to organization
theorists. As Fineman notes on the first page of his introduction to Emotion in
Organizations (1993, p. 1): “Writers on organization have successfully ‘written
out’ emotions, to the extent that it is often impossible to detect their existence.
A scan of the indexes of recent texts on organizational behavior reveals no
direct entries under ‘feelings’ or ‘emotions’.”
This preoccupation with the rational side of organizing seems even stranger
when you contemplate that there is indeed a vast literature on human emotions
and their significance in social life. Not only can we look to the discipline of
psychology; the study of history is also rife with examples of how emotions
have decided or heavily influenced the outcome of social and political conflicts
with far-reaching consequences.
Going still further, we can draw upon the literature of the world, or indeed
the total body of art produced throughout human history, as a powerful witness
to the sway that emotions hold over human actions. In stark contrast to the
modern classics in the field of organization theory, which treat emotions either
cursorily or not at all, Plato was very concerned with subject. In his three books
The Republic, The Statesman, and
4Interestingly enough, today, we tend to describe this our emotional self as our "human" aspect—as
opposed to our logical faculties, which we tend to perceive as machine-like: hence the
contemporary fascination for the impulsive, emotional, expressionist personality, capable of loving
and hating with equal intensity. People ruled by logic and reason are frequently depicted as cold and
indifferent to other people’s sufferings. This represents a turnabout from the time of the ancients,
who looked at our emotions and instincts as something generally despicable that resembled animal
nature, hailing reason and logic as our virtuous, “human” aspect – that part in us most resembling
God.
The Laws—arguably the first books on organization ever written (around 2400
years ago)—we find that control of emotions, especially destructive emotions, is
a central theme in the struggle to achieve justice.
Plato concluded that no one could rule justly without an understanding of
what justice really was; in other words, the ruler must be so thoroughly trained
in philosophy and so advanced in his thinking that he would be totally governed
by reason, unmoved by all kinds of desires, always working for the best of his
subjects and never serving his own interests. This view was initially presented
in Gorgias, but received its full expression first in The Republic (where the
philosophers are appointed rulers) and later in The Statesman (where rule is
effected partly by law and partly by philosopher-statesmen) and The Laws
(where an almost immutable set of laws is set to provide the rule that fickle
human nature cannot).
Plato, then, who was much more uncompromising in his fervor for reason
and logic as the governing principles for organizing than the modern classics of
organization theory ever were, at the same time fully realized that it was the
emotional side of human nature that was his worst enemy, and devoted large
parts of his works to discussions of how the unwanted part of those emotions
could either be eradicated, suppressed, or controlled. He certainly also realized
(at least as he grew older) that his goals were utopian, and that the best one
could do in practice was to enlighten prospective politicians as much as
possible, hoping that this would moderate their behavior when in office. At
least, that was one of the practical functions of his Academy, which attracted
students from throughout the Greek world.
Indeed, then, as Fineman suggests, the time is long overdue for bringing the
subject of emotion (back) into the discussion of organization, though the scope
of this text does not permit a detailed discussion of the subject. However, I
believe it is too important to leave out altogether, and all the more so because its
importance seems to be grossly underestimated in the debate on the use of
information technology in organizational settings. I will therefore include those
aspects of emotions that I think are the most important in the following
discussions.
As Flam (1993) points out, our emotional self is constantly interfering with
the rational and normative parts of our mind. 5 Fear, for instance, which is the
subject of Flam’s discussion, can cause an individual to rationally plan and
perform actions that are in direct conflict with the
5Flam herself contrasts this trichotomy with Etzioni’s (1988) merger of norms and emotions.
However, the division of the human consciousness into a rational, a normative and an emotional
part is an old and established way of conceptually dividing human consciousness into partly
conflicting selves. For instance, it roughly corresponds to Freud’s ego, superego, and id.
normative self, and pride can cause an individual to obey the normative self
even if it means death, thus overriding rational deliberations. As we all know,
love and desire can also have devastating effects on rational behavior.
In a small hunter/gatherer band, our most probable “natural” state, the strong
influence of the emotional self poses few organizational problems. On the
contrary, organization in such a band is indeed structured around affectionate
relationships, as well as real and ritual family relationships, myths, and religious
conceptions—and emotions constitute a large part of the glue of such
relationships. Emotions here ensure the stability and predictability of both
structures and lines of authority, and can thus be said to constitute a
fundamental human organizational tool.
That this basic mechanism is still very important can be seen from the fact
that family ties continue to be of great importance in most societies of the
world, both in private organizations (businesses) and in politics. Even in modern
democracies, family ties continue to have considerable significance. Another
class of emotional bonds, friendship ties, is also very important, and in the
headlines of newspapers and newscasts we are constantly reminded of the
immense power of tribal (ethnic) identification—both for the better (national
unity in crisis or celebration) and for the worse (racial discrimination, war and
ethnic cleansing).
As a tool for organization, emotions are definitely most appropriate in the
small group—such as the hunter/gatherer band, where it probably awarded an
evolutionary advantage. In larger organizations, however, emotions may give
rise to problems such as factionalism, when loyalties and interests defined
locally clash with those defined on higher levels or elsewhere in the
organization.
Emotions also make us less reliable in many ways, and harder to predict.
They often bend our memory, shift our focus of attention, create interpersonal
tensions, give rise to tactical behavior and generally mess up our performance as
organization members. Organizations become not only rational means to
legitimate ends, but also, as Morgan (1986) shows, arenas for display of
ambition, pursuit of individual goals and fights for control. They become
instruments of domination and vehicles for status. Some of the most extreme
demonstrations of this phenomenon are presented by the takeover kings of our
modern market economies. Driving their aggressive manipulations are the same
desires, ambition, and thirst for power and status that drove Genghis Khan,
Alexander the Great and Harold the Fair-haired. Some of the modern warriors
are no less ruthless than their ancient brothers-in-arms either, if judged relative
to the accepted standards for chieftain conduct in their respective epochs and
societies.
However, it is perfectly in line with the dual edge of our emotions that they
sometimes make people extraordinarily reliable, as in times of war or other
great danger, where, as already indicated, social bonding and emotional ties can
induce people to remarkable selflessness and courage even in the face of torture
and death.
Coping With Reality

So, as naked animals, we have our limits. We have nevertheless been able to
survive and prosper in a complex world even without our present sophisticated
tools. Indeed, we still have to trust our basic capabilities for many of our
activities. How do we cope with the complexities of life, with the avalanches of
information of all kinds that hit us every minute, both through raw sensing and
through symbol interpretation? What about the innumerable large and small
decisions we have to make every day—like figuring out which bus to take to
work, what to eat for dinner, what to wear, whom to greet (and how), and so on?
Why are we not permanently bogged down in decisions? If we had to analyze
every new situation, every new challenge from scratch, we would be left in
constant bewilderment—our brain would simply experience permanent
overload. Luckily, we have some effective strategies for coping. The two most
prominent ones are imitation and the compilation of mental sets.
Imitation
Imitation is the most obvious of the two. It lies at the very base of human
learning and has been discovered to occur even in infants only a few days old
(Hofer 1981). The socialization process is nothing but a transfer and subsequent
internalization of standard procedures and norms for everyday life in society.
The standards are not immutable, but they normally show great resilience
against sudden change. Much of what our great-grandparents considered proper
conduct is still endorsed by the great majority. By accepting established norms,
we can relieve ourselves of an enormous amount of decision-making—we greet
another person without thinking about how to do it, we do our shopping without
fussing about how to behave toward staff and fellow shoppers, we automatically
behave differently in a funeral from during a rock concert, we know how to
conduct meetings in our local residents’ association, and so on.
In organizations, we learn the local mores as we go, internalizing their
traditional way of conducting business. The importance many people accord to
this organizational socialization can be judged from the fact that numerous
organizations have adopted a practice of only recruiting
managers internally—thereby avoiding the potentially disruptive consequences

of putting people with deviant norms in positions of power. The downside of
this approach is the risk of groupthink and blindness to alternatives, which can
be very dangerous—especially in rapidly changing environments.
Organizational structures themselves are, as Stinchcombe (1965)
convincingly argues, almost always imitations, most often of previously
established organizations in the same line of business—copying organizational
structure and business conduct from existing, successful operations. It is of
course a lot more convenient just to roll out something one knows will work and
is familiar with, than to use a lot of time and energy constructing something new
and untried. Your financial backers may not approve it, either. On the other
hand, you can also (as always in human affairs) find examples of the opposite:
experienced people breaking out from a traditional operation to start a
competing organization with a novel organizational approach as their main
weapon.
In the same paper, Stinchcombe furthermore points out the conditioning
effects of the prevailing social structure, which affects all contemporary
organizations to a considerable degree. According to Meyer and Rowan (1977),
organizations also tend to inherit formal structure from their society’s
institutional myths—often resulting in a formal structure that is out of step with
the actual, day-to-day work procedures.
Imitation is a very economic way of building an inventory of responses to
common problems and events, and it allows knowledge to accumulate and
spread with significant speed.
Mental Sets
To a newborn baby, the world must be a bewildering chaos of light, patterns,
and sounds. Although it can already recognize some sounds heard before birth
(especially the heartbeat of its mother), it has few possibilities of understanding
what it senses—it has no established pattern “library” to which to relate its
impressions. Before it can recognize objects and sounds, it must build such a
library, synthesizing similar, concrete, perceived patterns into generalized object
classes, which can then provide the templates needed to recognize a particular
instance of the class and ascribe the proper rules of behavior to it. It is exactly
the class concept that allows us to recognize a particular car as a car, even if it is
a model we have never seen before. We also know that we cannot expect it to
stop on a dime—we assume that it will exhibit the general properties of its class,
of which an approximate inertia and braking power are among the most
important for pedestrians and drivers alike. The classes, their properties, and
their relations to each other must be extracted from what we see and
experience, and then stored in memory to allow later use. As Cohen notes
(1980, p. 116), “If we are to be able to apprehend the world around us, this
apprehension must be guided and shaped by our cumulative store of experience.
In short, memory may be said to be the organ of perception.”
Similarly Hofer (1981, p. 138-39) notes that “within the newborn’s
capabilities . . . lie all the building blocks for the mind as we know it. The
sensory plan by which certain information is selected, together with the related
action pattern, may be referred to as a schema . . . The essence of such processes
is to form inner representations of the outside world and to make ‘predictions’
as to the outcome of actions directed at that world.”
The speed with which the child advances in its early synthesizing efforts, its
establishment of schemas, is a proof of the very powerful pattern recognition
and integrating faculties of the brain. As the basic, physical patterns are
synthesized, a child must also build the even more subtle models of the objects’
properties, their normal behavior, the settings in which they occur, the relations
between different objects, and so on. When it slowly realizes its own position as
a separate entity with a certain freedom of action, this exhilarating fact must be
integrated with its views of the outside world. It must start to build its own
implicit theories of action—its own theories-in-use (Argyris 1980).
As we advance from the concrete, physical level to the abstract and symbolic,
the synthesizing process becomes more and more demanding. It takes many
years to build an adequate set of schemas for understanding human behavior
and the proper responses in different situations, and quite a number of people
seem to have problems ever acquiring a suitable understanding of the intricacies
of human interaction. Likewise, establishing an adequate understanding of a
branch of science is no easy matter, and beyond the reach of many people. You
get a renewed taste of this basic experience every time you enter a new field of
knowledge: You are not able to judge what is important and what is not, or see
what constitutes quality and what is more doubtful. You have to “get your
bearings” first, to develop a “map of the terrain,” a “feeling for the subject,” so
that you can judge and remember by relating to things you already know.
As we grow older, we build up an extremely rich complex of schemas
covering the different aspects of the world and our lives, from the most minute
details to a general world view. The schemas can relate to objects, persons,
animals, acts, sensations, symbols, and so on, or combinations thereof. They
tend to be organized in clusters, covering the totality of common situations. If
we follow Goffman’s (1959, 1970) analysis of human interaction and accept his
metaphor of the theatrical performance, it seems natural to label these
amalgamations of schemas as mental sets. A mental set defines the totality of
the situation we confront and tells us
what kind of objects, persons, acts, etc. are likely to occur, and which types of
actions and responses that are appropriate on our part. It thereby guides our
perception and decision processes, provides us with an arsenal of preconceived
solutions, and makes it generally possible for us to scan and evaluate the
avalanches of information constantly bombarding our senses, and react to it in
real time.
Sets can exist on different levels. At home, one set is activated, covering our
domestic activities. Engage in political or philosophical debate, and a more
sweeping set may be invoked, called ideology. Join in scientific research, and
you will soon discover the reigning set of that science—what Kuhn called its
paradigm (1970). And while much of our set building is original, in the sense
that we synthesize our own sets on the basis of original experiences, we also co-
opt (imitate) sets or parts of sets built by others. That is especially true for the
more abstract, symbolic sets—for instance, ideologies and scientific paradigms.
As a student, only the foolishly self-confident or the true genius will dare to
deviate from the basic set (paradigm) agreed on by the canons of the science in
question.
The Constraints of Sets

Our mental sets are powerful and indispensable. Set building is a very efficient
way of coping with reality, and we could literally not survive a day without
them. But, like all simplifications (and they are indeed simplifications of
reality), they are also constraining, because we tend not to perceive events or
objects that fall outside our sets (or, if we perceive them after all, we are
inclined to judge them irrelevant). Our thoughts and actions tend to occur inside
the set and consider it given. In real life, breaking out of the set requires
considerable energy, and will often be felt as disturbing and threatening.
Thomas Kuhn (1970) has convincingly demonstrated this effect in the realm of
science, but it is just as true in other spheres of action. Business history is full of
companies going bankrupt because reality was changing, whereas the managers’
mental sets were not, with the consequence that new, crucial developments were
overlooked. You may as well talk of business paradigms as scientific ones.
Consider for instance the example of the Swiss watchmakers: their business
paradigm was built around mechanical clockworks. They strived to become
better precision mechanics. Unconsciously, they assumed that watches were in
their essence mechanical devices. Accordingly, when a superior technology for
timekeeping came along, they did not recognize it: electronics was not their
business; it was not part of their paradigm. The customers thought otherwise,
though, and electronic watches almost wiped out the entire Swiss watch
industry. Their business was cut by two thirds in just a few years, and the
Japanese took over the hegemony.
The electronic revolution in timekeeping also tore the timekeeping function

away from watches altogether, and we got timekeeping pens, calculators, and
radios—even coffee machines! The business paradigm of the victorious
Japanese watchmakers, however, was strictly a technological one—it did not
contain the notion of a watch as a piece of apparel that happened to measure
time. It took a shell-shocked Swiss to think of that, which shows that when a
paradigm first breaks down, radical change in many directions becomes possible
(indeed, for people who feel stifled by the old paradigm, its collapse is often
experienced as a liberation). Combining the knowledge that people always like
to dress smart with the low price of electronic watches, the stricken Swiss watch
industry spawned the Swatch, and the ever-changing collection of funny,
colorful watches soon swept the world.
It is also noteworthy that IBM almost did not enter the computer business, at
least not as early as they did. According to Cuthbert C. Hurd, then a coworker
of Tom Watson, son of the legendary Watson Sr. and champion of the computer
cause within IBM, IBM’s planning department in 1950 vigorously opposed
going into computers (Augarten 1984). “Because they could not imagine classes
of problems different from those already treated by punched-card equipment,”
Hurd wrote, the planning department “told me throughout 1950 that no
computer could ever be marketed at a price of more than $1000 per month.”
At about the same time, Watson Sr. is credited for saying that “the United
States will never need more than twelve computers.” Eventually, IBM delivered
eighteen of its first computers, the 701, at a cost of $15 000 per month, and most
of them to private corporations! Even after this remarkable success, the
planning department kept repeating their “You can never sell a machine that
rents for more than $1000 per month,” now modified by the extension “except
to scientists.” They kept on resisting the construction of the 650 all the way to
its release in July 1953, when it was an instant hit at around $3500 a month.
Fifteen hundred machines were manufactured altogether before the 650 was
phased out in 1969.
You may also speak of national or societal mental sets and paradigms. A
contemporary example is the development in the newly liberated countries of
Eastern Europe, where the communist paradigm has been officially discarded.
The common mental sets created by this paradigm linger on, however, and are
still the main obstacles for turning the economy around. Typically, those
countries that fare best are those that experienced a period with a modern
capitalist economy before they were occupied, and where capitalist/liberalist
mental sets therefore still exist in the population.
As the communist sets slowly disintegrate under pressure of the new realities,
chaos is threatening, as it is extremely difficult to build new sets
shared by all in such a short time, even when there are obvious (and indeed
identified) models to be found elsewhere. Most epochal of all, the paradigm of
the Soviet Union as a strong, centralized empire broke down—not only the
paradigm of communist party leadership, but the very concept of the union.
The breakup of empires is a risky business, and both revolutions and wars as
well as other dramatic upheavals show us that we are not masters of the very
complex—unintended consequences proliferate and surprises abound. Basic
mental sets represent integrated knowledge, tested through centuries of
unforeseeable incidences. We know they work in their proper settings;
therefore, they provide us with much needed stability and predictability in
human affairs. That is their strength but also their weakness, since stabilization
also means a bias toward the status quo.
89
5 The Dawn of Organization
“My notion is, I said, that a state comes into existence because no
individual is self-sufficing; we all have many needs.”
Socrates, in Plato, The Republic, c. 380 BC
Evolving from the Primate Stage

As naked animals, then, we humans are in many ways constrained in our
organizational abilities, even if we far outperform all other animals. We cannot
process important matters in parallel, we have a limited memory, and there are
many important constraints on our capacity for problem solving. Our
communicative capabilities are restricted by narrow bandwidth and short range,
and our more basic, primate nature poses many obstacles to the rational
behavior required for large-scale organizing, especially when it takes place
outside the domain of the family or the local band.
What kind of organization did man then build in his prehistoric or “natural”
state? To answer that question, we can either look to the studies of
contemporary primitive societies, or we can consult historic evidence—or,
preferably, both. There are a couple of problems, however.
First, if one wants to study preliterate organization, one confronts the same
main problem as when studying preliterate history—there are simply no
firsthand accounts available, because all written material must by definition be
secondhand renderings of knowledge passed on from an oral tradition. What we
do have preserved are myths and legends, such as folk tales, religious myths,
and epics (e.g. the Iliad and the Odyssey). All is not lost, however. Although we
know that myths are not accounts of historical facts, and legends are notoriously
unreliable in details, both myth and legend preserve important background
information about the societies that created them (Shotwell 1961). It is highly
unlikely that myths and legends will operate with basic social and
organizational patterns that
are totally different from those of the societies that produced them, and
archeological evidence can further corroborate the evidence they contain.
The myths and legends that have made it into writing, however, are largely
the creations of the most advanced societies—those that made the transition to
literacy. They therefore probably reflect social organization at a fairly mature
stage. To gain insight into the conditions of man before what we call
civilization, to grope for the very beginnings of human society, we are, as
Wilson (1988) remarks, invariably drawn toward the simplest societies we
know—the contemporary hunter/gatherer societies of the third world. This has
its own problems, since the primitive societies that have survived to be studied
in our own time may not be representative of man’s prehistoric past—as Morris
notes (1967), the tribes that still remained at the Paleolithic level in the
twentieth century were probably more representative of the dead ends in human
cultural development than of the mainstream strains of creative, exploratory
human societies.
However, everything we know about prehistoric man suggests he was a
hunter/gatherer, and even if parts of the culture or environments of
contemporary tribes have served to hold back their development, it is highly
probable that they have enough in common with our (and their) distant
ancestors that we can learn a lot about the conditions of prehistoric man by
studying them. It is the closest we can get.
Present-Day Hunter/Gatherers
Hunter/gatherer societies—at least those that remained in the last half of the
twentieth century—are extremely simple and small scale. According to Wilson
(1988) the bands are small, consisting normally of 25 to 50 people, and they
have no permanent place of residence. Neither do they recognize exclusive
territories or formal boundaries. Although bands normally move within a
geographically restricted region, the regions overlap, without this giving rise to
territorial conflicts. There is nevertheless a definite association between the
people and the territory, but it centers around features of the landscape rather
than the stretches of land between them. Paths, tracks, water holes, and sacred
sites are the landmarks of the hunter/gatherer bands, and serve them as base
points for mapping their relative positions as they move about.
Because they are constantly on the move, they have no permanent
settlements—they erect temporary camps that may last only a few days or
weeks. Shelters normally take little more than an hour to construct, and in some
cases they even live around the fire in the open. They live in very close physical
proximity to each other, with almost no privacy as we know it. Not surprisingly,
conflict management and control are well developed in these societies—conflict
is disruptive and must be avoided. If things get
5 The Dawn of Organization 91
too tense, the exit option is always there—it is perfectly legitimate to leave the
band and join another.
This relaxed attitude to group membership seems to permeate hunter/gatherer
society. Wilson (1988) holds that modern research seems to bear out that
hunter/gatherers have more flexible and fluid relations than conventional theory
has acknowledged. Kinship ties are weak, says Wilson, even between parents
and children. Pandam children, for instance, are free to leave their parents after
the age of about six. Marital relations change, and people frequently change
dwelling for what seem to be just desires and whims. The strongest criterion
seems to be personal affection and feeling of friendliness, and kinship ties have
significance only as far as they are reinforced by affection and physical
proximity. Even for people in the same band, kin is generally not reckoned
beyond the second degree of collaterals. At the third degree, people start to
forget kinship ties. However, kinship still seems to be the most basic structuring
mechanism. It can easily be overridden by affection, but in most hunter/gatherer
societies, it always exists as an independent factor (the Pandam appear to be at
the extreme end of the kinship importance continuum).
Above all, however, the hunter/gatherers seem to value independence—it is
encouraged in children from the start. Dependence on others is looked down
upon. The ideal is that everyone should be independent and self-sufficient.
Sharing of food is nevertheless ubiquitous, especially the meat of larger
animals. The bands are extremely egalitarian, and any attempt on the hunter’s
side (after killing big game) to boast about his skill is immediately put down by
the others. Great care is also taken to avoid recognizing the lucky hunter as a
benefactor, as someone to whom the rest should owe favors.
It follows from the fluidity of group membership and personal relations as
well as the independence ideal that the hunter/gatherer bands have no real
central authority structure. There is no chief in our meaning of the word, and the
social order is upheld mainly through consensus and group pressures. People
breaking the consensus are more or less ostracized—they are in reality forced to
comply or leave the group.
The social structure is minimal and shows a definite resemblance to the
roving bands of our relatives among the primates. Indeed, some of the apes have
a clearer central authority (in the form of a dominant male) than a number of the
bands described by Wilson. There is a degree of organization—common tasks
are undertaken, food is shared—but on a very small scale, and based mainly on
direct personal relationships reinforced by affection. The size and scope of
organization is limited by the extremely low overall population density in
hunter/gatherer territories, by the small size of the bands, and by man’s intrinsic
physical limitations.
There are, however, hunter/gatherers that have developed more advanced

social structures—for instance, the aborigines of Australia. When Europeans
first encountered them, they had been living undisturbed for perhaps as long as
30 000–40 000 years and had developed a richer culture and more elaborate
organization than most of their remaining “colleagues” on other continents. But,
even if they were more advanced socially than the Pandam, Naiken, Hadza,
!Kung, and others described by Wilson, they were still distinctly “primitive”
and without any trace of sedentism.
As Lauriston Sharp (1952 1) described them, they lived a roving life in fairly
small bands. They had domesticated the dog, but no animal that served as a
source of food— foraging was their dominant activity. They did not know
metals, and even their stone tools were primitive compared to the refined flint
implements of the mature stone age cultures of ancient Europe.
Their social organization was structured along clan and kinship lines, and
closely associated with their religious concepts and their perception of the
world. Aboriginal belief divided time into two great epochs—the first a distant
and sacred past, populated with mythical ancestors, and the second a new and
more prosaic order comprising the present. The mythology held that everyone
and everything present had a corresponding archetype in the mythical epoch,
and that everything that happened today was just a reenactment of the actions
originally carried out by the mythical ancestors. A man was a member of a
particular clan because his alter ego among the ancestors was so, his name was
the same as the ancestor’s, he performed the same duties, married a certain
woman from a certain clan because his ancestor had done likewise, and so on.
These relationships even transcended the local group, because of ritual and
trade relationships between groups. In northeastern Australia, the most
important items of trade were stone axe heads, coming from quarries in the
south, and spears made from the barbed spines of stingrays, originating from the
coast-dwelling groups in the north. This string of trade relationships may have
extended up to a thousand kilometers, involving a large number of separate
communities. The trade relationships between pairs of persons from different
communities were defined within the ancestral system in kinship terms,
although no actual kinship was
1Sharp’s paper is about the Yir Yoront and their neighboring groups on the Cape York Peninsula in
Queensland, Australia, in the 1930s. It has inspired several comments about the important interplay
of technology, culture, and organization, for instance Peter S. DeLisi, “Lessons from the Steel Axe:
Culture, Technology, and Organizational Change,” Sloan Management Review, 3, 1990, pp. 83–93.
involved. Trade was carried out mainly during the great ritual celebrations in the
dry season, which often attracted hundreds of people.
The aborigines had no conception of a future different from the present—
their view of history was circular rather than linear. They believed that nothing
new ever happened, that their total universe of people, actions, and artifacts was
defined and laid down in the sacred epoch. No new actions or artifacts were in
their view possible. This meant for instance that they did not use any form of
boat or raft, even though their neighbors 70 kilometers to the North made and
used bark canoes. They knew about the canoes, the materials were readily
available, but they also knew, they explained, that their mythical ancestors
never built or used canoes, and that was the reason why they lacked it. They
assumed that the canoe was a part of the ancestral universe of their neighbors
and regarded it therefore natural for them to have it.
The resilience of this system and was so great that even the account of a
dramatic encounter with a party of cattlemen in 1864, where at least 30
aboriginals were killed, was effectively suppressed—not a trace of the event
could be found in any of the stories containing the history of the group when
anthropologists studied the group 70 years later. It was as if a collective
suppression of the fact had taken place because it did not fit in their view of the
world: None of the mythical ancestors had ever been attacked by white men and
killed in scores by firearms. It is tempting to interpret this in the light of
Morris’s comments and the concept of mental sets described in the previous
chapter: The Yir Yoront had developed a religious/cultural system and an
accompanying mental set that blunted curiosity, blocked developments in
knowledge and technology and locked them into an evolutionary dead end.
We may similarly question the extremely weak structures of the
hunter/gatherer societies described by Wilson: Maybe it is their aversion to
social control and obligations—a necessary complement to more permanent,
close cooperation—that explains why they have remained at the hunter/gatherer
stage.
Domestication
Ancient tribes similar to the Yir Yoront did not by any means exhaust the basic
human potential, however. The evolutive process continued, and man settled
down. Even if the hunter/gatherer society is mankind’s starting point, the
overwhelming majority of humans have been living in permanent settlements
for the last 10 000–15 000 years. Historically, this has been the natural way of
life for all important civilizations. When inquiring into the roots of human
organization, we must therefore include some reflections on the changes
wrought by sedentism—which we
know fairly well both from archeological evidence and the study of simple
sedentary people from the twentieth century.
When humans became sedentary, two important things happened. First, they
began to build more sturdy dwellings, and second, the concept of property was
extended. At the outset, it was not necessarily a question of individual property
or the ownership of land or animals—many of the earliest known settlements
were situated by the seashores or along lakes and rivers, and their inhabitants
probably relied on fishing. But the village itself, at least, became the property of
the community, as well as the increasing number of personal and family
belongings such as tools and household utensils. Later, when horticulture,
agriculture and domesticated animals became the economic basis for most
societies, the rights to tillable land, herds, grazing areas, and tools became not
merely important, but crucial for survival. Humans became fiercely territorial,
defending what they had, and often engaging in war to seize new land.
When rights to land became established, kinship took on an important new
dimension. The fluid arrangements of the hunter/gatherer bands were simply not
adequate anymore, since kinship regulated the access to land—collective or
individual. Land rights were inherited on the basis of kinship and village
affiliation, and the exit option was not so easily available anymore—individuals
could not simply leave their native village and expect to become a full member
of another one. Leadership became more pronounced, either in the form of
chiefs or councils.
There were (and are) many variations, however. The status of chief may be
inherited, or a chief may be chosen. Councils may consist of family heads,
elders, or combinations thereof. There is even evidence that the same societies
may oscillate between the two forms, which Leach (1970) reports as a likely
explanation for the existence of two parallel systems of authority among the
Kachins of highland Burma: Some villages were hereditary chiefdoms; some
were ruled by councils of family heads. However, the evidence suggests that
chiefs who stretched their powers too far could be deposed and supplanted by a
council, and a strong natural leader eventually emerging in a council could in
his turn succeed in establishing a new hereditary chiefdom.
This new importance of kinship gave sedentary communities a much more
permanent and substantial structure than hunter/gatherer society, a structure that
was further elaborated and strengthened in societies that were systematically
able to produce a food surplus. Surplus production of food made room for
craftsmen, merchants, religious specialists, and ruling classes, and made large
construction projects possible—some of which (like the extensive irrigation
projects of the ancient civilizations in the Middle East) increased the fertility of
the land further and thus contributed to the development of even larger and
more complex societies.
Surpluses invariably lead to social stratification, an uneven distribution of

property, and a more stable social structure. Again, we can turn to Leach (1970)
for evidence: the Kachins, with their structural instability, lived in the steep hills
in the highlands, more or less at the subsistence level. The villages could
produce little or no surplus; there was simply no economic room for a class of
landowners.
In the valleys, on the other hand, the conditions for agriculture were much
better. The people living there, the Shan, had a more sophisticated culture and a
stable social structure, forming feudal 2 states based on hereditary positions tied
to the ownership of land.
This new, structured society imposed a much wider set of rights and
obligations on people and developed a rich set of rituals and routines to enforce
them. Unlike in hunter/gatherer society, routine is a hallmark of the sedentary
community. Most of the day is spent doing programmed tasks that are necessary
to fulfill one’s obligations toward others or tending the land and animals that are
the basis for one’s subsistence. The social and political structure is thus cast in
a stable pattern of actions that is constantly enacted and becomes thoroughly
ingrained in people’s minds. Periodical religious or other feasts and rituals
contribute to this and give the status quo a more solemn blessing. Often, a
period of religious training followed by initiation rituals becomes a part of the
upbringing for all children.
When the development of human society reached this stage, technology and
techniques had already started to make a difference. Domestication of the horse
in many societies and the emergence of ships improved communication, and
buildings were used not only for shelter, but also to encode information—
especially information of a ritual character and with a bearing on the social
structure. Mankind was approaching its first real technological revolution, the
invention of writing—to which we shall return in Chapter 6. The rest is, literally
speaking, history.
Theory for Simple Organization

Most organization theorists are not very interested in simple organizations—in
exploring the primal space of organizations. That is perhaps not so surprising,
since organization theory as a discipline sprang from the problems experienced
in building and running the complex organizations
2There are some nuances in how different people use the term “feudal.” The most restrictive reserve
the term for the political system of the kingdoms of medieval Europe, others think (as I do) that it is
meaningful to extend the term to cover all hierarchical political systems where lineage and control
of land are the main structuring elements.
that arose in the late nineteenth century and grew to prominence in the
twentieth. Small-scale societies and simple organization have by and large been
left to anthropology.
Another reason is probably the distinction between formal and social
organization set up by the classical theorists. Organizational theory was
restricted to the former—the latter, especially the family, was seen as something
quite different. To me, this distinction is artificial. Organization lies at the
bottom of human existence, and the repertory of behaviors that formed our
societies from the earliest times are the same that underlies the more advanced
formal organizations of the modern era—even if they have evolved
considerably, and have come to depend in large part on tools not available to
humans in the “natural” state. Even today, when formal organizations, voluntary
organizations, and family life all have different “frames” (Goffman 1974), we
are not able to separate them fully. Experiences and prescriptions from one
frame tend to spill over into the others, and our situation in one of these
domains always interacts with our situation in the others.
One of the few theorists who does discuss simple organizations is Henry
Mintzberg (1979). His classification of organizations contains two forms that
encompass small, simple organizations: The Simple Structure and the
Adhocracy.
Of these, the Simple Structure is the intuitive small-scale organization with a
strong leader, often charismatic and entrepreneurial, leading the organization
through direct supervision effected through informal contact with its members
(Mintzberg 1979, p. 306, bold type from original):
The Simple Structure is characterized, above all, by what it is not—

elaborated. Typically, it has little or no technostructure, 3 few support staffers,
a loose division of labor, minimal differentiation among its units, and a small
managerial hierarchy. Little of its behavior is formalized, and it makes
minimal use of planning, training, and the liaison devices. It is, above all,
organic. In a sense, Simple Structure is nonstructure: it avoids using all the formal
devices of structure, and it minimizes its dependence on staff specialists. The latter
are typically hired on contract when needed, rather than encompassed permanently
within the organization.
Mintzberg gives the Simple Structure a much wider span, however—it ranges
from the small entrepreneurial start-up, where everyone works in the same little
room (which he calls the simplest structure), to the large, autocratic
organization run by the iron-willed founder/owner. It also includes Thompson’s
(1967) synthetic organizations, ad hoc organizations
3The professional part of the staff, where you find the analysts who monitor and analyze the
environment and plan and standardize the work that the others are doing.
set up to handle unexpected crises, such as natural disasters, and headed by

strong leaders with comprehensive authority.
It is thus clear that many of the organizations falling into the Simple
Structure category (the large ones) are anything but simple in administrative
terms, and that they require sophisticated, technology-based infrastructures
(such as paper-based archives and telegraph lines) to function. Prime examples
are the large American trusts of the late nineteenth century, which Mintzberg
classifies as Simple Structures because of their total dominance by single
owners/entrepreneurs.
The Simple Structure, on a sufficiently small scale, has probably always been
an extremely common organizational structure in human societies, as indicated
by the foregoing discussion. This conclusion is also supported by recorded
myths and legends. From the matriarchal queens of Neolithic Europe to the
nineteenth century chief and his tribe, to the master craftsman with his
apprentices, and the patriarch with his family, the Simple Structure abounds.
But, as indicated earlier, other structures have also existed from time
immemorial—from groups of cooperating hunters to more or less democratic
villages and tribes with a council of elders or family heads as the supreme
authority. The extremely simple bands of the Pandam, for instance, do not have
sufficient leadership to qualify as Simple Structures. They can only be described
as very loose Adhocracies. The more democratic variety of Kachin villages are
a sort of a mixture, with family heads (Simple Structure) forming a governing
council (Adhocracy).
With the term Adhocracy, Mintzberg mainly denotes innovative
organizations with a high content of professionals and experts (Mintzberg 1979,
pp. 432–33, bold type from original):
In Adhocracy, we have a fifth distinct structural configuration: highly organic

structure, with little formalization of behavior; high horizontal job
specialization based on formal training; a tendency to group the specialists in
functional units for housekeeping purposes but to deploy them in small
market-based project teams to do their work; a reliance on the liaison devices
to encourage mutual adjustment—the key coordinating mechanism—within
and between these teams; and selective decentralization to and within these
teams, which are located at various places in the organization and involve
various mixtures of line managers and staff and operating experts.
To innovate means to break away from established patterns. So the innovative
organization cannot rely on any form of standardization for coordination. In
other words, it must avoid all the trappings of bureaucratic structure, notably sharp
divisions of labor, extensive unit differentiation, highly formalized behaviors, and
an emphasis on planning and control systems.
This connotation is natural in the modern world of formal organizations. But the
basic structural properties of the Adhocracy are found in the
small, egalitarian human group, where problems are solved as they crop up,
decisions are made by consensus, and coordination is taken care of by mutual
adjustment. When society develops and grows beyond the limits of the mutually
coordinating band, especially when it becomes sedentary, the Adhocracy
evolves into the form seen in the Kachin villages, where the basic structure is
the family, but where authority on the societal level is still created by mutual
adjustment—now through the institution of the council. Following Mintzberg’s
terminology, we might well call this second-order form of the Adhocracy a
Councilcracy.
If a strong natural leader emerges in a Councilcracy, it may change to a
Simple Structure, but there may also be strong norms that inhibit such
transformations (as in our own societies) or effect a return to Councilcracy
when the leader dies, is deposed, or otherwise discredited.
Looking at the anthropological and historical evidence there is much to
suggest that those two configurations—the Simple Structure and the
Adhocracy—are the two basic organizational configurations of the human
race.
In their simplest forms, they are also clean representations of two of the
fundamental solutions to the problem of coordination (see Figure 3-1 on page
51): the Simple Structure achieves coordination by empowering one person to
direct the others, and the Adhocracy by letting all the group members know
what the others are doing at all times, thereby allowing each one to continually
adjust his or her behavior accordingly. Coordination in the Simple Structure is
thus focused on directing work and does not require that people are equally
competent or informed (indeed, most leaders in such organizations will prefer
that they are not). The Adhocracy, on the other hand, requires not only that
everyone in the group knows the common goal, but also that they agree on it,
have a common understanding of it, and are in reasonable agreement on the
means. If not, their self-administered actions will simply not fit together. We
may therefore say that coordination in the Adhocracy is focused on sharing
information, and it requires that the participants are on the whole equally
competent to act on that information.
The two other basic coordinating mechanisms delineated in Chapter 3,
standardization of work and standardization of skills, did not give rise to
separate organizational forms in preliterate societies, but they were obviously
operative in rudimentary forms. As long as there have been humans, there must
have been tacit knowledge and routinized ways of executing recurring tasks—
indeed, as noted in Chapter 4, this is perhaps our main trick for surviving with
limited cognitive capacities in an information-rich world.
Enlarging a Simple Structure is fairly straightforward, in the sense that the

means are a part of an age-old human repertoire of social roles: if the
organization becomes too large for the leader to oversee, he or she can delegate
authority to trusted persons who are inferior in status and have clear loyalties.
Insofar as such persons can be seen as direct extensions of the leader’s own
person and authority, the capacity for direction and control can be substantially
increased, whereas the line of command is kept unequivocally intact. We shall
soon see how that was done in preliterate societies.
The Adhocracy, on the other hand, is much more difficult to extend. In an
organization where everyone communicates directly with everyone else, the
members’ communication capacity rapidly becomes saturated as the group
grows, and all answers to this problem must compromise on the basic form to a
much larger degree than for the Simple Structure. Because the only possible
answer is to divide the growing group into subgroups, the all-to-all
communication and direct, mutual adjustment is irretrievably lost. Even if
direct mutual adjustment is preserved within the groups, coordination between
them can be achieved only through group representatives. The method of
representation then becomes an important parameter. The basic Councilcracy
solved this by combining the Simple Structure of the patriarchal family with the
Adhocracy of the council. The Greek city states evolved it into a combination of
the town meeting and representative democracy (however, only proper
citizens—free men who owned land—could vote), a tradition further elaborated
to create our own modern representative democracies.
Adhocracy has not been a favored form of organization in more elaborate
societies up through the ages. The reasons are probably mixed: in addition to the
scaling problem, which is serious enough, the Adhocracy’s democratic form
could not survive in the authoritarian cultures that have dominated every large
and sophisticated society until quite recently.
The Problems of Organization Building in Preliterate Society

There are three questions that seem especially interesting regarding organization
in illiterate or semiliterate societies (societies where the art of writing is known,
but where so few are skilled in it that most activities and organization building
have to be conducted without its help): what are the domains of the
organizations we find, how are they structured, and what happens when they hit
the upper limits of human memory, communication, and information
processing—when they become too
complex to handle by mutual adjustment or by direct supervision (the

commands of a single person)? How does a kingdom grow larger than the
number of persons the king and his immediate helpers can oversee?
The Organization Domains and Their Structuring

The domains of organization in illiterate societies are few. The first and most
basic domain is the family, then we have the society’s authority structure—
whether the boundary of the society is the village, the tribe or the state. In more
advanced societies, there may also be specific religious domains with separate
organizing, and even small organizational domains of a craft or commercial
nature.
Some readers may balk at the idea of calling a family an organization. We are
used to reserving that concept for formal organizations (see, for instance, the
discussion in Silverman 1970). As I just said, I think this distinction is artificial.
The formal organization, be it business or governmental, is a fairly new
phenomenon. In preliterate society, the family was no doubt the main
structuring element of society—indeed, even in Western societies, it kept this
position until quite recently. In Taiwan, which has built a modern economy
while retaining traditional Chinese values, the family can still be regarded as the
main structuring element of ownership in business (Hamilton and Biggart
1988).
There are several reasons for this. First, the family was not only an informal
group bound together by affection; it was a structure with very formal and
material purposes: to uphold rights to land (or condemn to serfdom), and to
channel political power. Rights to land—allocated through the operation of
kinship, inheritance, and marriage, not through conveyance, contract, sale—
were almost the sole source of economic and political power until trade became
so abundant that a rich merchant class emerged (Nash 1966). Often, however,
citizenship continued to be tied to the rights to land, as it did in the Greek city-
states.
Second, the family structure was already there—an important feature in
illiterate society, where the burden of retaining administrative information was
formidable: everything had to be remembered by someone, and, preferably, by a
number of people, should claims be contested. Family relationships were widely
known and easily remembered and were therefore a convenient infrastructure
for other purposes. Third, loyalty could best be counted upon from members of
one’s family, where both affective and economic ties were present. And, of
course, let us not forget that ancient humans in all probability had the same kind
of affectionate feelings for their family, especially their offspring, that we have.
In a society without the social and judicial safety nets of modern industrialized
nations, it is quite natural for people to protect and support their nearest and
dearest first.
Circumventing the Barrier of Cognitive Capacity

The rule by family head, chief, or council is adequate as long as the family,
tribe, or city is below a certain critical size. This critical size is not possible to
determine in any exact way, because it varies according to contextual
circumstances (geographical extent of domain, fertility of land, hostility of
neighbors, etc.). However, as our assessment of man’s cognitive capacities
indicates, the critical size cannot be very large. The barrier must have proved a
formidable one, and most prehistoric societies were probably small-scale
structures, just like a large number of the primitive societies studied by
anthropologists in our own time. The old Norse religious myths, for instance,
tell us that all the gods, æsene (the Aesir), lived together in Åsgard, with Odin
as their chief, and even if they all had their separate duties, there is no mention
of a larger hierarchy based on territory.
But, somehow and sometime, such hierarchies emerged, and we can find
evidence of this also in the myths, as for instance the Greek creation myths,
where a Titan and a Titaness are placed in charge of each of the planets (Graves
1960). In the very old Mesopotamian epic of Gilgamesh, believed to be written
down 1500 years earlier than the Iliad and the Odyssey (Sandars 1964), we can
also find passages that indicate arrangements of a feudal character, as when
Ishtar, Queen of Heaven, the goddess of love, fertility, and war, tries to
persuade Gilgamesh to become her husband (p. 83):
When you enter our house in the fragrance of cedar-wood, threshold and throne
will kiss your feet. Kings, rulers, and princes will bow down before you; they shall
bring you tribute from the mountains and the plain.
Even the myths of the Aztecs tell about how the supreme god, Omeyocán,
divided the heavens into different regions and created a god to head each of
them (Beals 1970). Further evidence of the general nature of this arrangement
can be found in anthropological studies. If we return for a moment to the
Kachins of Burma, a village was often part of a group of villages, and the local
chiefs were subordinate to one supreme chief. The valley Shans, on their part,
had a social structure that was even more distinctly feudal, where the local
nobles were subordinate to higher lords and finally to the king himself.
In Africa, such structures were quite common in the kingdoms that existed
when Europeans colonized the continent (Lloyd 1965). There were variations in
the mechanisms by which the King appointed his
subordinate chiefs, and the ruling class might be closed or open for upward
migration from commoners, but the main principle was a division of
responsibility by means of geography. Historic variations on the feudal structure
have been the common political frameworks throughout the world, and these
were probably developed to their most stringent and elaborate form in the
kingdoms of medieval Europe.
The beauty of such a system—from a cognitive point of view—lies in the
extreme economy it offers with respect to information processing, memorizing,
and communication. Based on land rights and family lineage, it contains its own
structuring information; information that is constantly enacted in everyday life
and thereby reinforced in everyone’s memory. By delegating to his vassals total
authority over their fiefdoms and the people who live there, the king effectively
encapsulates the information required to run the fiefdoms and shields himself
from it. He now only has to worry about the loyalty of his vassals (often no
small worry, however), the size of their tributes and their contributions to his
army. The number of people with whom he must deal is drastically reduced.
The king normally does not interfere in the way his vassals run their affairs, as
long as they are sufficiently competent, loyal to him, and not so cruel or
unreasonable to the people as to inspire massive uprisings.
It is thus a system that can exist without the help of writing, and, from an
information processing point of view, there is no theoretical limit to the size of
states built on these principles. However, we know of few large nonliterate
states, and only one really large one: the Inca state overthrown by Pizarro in
1533. After extensive conquests in the two centuries preceding the Spanish
invasion, it covered what is now Peru, as well as parts of Bolivia and Chile, an
area of approximately one million square kilometers—about twice the size of
France, or nearly one third the size of India. The total population of this empire
was about 4 million.
The Incas’ state organization was of a feudal type, and their expansion came
almost exclusively in the coastal areas and in the hills where sedentary farming
dependent on irrigation had already produced a society with feudal organization
resembling their own. There they could rule through the established structure,
through the old leaders, if these leaders agreed to submit to Inca rule (which
they often did). The Inca king (or the Inca, as he was called) supplied the
vanquished lord with a new first wife from his own lineage and in return
accepted one of the new vassal’s daughters into his harem. In such a way, the
local lord was tied into the dynastic kinship system (Murra 1986). Note this
prime importance of lineage as a structuring element: When such links were
absent, they had to be at least formally established to foster loyalty and
reestablish the customary congruence between lineage and political power.
Where the social organization stopped at the village level, and the villages
themselves moved about, as in the semi-sedentary tribes adjoining their empire,
the Incas had not managed to conquer and establish permanent control—the
information economy of the feudal structure was not available and the task
became too complex.
However, even if the Incas were not literate, they had a system for numeric
records, called the khipu, to help them. It consisted of a bundle of knotted
threads, where threads and knots of different colors had different values. It was
primarily used for taxation, but as the empire expanded and the king’s needs for
soldiers, food and other goods and services came up against the limits of feudal
organization, the khipu’s decimal organization started to be used also for
organization purposes. Murra (1986) notes that, in the last years before the
Spanish conquest, the Incas seemed to try out new local subdivisions based on
numbers rather than on lineage and ethnic origin.
If there were other preliterate states of the same magnitude, knowledge of
them has not survived to be written down. Judging from the empires we do
know, we can only conclude that very large states seem easier to build and
control when the rulers have access to a literate class of administrators—a
conclusion that seems perfectly plausible. Maybe the Incas represent the
extreme accomplishment in organization building for an oral culture. Judging
from the great importance that the khipu had for their administration, we may
even view them as a borderline case—not truly oral anymore, but not quite
literate either. If left alone for another century, they might well have developed
their own script as an answer to their pressing administrative challenges. After
all, the Mayas and the Aztecs already had—in 50 AD and 1400 AD, respectively
(Ong 1982).
The Feudal Type Organization

The feudal system builds upon the rule of one, on the concepts of the family
head and the chief. Because there is a hierarchy in the family (father-son,
mother-daughter), this concept lends itself readily to build a hierarchical social
structure as well. More democratic systems had no such blueprint to follow,
and, anyway, it goes against the grain of democratic assemblies to delegate
power upward. It is thus no accident that societies with a more democratic type
of government have been less prone to build empires than autocratic states.
When a Greek city-state founded a colony, for instance, it was at no point
supposed that the new city should obey the authorities of the mother city (Kitto
1951). It was certainly considered distasteful to enter into conflict with the
mother city, and her citizens usually enjoyed certain privileges when visiting,
but the new city was by all parties considered politically independent from day
one. Another indication of this can be found in the reluctance of the citizens of
most EU-countries to move toward further political integration in the European
Union.
From this perspective, it is also noteworthy that when countries in Europe
managed to sustain colonial empires even when they themselves became
democracies, they did so by basing their rule on ideas of racist and cultural
supremacy. As soon as those concepts crumbled through moral debate and
reflection in the ruling countries, the empires also disintegrated. The last
example is the recent collapse of the Russian empire. Built by the feudal and
imperialist czars, it was upheld and even extended by the Communist Party,
which (though democratic on paper) ruled on the basis of their own supremacist
ideology: the theory that only the party cadres were politically conscious and
could understand the true interests of the people. By demolishing that ideology,
the Soviet leadership destroyed the ideological basis for the union, and the
republics and regions were bound to claim sovereignty.
How does the hierarchy of the feudal-type state fit into modern organization
theory? It has certainly evolved from the Simple Structure, but it no longer
belongs to that category. On the surface, it may seem, it is just a case of the
universal hierarchical form, found today in military organizations and in the
large bureaucracies of government and private business. However, there are in
fact significant differences. Modern bureaucracies rely on advanced
administrative technology, primarily the art of writing and its associated tools.
They are specialist oriented, with each level and department having definite
responsibilities that fit together in a complex task structure. Work flows trough
it in an orderly fashion, according to centrally administered plans and rules for
execution. There are large flows of information both vertically and horizontally.
The original feudal hierarchies did not have the instrument of a written
language at their disposal. They were not at all specialist oriented; on the
contrary, the parts of a feudal state (on the same level) are by definition similar.
Communication, both vertically and horizontally, was kept to a minimum. The
whole point of the feudal structure was precisely to simplify complexity as
much as possible, so that the unaided human mind could handle it—the idea
was to abolish, as far as possible, the need for coordination in the first place.
If we can compare it to a modern form of organization, it must be what
Mintzberg (1979) terms the Divisionalized Form. A division, in the original
sense of the word, is a largely self-contained and self-sufficient part of the
organization, with a broad objective (usually to serve a particular market) and a
minimum interface with corporate management—ideally limited to passing
general goals, budgets, and profit goals downward, and status information
(mostly in the form of financial reports) upward. “In
general,” says Mintzberg (1979, p. 381), “the headquarters allows the divisions
close to full autonomy to make their own decisions, and then monitors the
results of these decisions.” Not infrequently, the divisions are organized as
separate legal entities (corporations).
The key elements, from our point of view, are the separate objectives of the
divisions, the duplication of operating functions across them, their resultant
quasiautonomy and elimination of the need for interdivisional coordination.
Real coordination is not necessary, as evidenced by the use of standardization
of output, which, as we concluded in Chapter 3, is not a proper coordinating
mechanism, but only a tool for controlling the level of performance.
The goal of this method of organizing is obviously to limit the need for
information flows across division boundaries and thus obtain a wide span of
control for top management. Viewed in this perspective, the Divisionalized
Form is no more than the modern commercial version of a primordial Feudal
Form of organization, which developed out of the need to build large
organizational structures with minimal requirements for memorizing,
information processing, and communication. In the Feudal Form, the subunits
were fully independent from each other, answering only to the higher level, they
had their own administrative apparatus, and their objectives, although usually
similar, were separate in the sense that they did not depend on each other or
require any coordination with other units on the same level. Finally, control was
maintained through the use of established standards of output (with the size of
the tribute or taxes and the number of men for the army as the most important)
supplemented by the occasional royal command.
Except for the disagreement of the status of standardization of output as a
coordinating mechanism, this is in no way in conflict with Mintzberg’s
definition. On the contrary, he emphasizes that divisionalization does not mean
(1979, p. 381) “. . . a complete structure from the strategic apex to the operating
core, but rather a structure superimposed on others. That is, each division has its
own structure.” He goes on to say that the Divisionalized Form tends to draw its
divisions toward the Machine Bureaucracy configuration, but that it is not a
necessary condition, since the focus of the Divisionalized Form configuration is
on the structural relationships between the headquarters and the divisions.
Military Organization
Military organization is an interesting chapter in itself, and one that has
undergone radical changes that are not (even today) fully acknowledged in the
code of command.
At first, the military organization was only a mirror image of the feudal state,
with a substitution of group (unit) for territory. There was little differentiation
among the soldiers, and they were commanded by their feudal lords. The
hierarchy was thus structured after the civil society, and, as the civil structure
was fashioned to minimize information processing and communication, it also
served the needs of war well: during battle communication is extremely
difficult, and the need for it must be reduced as much as possible.
Even the feudal army was thus similar to the Divisionalized Form, with one
important exemption: The King now needed to coordinate the actions of his
subordinate lords, and he needed to do so in real time as far as circumstances
would allow. In this respect, we may say that the feudal nation in war reverts
partly to the Simple Structure by strengthening central coordination as much as
possible. In preliterate society, however, it is not feasible to revert fully to the
Simple Structure with a large army—the administrative load on the central
command would then totally overwhelm its capacities. The modern, specialist,
bureaucratic armed force with its abundance of communications equipment and
its great capacity for detailed planning is therefore closer to Mintzberg’s
definition of the Simple Structure than the army of Genghis Khan was.
The small differentiation between soldiers that persisted up through history,
even into our own century, also meant that an officer could successfully
command almost any military unit, and it was from such a structure that the
military code of command grew, whereby any officer had authority over all
personnel below him in rank, regardless of unit. It was probably Frederick the
Great of Prussia who really established this principle, through the standardized,
elaborate training he mandated for all his soldiers.
Today, this tradition is rapidly becoming meaningless. The specialization in a
modern armed force is just as extensive as in a modern corporation, and the
notion of universal authority in war today makes no more sense than it would in
industry—such as authorizing a vice president of finance in a manufacturing
company to issue direct orders to a foreman on the shop floor about the way he
should run his robotized paint line. In theory, the unity in military command is
still in force, but, in practice, the sensible officer will always yield to the
specialist knowledge of a subordinate.
The Basic Principles of Preliterate Organization

We have now explored what was defined as the primal space of organizations in
Chapter 2. Before the art of writing permanently changed the premises for
human reflection and human society, central planning and
minute, standardized directives were not available as organizational tools.

Preliterate societies had to use solutions with much greater information
economy, structures that could rely on human memory alone, and that required
an absolute minimum of communication between the levels in the hierarchy.
The basic forms relied either on the rule of one or on a form of rule by
consensus or council. For larger structures, where one ruler or one council could
not manage the complexity, the iron constraints on human memory,
communication, and information-processing capabilities forced a reliance on
two principles: the delegation of authority and the encapsulation of information.
109
6 The Power of Technology
“It must be confessed that the inventors of the mechanical arts

have been much more useful to men than the inventors of
syllogisms.”
Voltaire, “Philosophy,” in Philosophical Dictionary, 1764
The Nature of Tools

In many ways, this book represents an inquiry into the development and use of
tools. Before continuing, I would therefore like to reflect for a moment upon the
nature of tools—the implements we have invented to enhance our powers.
A tool is most often conceived of as something extraneous to humans—
indeed, new and revolutionary tools are sometimes even seen as unnatural, alien
and threatening. When talking about the natural state of humans, most people
even today seem to envisage a primordial hunter/gatherer society, where man
lives in peace with himself and nature—something like Rousseau’s “noble
savage.” But even if hunter/gatherer society may well represent humanity in its
primordial, natural state, such societies ceased to be typical representatives of
our species thousands of years ago—and those that survived may represent no
more than the longest surviving dead ends of cultural evolution.
As Morris (1967) argues, judged from an evolutionary point of view, the
most successful and powerful human societies today are the modern
industrialized societies of Europe, Asia and North America. In the course of
their development, they have physically overrun and displaced other societies
they encountered, like the Indians of North and South America and the
aborigines of Australia. When they have refrained from doing the same
everywhere it is mainly because of conflicts between themselves, their need to
achieve viable balances of power and (not least) their own political and moral
philosophical development. Today, most other societies do their best to become
more like them. The culture of the
Western industrialized countries is therefore overwhelmingly successful in

terms of diffusion and adoption.
It is also evident that cultural evolution has long since replaced biological
evolution as the driving force in human development and competition—it
probably happened when sedentism started some 20 000 years ago.
Interestingly, the ecologists Robert Boyd and Peter J. Richardson (1985) argue
that cultural evolution has many traits in common with biological evolution and
follows many of the same laws—a fact that should perhaps not surprise us too
much, given the basically systemic character of both spheres. Berger and
Luckmann (1967) also argue along these lines.
Nevertheless, there seems to be a peculiar reluctance to acknowledge tools as
something intrinsically human—especially advanced tools. But tools are not
delivered by fate and do not spring spontaneously from lumps of raw material.
Rather, they are conceived, designed, and crafted by humans; they are socially
and culturally dependent constructions—material expressions of culture and
knowledge (Bijker and Law 1992). Tools remain an intrinsic and natural part of
human development, they are true expressions of the human mind.
If we accept that the physical and biological properties of the modern and the
prehistoric human are not very different, there must be solely cultural reasons
for the differences between hunter/gatherers and the modern industrialized
societies—including the immense differences in size and scope of their
organization. The basic preconditions must thus have been considerably
augmented by knowledge, techniques and technology in the course of the last
millennia, since we humans have totally transformed ourselves from intelligent
apes, outwardly not that much different from their primate cousins, to beings
with a knowledge and power that set us utterly apart from all other animals. In a
mere blink of an eye on the evolutionary timescale, we have progressed from a
position as unobtrusive members of the fauna, predators among many others
and definitely no serious threat for any other species, to a position as the total
masters of the earth and all large animals.
The developments of knowledge and methods on one side, and of tools on the
other, are of course not separate processes. On the contrary, they are intimately
linked—even more so than we normally recognize. Our intellect is not
something that is suspended in a pristine, spiritual capsule; it has been and still
is developed only in our ongoing interaction with nature and other humans.
The basic nature of this iterative, recursive process is aptly illustrated by
Rachlin (1989, pp. 248–49) when he describes the opening sequence of a film
about Picasso. Picasso is filmed painting, but the camera is set up to take single
pictures about every other minute, compressing about a week’s
6 The Power of Technology 111
work into a few minutes. The picture starts with a few bold lines, but what
follows is a constant metamorphosis, as Picasso tries one approach after the
other: “A fish becomes a chicken. A bird becomes a woman. He keeps on
working.” After the week has passed, Picasso finally says, “Now I know what I
was trying to do,” and starts again—almost from scratch. Clegg, discussing the
technical aspects of factory production, is on to the same phenomenon (1990,
pp. 186–87):
Technique is not simply a commodity to be bought, but a vital aspect of

organization. This is clear in the sense that applied technique includes the human
organization or system that sets equipment to work. Equally importantly the
concept includes the physical integration of a new piece of equipment into a
production process and its subsequent refinement and modification at the hands of
the technically skilled workforce. Many manufacturers have come to grief on the
belief that technical solutions can be bought pre-packaged. This is to ignore,
precisely, that in operation these are always socio-technical solutions. What is at
issue is precisely the “cultural” context in which these solutions have to work.
Studies have shown that equipment users rather than makers develop major process
innovations (thus stealing a march on their competitors) and that small,
imperceptible “everyday rationalizations” account for the lion’s share of
productivity gains in an ongoing manufacturing business.
This is the basic human approach to discovery and innovation—to act, to try
out, and then gradually modify until it is “right.” In fact, this process is so basic
that it is even reflected in the physical development of the brain: Experiments
have shown that animals growing up in a complex and changing environment
attain larger brains with more interconnections than similar animals growing up
in extremely simple and stable environments (Hofer 1981). And it is important
to note that this difference occurs only when the animals are allowed to engage
in sensorimotor interaction with the environment, that is, merely living in a cage
in a complex environment does not foster brain development; the animal must
be able to move freely around, interacting directly with it, physically
experiencing the changes and the complexity.
Our tools thus constitute our minds’ projection into the physical world; they
allow our minds to engage the complex world on a much grander scale and in a
much more sophisticated way than our unaided hands and feet. The feedback
our minds receive is correspondingly advanced, and it drives our thinking and
investigations toward ever higher levels. The mind creates the tool; the tool
allows us to do new things and shows us new constraints, thus posing new
challenges and riddles to solve—both technological, scientific, and moral. This
represents the essence of human progress—the concept of iteration, the endless
number of small steps, some erratic, some successful, but always spawning new
insights, new
ideas, new experiments. The history of human development is therefore also the
history of tools, and probably much more so than historians generally recognize.
The pace of this development is not even. Some inventions are more dramatic
than others, carry within them the potential for far-reaching changes and release
spurts of rapid development. Other phases are characterized by the slow trickle
of improved details. Astronomy, for instance, was clearly stagnating when
Galileo set the stage for new growth by creating the first astronomical telescope
in 1609. The revolutionary initial discoveries led to the construction of larger
instruments to gather more light and resolve even finer detail. Development of
telescopes was then gradually dampened by the law of diminishing returns, until
we suddenly were able to lift them out in space, above the diffusion of the
athmosphere, and start a totally new and exiting trail of innovations.
Pasteur, on his part, could not have developed his theories about bacteria
without the microscope. On the basis of its new importance, the microscope was
then subsequently developed to the limits of the resolving power of visible light.
Further developments led to breakthroughs like the electron microscope and
later to the scanning tunneling microscope, which can resolve—and
manipulate—individual atoms. Nuclear physics on its part rests heavily on the
construction of particle accelerators, the king of which at the moment (and
probably for many years to come) is the Large Electron-Positron Collider of the
CERN laboratories near Geneva. Arguably the largest tool in the world, it is
located in a ring tunnel with a diameter of 27 kilometers and a cross section of 4
meters, running from 50 to 150 meters below ground.
In practical engineering, the limitations of Thomas Newcomen’s steam
engine became the starting point for James Watt’s improvements, subsequently
leading to Richard Trevithick’s high-pressure machine, the first really modern
steam engine. (Watt, by the way, believed high-pressure engines would be too
dangerous, and held back development until his patents expired in 1800.) In our
own time, Shackleton’s, Bardeen’s and Brattain’s crude transistor turned out to
be the necessary stepping stone for the integrated circuits of modern
microelectronics.
In business, we have seen the mutually dependent evolvement of modern
communications and the modern industrial and retailing organizations from
about 1850 (as described by Chandler, Beniger and others). And, to slip in a
Norwegian example, one man’s quest for synthetic saltpeter and another man’s
failure in building an electrical cannon led to the invention of an electric arc
furnace for the manufacture of nitrogen oxide (the key to salpeter synthesis),
spurring a large-scale development of hydroelectric power, which in turn
created the basis for a vigorous metals industry in Norway.
Indeed, even the development of moral philosophy is more closely related to

our tools and material advances than generally acknowledged. Our ideas about
war, for instance, have changed considerably the last 100 years, as we have
developed modern weapons of mass destruction and experienced their effects.
Before World War I, war, even between major nations, was much more readily
accepted as a legitimate extension of politics. Even more obvious is the recent
debate over the ethical implications of changing the genetic composition of
bacteria, plants, animals, or even human beings. Before we actually had the
tools and knowledge necessary to do it, it was not on the agenda at all—
philosophers of previous centuries were not even aware of the possibility. The
case is the same with global ecology—before human activities actually started
to interfere seriously with nature, the question was not even raised.
Always, however, it seems to be easier and quicker to exploit innovations
with a predominantly material and concrete purpose (like the steam engine) than
innovations of a partly or predominantly intellectual nature—like writing. That
should not surprise us, since manipulation of the physical world has been our
main preoccupation and means of living for almost our whole existence as a
species. All that we can see and touch is so much easier to comprehend, and
dramatic developments may take place in decades. For matters of the mind, for
the refinement of concepts, methods and social practices, we may need
centuries.
Like writing, the computer is another invention that is partly material and
partly intellectual. For the material part, the development races ahead at
enormous speed, with hardware performance doubling every year or two—but
when we strive to make good use of it, especially in the complex world of
organizations, we obviously progress at a much slower pace.
Thus, our tools are our destiny, for better or worse. To paraphrase William
Ralph Inge 1, man may never succeed in becoming lord of himself—there will
always come a new tool, a new capability, a new insight that will tax our ethics
and invite questionable actions. And we can never tackle the problems before
we experience them; to think we need to be confronted with the physical world.
Perhaps this is why so much of the discourse at our universities borders on the
sterile, and why so many exiting things happen at their fringes—where
intellectual curiosity and analytic power is more directly exposed to practical
problems.
1“Forbetter or worse, man is the tool-using animal, and as such he has become the lord of creation.
When he is lord also of himself, he will deserve his self-chosen title of homo sapiens.” William
Ralph Inge, “The Dilemma of Civilization,” in Outspoken Essays: Second Series (1922).
The Breakthroughs
The development of our tools as well as our knowledge and conception of work
has gone through many stages, and some carry significantly more momentum
than others. There may be several ways of classifying them and, conceivably,
some disagreement as to their relative importance. For our purposes three
breakthroughs stand out:
1. The invention of writing

2. The Industrial Revolution, with its sweeping developments:
- abundant energy and mechanized production
- new means of communication, from railroads to telephony
- mass literacy, cheap printing, and the knowledge explosion
3. The invention of the digital electronic computer
In this chapter, we shall look at the first two; the third must wait until Chapter 8.
The Externalization of Memory

The Struggle to Remember
Cicero, in his De Oratore, tells a story about the famous poet and orator
Simonides, who was having a meal with a number of friends (Eriksen 1987).
The building they were in suddenly collapsed, and the diners were crushed to
death under the tumbling stones. Just before the roof caved in, however,
Simonides was rescued and brought outside by the gods Castor and Pollux. The
only survivor, he was able to tell the rescuers who had been present, because he
recalled exactly where they had been seated at the table. He then realized that
much could be remembered if one located pictures at places in memory.
That was purportedly the wellspring of the classical mnemonic techniques so
widely used from antiquity to the Renaissance, which involves building a
permanent mental picture of a house with a number of rooms in a fixed order.
For each occasion, the rooms can then be “filled” with objects or events that
serves to recall a part of the speech. Advanced orators managed to build and
maintain “houses” with a very large number of rooms, and was able to
remember even very long speeches almost word by word.
However, mnemonics hardly originated with Simonides, who lived (from 556
to 468 BC) in a period when the Greeks had already started down the road to
literacy, after the full Greek alphabet was developed from the Phoenician
sometime around 700 BC (Havelock 1986). All
nonliteral societies above the basic subsistence level of the simplest

hunter/gatherers must devote considerable time, effort, and ingenuity to keep
the vital communal memory from fading, and would have had techniques for
this. Both myths and religious conceptions aquire meaning in this perspective—
adding pictures, stories, and emotions, thus making it easier to remember.
Alternatively, memory may be kept alive by rituals and ceremonies, with role
playing and drama. The Yir Yoront are a prominent example of both
approaches. The rich pantheon of the aboriginal sacred epoch can from this
perspective be interpreted as a mnemonic device, where the mythical landscape,
figures, and events serve to visualize and thereby fix in memory the elaborate
structure of aboriginal society. Their seasonal gatherings then served as
opportunities to reinforce the myths by ritual acts.
Nevertheless, the life of nomadic hunter/gatherers did not lend itself to really
elaborate social structures, and it was fairly well served by unaided memory.
Sedentary life brought great changes, however, with property, larger
communities and a much more complex social and economic life. The
limitations of memory then became a much more important constraint, and the
new physical structures also took on a role as mnemonic devices—the house
itself, for instance, became a powerful practical symbol, perhaps the most
important one until the invention of writing (Wilson 1988). Both private houses
and public buildings and monuments served as memory banks for common
values and organizational systems. By learning to live in a private house and
participate in the rituals and functions associated with public buildings and
monuments, a child would also learn the basics about the reigning social
organization and its own place within it. When property (especially the rights to
land) became the subject of inheritance, either for the individual or the group, it
simultaneously became the central structuring element in society, since persons
or groups possessing more or better quality land than others were able to
establish a permanent, superior position. The village itself, then, and later the
town, the city, and the nation, created a rich and evolving set of anchoring
points for social organization and stratification. Great engineering works
(especially watering systems) allowed for even more intensive agriculture and
larger cities, and technology proper became an important factor in the
development of human organizations. 2
2The architectural and construction technology of the ancient civilizations is impressive even today.
Everybody knows about the pyramids, but there are other examples. For instance, 18 canals with
fresh water went into Niniveh in Mesopotamia in the second millennium BC, the one best known
was 20 meters broad and 80 kilometers long. It was built in only 15 months. Asphalt and bricks
were used on the bottom, and in difficult passages the asphalt was covered with a foot of concrete,
lined with stone tiles (Dahl, 1984).
Yates (1966) continues this discussion. She argues convincingly that the
great cathedrals were designed along the lines of classical mnemonic theory—
with their numerous niches and room dividers, their carvings, friezes, pictures
and sculptures creating and furnishing the many “rooms” required to hold and
display the entire common memory of Christianity. And, as anyone visiting
national shrines like Westminster Abbey can witness, there was also room for
conspicuous representations of the ruling elites and their historical claims to
power. We might add that even modern churches, in all their stark simpleness,
still contain works of art depicting central themes in the Christian heritage.
This tradition of encoding organization and the social order in buildings is in
fact kept alive on broad scale, from private homes to office buildings, churches,
hotels and sports facilities. Parliaments in democratic countries invariably reside
in buildings of prestige at the very center of the capital; 3 banks and industry
erect pompous headquarters signaling their economic power; and hotels, the
new cathedrals of the international elites, today represent the most daring and
opulent architecture in the Western world. The encoding extends right through
the interiors, and seldom leave you in doubt—for instance—about the relative
rank of people you meet in their own offices.
Remembering must have put a growing strain on any society struggling to
escape from mere subsistence. For instance, an inscription from Greece in the
sixth century BC mentions the civic office of mnemones (literally,
“remembrancers”)—people entrusted with the task of remembering important
public information, such as rulings, precedents, and other events worthy of
chronicling. The office of mnemones illustrates well the iron-clad constraints of
oral society. Keeping records of key public information required the dedication
of many people’s memories. Just memorizing the minutes of day-to-day affairs
can be challenging enough, but then come affairs such as property rights, trade
agreements, debts, and obligations, even kinship relations beyond the immediate
family. As indicated in Chapter 5, from this perspective one can interpret feudal
3Interestingly, the Swedish parliament, in a fit of northern European rationalism, in 1971 moved out
of its cramped building in the historical central part of Stockholm to a new and supremely
functional structure in a more commercial district. The new building was very well suited to its
purpose, but the representatives discovered (to their own astonishment) that they resented it – the
symbolic value of its address, together with its businesslike modernity, collided headlong with the
parliament's perception of its own position. It really signified that the parliament was not so
important anymore, and that the power now resided in the cabinet and its departments, still located
in its traditional quarters in the area the parliament had left. After 12 frustrating years in “exile,” the
parliament rectified the situation by moving back to its old building in the geographical center of
power, now refurbished and extended through passages into adjoining buildings.
systems not only as power relationships, but also as mnemonic systems, which
structurally defined and thereby fixed in communal memory large sets of rights
and obligations.
The Art of Writing—an Administrative Technology

As the leading civilizations grew both in size and complexity, the limitations of
memory became a severe constraint, blocking highly needed developments.
Simple signs and representations of objects and numbers were not enough any
more, a new tool was needed—one that could express and preserve complex
information in an unequivocal way. And, it was administrative and commercial
needs that spurred the development—all available evidence suggests that the
first known script, the cuneiform of Mesopotamia (about 3500 BC), developed
directly out of a need for economic records in the growing economy of what
may be termed the world’s first states—for private business, for public
administration, and even for the economic side of the religious establishment
(Goody 1986): the first use for writing in the temples of Mesopotamia was
demonstrably not for recording religious material, but for temple administration.
Discourse, recording of poetry, and recording of religious and scientific material
all came later. In my view, this innovation—the art of writing—represents the
most fundamental, single technological breakthrough in the history of
organizations—at least until now, when the ripening information technology
may be able to equal or (given enough time) even eclipse it.
We may not normally think of writing as a technology, but that is only
because we are so used to it—it has become second nature to us, a complement
to speech. Plato, who lived during the crucial transition from an oral to a literate
society in Greece, thought otherwise—he considered writing an external, alien
technology, in just the same way as many people today characterize the
computer (Ong 1982). Its product is material (although with an immaterial
message content), and to produce it, one needs tools (chisels, styli, pens,
brushes, inkwells) and materials (stone surfaces, clay, papyrus, parchment,
paper, ink).
So, writing is indeed a technology—a most formidable technology. It gave
humans the immense power of absolute reminiscence and the ability to
communicate across time and space without loss of content or accuracy. It made
possible private records for business and personal use as well as public records
about property rights, taxation, and compulsory services. In principle, there
were no limits to the volume of information that could now be collected and
preserved, although in practice, externalized information has bulk—it occupies
space and is heavy to
transport. Clay tablets, for instance, were the medium of early cuneiform.
Technological innovations like parchment, papyrus and paper were therefore
just as important to writing as the invention of magnetic disks and semi-
conductor memory have been to computers.
The externalization of memory effected by writing was of course nowhere
near complete—it offered no way of representing the full web of memories,
with aspects such as sound, smell, vision, and pain, not to mention emotions—
except, of course, as verbal descriptions. But the aspect that was covered—the
ability to fix permanently any kind of verbal narrative and select pieces of
verbal and (in due time) numeric information—was of such importance for our
intellectual development and our ability to build organizations that it marked a
true watershed in human history.
The Significance of a Shared Memory

The enormous significance of the written record does not only rise from the fact
that it preserves information for an indefinite time, relieves the mind, and makes
it possible to accumulate information on a grand scale. Even more important for
the development of organizations is the fact that it creates a shared external
memory, accessible by a large number of persons—in fact, by any person so
authorized. An active file of written records allows many people to base their
work on the same information, and an update made by one person is
immediately available to the others and applies directly to their decisions. The
revolutionary new information integrity inherent in the active file of written
records provides therefore an implicit coordination of the people who work with
it—because of their common basis of information, they are now able to act with
considerable consistency without ever talking to each other. Since written
records will not disappear with people who work with them, it also means that
administration—public and private—can survive both normal and sudden
changes in personnel.
The permanence and accuracy of written records made it possible to build
much larger, more complex and more enduring organizations. Formal
explication of rights, obligations, laws, and treaties added the necessary strength
to relationships outside the bonds of immediate kinship and friendship, and
made the partnership and the firm viable over time (Goody 1986). As Goody
remarks (1986, p. 175), writing helps to make “the implicit explicit, and in so
doing extend the possibilities of social action . . . by creating more precise types
of transaction and relationship, even between trusted kin, that give these
partnerships the strength to endure in more complex, more ‘anonymous’
circumstances.”
Even if the the Incas proved that it is possible to build states and
organizations of considerable size without the help of writing, it is also evident
that once writing was available, much larger entities could be built and
maintained. The great Arab and Asian empires, the Roman empire, the Roman
church, and the Roman army were created and maintained by literate cultures,
with their written codes and files, their reports and orders unequivocally
committed to parchment, paper, or tablets. Even Genghis Khan, the warrior of
warriors, who conquered his great empire by the brute strength of the sword and
the horse, in the end had to rely on the delicate hands and minds of the Chinese
mandarins to maintain and run it.
We also know that the riches, power, and reach of the Hanseatic League in
Northern Europe increased many times over when writing was introduced as an
instrument of business, allowing the merchants to direct their increasingly
diverse and far-flung trading activities from their home bases, instead of
traveling with their ships to manage the trading directly (Buckmann 1991).
Wherever it was employed, writing also served to homogenize language and
establish national or cultural identities.
The Importance of Numerals

The revolution did not only encompass letters—numerals, numeracy, and the
development of systems and techniques for calculation have in many ways been
just as important as writing, although they usually receive less attention (maybe
because nearly all historians are men and women of letters rather than of
numbers!). The use of numbers is a crucial aspect of our economic and
scientific development.
The progress was very slow as long as Roman numerals were the only
instruments. The Arab numerals and notation we use today (which are really
Indian, and which did not become widely used in the Arab world until about
1000 AD) arrived in Europe around 1200 AD, but took several hundred years to
become widely accepted. They met with much resistance—one of the arguments
was that accounts kept with the new system were easier to forge, and in
Florence, as late as 1299, a person could be fined for doing his accounts in the
new system (Eriksen 1987). It nevertheless became popular with merchants, the
fledgling bankers, and the money changers, since they appreciated its flexibility
and efficiency much more than the theologically dominated scientists of the
day.
When Roman notation was finally superseded throughout Europe in the
course of the sixteenth century, development came rapidly, both in
mathematical theory and mechanical tools. The crowning achievement in
calculating techniques was the invention of logarithms, by the Scottish baron of
Merchiston, John Napier (his Mirifici Logarithmorum
Canonis was published in 1614), and mechanical calculators 4 were also

invented to relieve people from the drudgery of calculation. The new
developments made it possible to carry out much more computing-intensive
tasks than before.
Calculations remained labor-intensive, though, and continued to be a serious
bottleneck both in engineering and science, even if mechanical calculators were
considerably refined in the last half of the nineteenth century and the first half
of the twentieth. It was only when they were superseded by the electronic digital
computer that this constraint was really mitigated.
Technology Takes Off

The evolution of technology between antiquity and the Renaissance went
slowly. It is not without its interesting moments, and there was much
groundwork being done that would prove useful later. However, it is not
possible within the scope of this volume to delve into the developments of this
period, and it is certainly not necessary for our purpose, even if it is interesting
in itself. We shall therefore go directly to a number of major developments from
the Renaissance and onwards, and start with the first real mass production of
identical products, preceding the Industrial Revolution by more than 200 years:
the printing of books.
Printing and Mass Literacy

The invention of printing with movable type, unlike the invention of writing,
did not have much direct effect on our possibilities for organization building.
For instance, it did not affect the way most administrative records (both public
and private) were kept—tax registers and general ledgers still had to be entered
by hand. Correspondence was also still a manual affair. However, for society
the consequences turned out to be momentous, since printing had a
revolutionary effect on the economics of knowledge dissemination. By
liberating it from the constraints of hand-copied scripts, printing had profound
consequences for traditional structures of authority—both in society in general,
and in some of its most central organizations, especially the Church.
When most people were illiterate, knowledge resided in handwritten books
and a small number of persons. It could be—and was—controlled
4The first mechanical calculator was built by the German Wilhelm Schickard, not, as commonly
believed, by Blaise Pascal (Augarten 1984). Schickard built his machine in 1623, the year Pascal
was born.
by the hierarchies controlling the books. The Church was strong and unified
because the priests were the only source of the scriptures and their
interpretation. Dissenting interpretations could be suppressed by denying the
perpetrators access to the pulpit, or simply by disposing of them.
Gutenberg’s revolution meant that the written word became a mass medium
and thereby escaped control. By printing a book, one could reach a much greater
audience than by any previous communication method. There is little doubt that
the success of the Reformation rests squarely on the printing press and Luther’s
understanding of its power. The Reformation and the printing business fed each
other. According to Eriksen, only 40 titles were printed in Germany in the year
1500, and 111 in 1519. In 1523 the number had swelled to 498, of which an
impressive one-third originated from Luther himself. All in all, 418 had to do
with the Reformation! Not bad, considering only six years had passed since
Luther nailed his 95 theses to the church door in Wittenberg.
So the Holy scriptures were finally on the loose, and regardless of how much
the Church deplored the new situation, it could not reverse it. It tried to stop
unwanted books by banning and burning them, but the Index Librorum
Prohibitorum, intended to suppress dissent, functioned more as a medium for
advertising than anything else. Records show that a place on the index was the
best guarantee for commercial success—and Eriksen concludes (1987, p. 115)
that no Papal institution has contributed more to the promotion of science and
general enlightenment than the Index! The paramount religious authority of the
Church had received a deathblow, and it was clear that all monolithic political
and scientific authority was in similar peril.
The printing press revolutionized society by effecting two mutually
reinforcing changes. First, the printed book served as an extraordinary effective
medium for the accumulation and dissemination of knowledge—scientific,
political and, otherwise. Scientific treatises did not any longer have to circulate
in a small number of handwritten copies, but were suddenly simultaneously
available throughout the learned community as well as to the general public.
Second, the sudden availability of (relatively) cheap books greatly increased the
rate of literacy and pulled European societies through the transition from a
predominantly oral mindset to a literate one. Analytic and abstract thinking
became prominent in large segments of the society, notably the growing
bourgeoisie, sweeping aside the Church’s insistence on tradition and religious
mysticism and preparing the ground for a widespread acceptance of empiricism,
for new ideas and practical engineering—and thereby also for the modern
organization.
As scientific discoveries and theories now spread much faster, the pace of
scientific progress increased tremendously. Practical people
(the engineers of their day) published directions for crafts and industrial
processes, and the guilds began to lose their power. Innovation was stimulated,
and new ideas spread rapidly. It is safe to say that the immense economic and
intellectual developments of the centuries that followed would not have been
possible without printing. It is equally true that the development of democracy
owes a great deal to the dissemination of printed scientific and political
knowledge, first to the bourgeoisie and then to the population as a whole.
The printing press also led to a further normalization of language and
homogenization of national cultures. The later development of mass-distributed
newspapers served to develop an even stronger sense of national unity and
consciousness through the definition of national news and thereby the agenda
for national concerns.
Organization of Records
Even if printing in itself did not make much difference to how organizations
could be structured and run, there were innovations in record keeping as well.
There is scant documentation to be found, but we can at least establish that
elaborate systems for filing eventually were developed, with cards, folders,
filing cabinets, and Rolodexes. Vertical filing was in its time seen as a
substantial step forward (Yates and Benjamin 1991). The most monumental
contraptions developed (in the twentieth century) were motorized filing
cabinets, with storage bins in paternoster fashion and a height of several meters.
The modern offshoot of this traditional filing technology is the microfilm
reader, which greatly reduces the space needed. In principle, however, it is just
another filing cabinet.
Cross-referencing and indexing systems were also invented along the way to
help locate and retrieve recorded information. Information recording, storage,
and retrieval nevertheless continued to be a very labor-intensive activity, and
inventions such as the typewriter could not change that to any great extent.
But the crowning achievement of precomputer record keeping was of course
the punch card reader. The idea of punch cards as information carriers probably
originated in the textile industry, where the Frenchman Joseph-Marie Jacquard
in 1801 invented a large automatic loom, controlled by punch cards, for
weaving tapestries and similar complex textiles. (A demonstration program
weaving Jaquard’s own portrait in black and white silk needed 10 000 cards!). It
was, however, Herman Hollerith, the son of a German immigrant to the United
States, who put the punch card to practical use in record-keeping. His machines
(with electrical sensors registering the holes) were first used in the American
1890 census, where they were a great success. They allowed for far more
advanced statistics
than manual methods, including filtering and cross-tabulation, and delivered the
required information in a fraction of the time manual methods would have
demanded. In a test arranged by the Census Office before the 1890 census,
Hollerith’s machines tabulated data ten times as fast as the two fastest manual
methods. Even punching the cards went significantly faster than the manual
registration (Augarten 1984). The reason for the success was twofold: both the
mechanization itself and the new and much more stringent structuring of the
records that became necessary to use the machines (Yates and Benjamin 1991).
Both public institutions and large private companies—notably in the service
and utility sectors—invested in the new machines to ease the burden on their
record-keeping departments. Punch card equipment represented a significant
step forward, because it was the first invention that allowed mechanization of
record keeping—the “industrialization” of an old craft. These systems also had
many of the same properties as the first computer-based systems—indeed, for
three decades, punch cards (based on the Hollerith cards) were one of the chief
storage media for digital electronic computers.
In 1911 Hollerith’s Tabulating Machine Company merged with three other
companies to create the Computing-Tabulating-Recording Company (CTR),
which was eventually restructured and, in 1924, renamed IBM.
Communications Revolution
The other great field of organization-relevant innovation is the field of
communication. When memory was externalized and in principle no longer
limited, communication quickly became the bottleneck for organization
building. Communication has two aspects, both of which are important:
physical transportation of people and goods, and communication of information.
For a long time, they were (with very few exceptions) one and the same. Before
writing came into use, the only way to send information over any distance was
to send a person, a messenger. The advent of the written message made it
possible to do without him, but the message itself was still a physical object. So,
even though writing greatly increased accuracy, it did not necessarily increase
communication speed. For some of the very oldest texts, engraved on rocks and
large slabs of stone, the reader in fact had to come to the text rather than the
opposite.
Couriers and Mail Services

Clay tablets were more handy, though, and we know that both the Egyptians and
the Assyrians of northern Iraq were running regular
courier services as early as about 2000 BC (James and Thorpe 1994). In the
nineteenth century BC, the Assyrians had a dependable postal system operating
between their homeland and trading bases abroad. Excavations of one of their
merchants’ colonies, at Kultepe in Turkey, uncovered a mass of
correspondence, accounts and legal documents. The letters were small clay
tablets, complete with clay envelopes inscribed with the name and address of
the recipient. The service was even reliable enough for people to send money
along with their letters. Chinese postal systems are known from around 1000
BC.
The development of papyrus paper and parchment made transport easier, and
increased speed was sought by employing homing pigeons and horse riders. In
Egypt pigeons were used as early as the twelfth century BC, and regular, fast
horse transport from about 500 BC Cyrus the Great (550–530 BC), who built the
almost 2600 kilometer Royal Road from Sardis to Susa, also organized a regular
courier service with postal stations at regular intervals, and with relay riders
carrying messages around the clock. The whole distance was covered in a mere
nine days. A similar service, but much more extensive, was organized by the
Mongols in China after the conquest by Genghis Khan. Marco Polo reported
that Kublai Khan (AD 1260–1294) had 10 000 postal stations, with 300 000
horses employed on a regular basis (James and Thorpe 1994).
The great costs and extraordinary efforts involved in organizing these early
communication systems only serve to underline the vital importance that
communication has for large-scale organizations. Both Cyrus the Great and the
Mongol khans no doubt viewed their courier services as vital for keeping grip
on their empires, and they probably built on bitter experience.
However, none of these systems (with the possible exception of the early
Assyrian system) had the capacity to serve as an infrastructure for mainstream
organization building, and the beginnings of the first really comprehensive,
public mail service only took place in Europe millennia later. France was first in
1576, followed by England in 1590 and Denmark in 1624. International postal
cooperation was not institutionalized until 1874, when the world organization of
post offices was founded. Before that, one had to rely on merchants,
shipmasters, pilgrims, or other travelers if one was not wealthy enough to send
one’s own messenger.
Communications improved only at a very slow pace up through the
millennia. The Romans and Incas built extensive road networks, and sailing
vessels underwent a more or less continuous improvement. The improvements
were nevertheless not revolutionary, even if they were considerable. The great
clippers of the tea trade were very swift compared to their predecessors, but
they still needed weeks to cross the great oceans. To cut travel time from, say,
three months to six weeks was not enough to
significantly change the basic constraints of organization building. International

operations (the few there were), shipowners dispatching their shipmasters and
nations their ambassadors still had to operate like the Roman emperor when he
sent forth his governors: Choose reliable, sensible persons, give them general
instructions, and pray they can tackle the problems challenging them on their
own—since reports on critical events could not reach the headquarters until long
after the matter was settled.
Railroads, Telegraphs, Telephones and More

The 1830s produced the first real communications revolutions in historic time:
the railroad and the telegraph. The railroad got started first—it was officially
“invented” when Stephenson built “The Rocket” in 1829, even if it has older
roots (it developed slowly over about two centuries, starting with horse-drawn
carts on wooden rails in the sixteenth century). The railroad revolutionized both
travel and the transportation of goods. It is indeed a sobering experience for an
IT buff of today to read about the development of the railroads in the middle of
the nineteenth century. We like to think that we live in an age of unprecedented
technological change, and that never before in history have such great changes
taken place in such a short time. In a way, we are right—the pace of
technological development is both breathtaking and unprecedented. However,
the societal changes in the Western countries brought about by the combined
technological developments over the last 50 years are hardly as great as those
wrought by the developments of the railroads alone in the years from 1840 to
1890.
Consider the development of railroads in England: It “happened” almost
overnight. In 1836 about 20 short lines were in operation, most of them
concentrated in the Liverpool/Manchester and Newcastle/Middlesborough
areas. The longest single line of rail was about 50 kilometers. In 1848, only 12
years later, the whole of England was covered with a surprisingly fine-meshed
net of lines, connecting all major towns and scores of smaller ones. During the
busiest year of “the railway mania” (as it was called), 1845, Parliament
endorsed 623 new railroad projects, and more than 3000 kilometers of rail were
laid that year alone (Dahl 1984).
In less than 20 years all major diligence lines were eliminated, the canal
companies thrown from prosperity into financial difficulties, and travel times
drastically reduced. Suddenly, small country towns that had been living a life in
placid isolation were only hours away from the nearest big city, and large parts
of Great Britain was abruptly brought within one day’s travel from London.
In the United States, where distances were far greater and the pace of
development was even more furious (even if it came a little later), the effect
must have been still more revolutionary: In 1840 there were 4500 kilometers of
rail; in 1870, 85 000 kilometers; and in 1890, 263 000 (Beniger 1986). That
gives an average construction of 8900 kilometers per year from 1870 to 1890!
This is quite impressive, especially when one takes into account that America’s
population in 1890 was only twice that of England’s—or 62.6 million. In 1840
it had been only 17.1 million, about the same as England. The first American
transcontinental link was opened in 1862, and within the span of 20 years, travel
time from New York to Chicago was reduced from three weeks to three days
(Chandler 1977). Now, this is change!
The telegraph spread just as rapidly after its invention in practical form in
1835. 1851 saw the first undersea cable between Dover and Calais; the first
transatlantic cable (between Ireland and North America) came in 1866. Within a
few decades, national telegraph networks were linked all over the world, and
messages could be sped around the globe almost instantly. Semaphores,
beacons, and drums notwithstanding, this was the first time in history that
communication of information was truly separated from the transport of a
physical message, and could take place reliably and routinely over long
distances. In 1876 the telephone came, allowing people to speak to each other in
real time irrespective of location. The invention of radio in 1896 increased the
flexibility and range of telecommunications even further.
The first industry to profit from this double communications revolution was
the newspaper industry, which in the second half of the nineteenth century
turned a fairly exclusive phenomenon, the daily newspaper, into a mass product.
But the telegraph also had impact on business and proved especially important
for the growing railroad companies, notably because of safety considerations
(Chandler 1977, Beniger 1986). Most lines were only single-track, and the
movements of trains in both directions had to be closely coordinated. It was also
necessary to monitor railroad cars traversing the different networks to offer
point-to-point transport of goods without reloading. The telegraph was not fast
or flexible enough to have really decisive influence on business, however. It
was the telephone that really changed things. After its invention in 1876, it
spread rapidly, especially in the United States, and played a crucial role in the
growth of the great, national enterprises.
Of course, the communications revolution did not stop with the telephone and
the railroads. The development of steam and motor ships speeded up sea travel;
the automobile gave the individual the freedom of unrestricted, rapid personal
transport; and the airplane gradually achieved primacy in the long-haul market,
shrinking travel time to a
small fraction of what even the swiftest train could offer. The telephone became
ubiquitous; the telegraph was supplanted by telex; and radio provided even
ships, airplanes, and remote communities with the benefits of direct, real-time
communication. The printing press was supplemented by radio and TV
broadcasting, records, and video.
The combined effects of these developments were quite significant. Indeed, it
became possible to run large, international organizations as very tight ships,
with strong control exercised from a central headquarters, for many decades
before the computer became emerged as a widespread and significant tool.
Harold Geneen, for instance, did not need computers to keep his legendary, 20-
year iron grip on ITT—jet planes, telephones, and mammoth management
meetings were quite sufficient (Schoenberg 1985).
Today, cellular radio and satellite technology is severing the last physical and
geographical bonds of technology-mediated communication, at last giving us
the capacity to communicate freely without spatial restrictions. From early in
the twenty-first century and on it will be possible to reach and talk to people
almost anywhere in the world without even knowing where the person in
question is—as long as he or she has a cellular or satellite telephone and has
switched on the receiver to signal availability.
The Technology-Augmented Preconditions

We have already noted a number of significant improvements in our organizing
capabilities brought about by major technological innovations over the last 5500
years, beginning with the invention of writing in old Mesopotamia. However,
before we can move on to look at the organizational fallout, we need to
determine the extent of the improvements in more precise terms.
Memory
After the invention of writing, the amount of information that can be stored is in
principle unlimited. In addition, it does not deteriorate over time, unless the
physical medium itself deteriorates. Medium deterioration is of course a
problem, especially for works of art, but important information can always be
renewed. Information in verbal (written) or numerical form can even be
renewed without any degradation or loss of content. The stored information is
therefore accurate in the sense that what we get out is by and large equal to what
we put in. The reservation expressed by “by and large” does not pertain to
symbols themselves—the words in a
book printed a hundred years ago are exactly the same today, the book has not
“forgotten” any of them—but to their interpretation, which may change over
time, and even change so subtly that we are not aware of it. Normally, it is not a
serious issue, however. Pictorial information, on the other hand, remains
problematic—photographs, drawings, and paintings can be copied, but
information is invariably lost in the process, and there is a definite limit to the
acceptable number of generations.
Even if there are no theoretical limits to the amount of information we can
store, there are practical limits since externalized information has bulk. What we
have in our heads we can easily carry around; what we store in books is another
matter. Even paper-based information requires space, and there is a definite
limit to the amount of information you can keep handy. Microfilm shrinks the
physical bulk of information tremendously, but because of updating problems
and cost, it has been practical only for special purposes.
Then there is the problem of retrieval—externalized information can be
retrieved only when we know where it is. For limited amounts of systematized
(indexed) information, retrieval is unproblematic. It does not take long to locate
a card in a card file kept in a desk drawer, for instance, and most accountants
will retrieve a particular bill as long as you can inform them of the approximate
date. For large amounts of information, and information that does not lend itself
readily to indexation, the problems are considerable. Filing systems, library
systems and punch card equipment represent our best pre-computer attempts to
overcome this difficulty, but even in extremely well-run (non-computerized)
libraries or archives, one can only search on authors, titles and a very limited
number of keywords. The necessary indexing systems are also complicated and
very laborious to establish and maintain. So, even if we can store massive
amounts of information, we encounter the same problem as with our biological
memory: We can only access a tiny fraction of it. The difference is that the
constraint on access does not reside in an inscrutable biological mechanism, but
in the prohibitively long time it takes to search through large amounts of paper-
based information.
Punch card equipment offer significant improvements in certain instances. If
you want a single card (information on a specific customer) you still have to
look it up manually (it will take much longer to run all your cards through a
reader to locate it). However, if you want to send a letter to the 14% of your
customers who happen to own cars of a particular make, then the card readers
can sort them out for you in a small fraction of the time you would need by
manual means. You can also count and produce simple statistics fairly easily—
that is what punch card equipment was invented to do. But these improvements
contribute little to change the preconditions for organizing. The main
contribution of
files and archives, the possibility for implicit coordination, is not significantly
augmented by punch card equipment.
Information Processing
As noted in Chapter 4, the working memory is not able to hold the many
variables of a complex problem in an active state simultaneously—its capacity
is very limited. By providing an external memory that can hold and display a
much larger number of variables, writing helps us extend our working memory
and has proved to be a formidable technology for tackling complex problems.
Strictly speaking, our internal capacity for problem solving has of course not
been increased. But writing lets us utilize that capacity much better by relieving
it of the task of memorizing all the information relevant to a problem.
We use writing extensively for all kinds of mental work—whether we write
or sketch on a blackboard, a pad of paper, or a computer screen. This is both
true for individual and group work—groups frequently use blackboards,
whiteboards, flip-overs and overhead projectors to provide a common, external
working memory to aid their processing. For groups, it also helps synchronize
the minds of the participants and foster a common understanding of the problem
at hand. In larger projects the more formal decomposition of tasks, with its
problem definitions and job descriptions, serves the same purpose. In addition to
serving as an extended working memory during problem solving, storing ideas
and intermediate results, writing also allows us to do “preprocessing” by
collecting relevant material, systematizing it, and thus creating a platform for
analysis and decision making. Elaborate problems can be broken down into
smaller chunks and distributed over time as well as between members in an
organization.
Our processing is also helped by the phenomenally large vocabularies of
written languages, allowing for great precision in descriptions and arguments.
Ong (1982) considers this difference between an oral-only language and a
language with a literate tradition so fundamental that he gives the latter a
separate name: he calls the language of a truly literate society a grapholect, to
emphasize its dependence on writing both for its richness in words and its style
of expression. 5
In the modern world, problem solving and decision making are most often
part of a group process. We have already noted the concept of the common
extended memory for groups, and even more important is the
5Orallanguages can consist of as little as 5000 words. A rich literary language such as English now
has maybe as many as 1.5 million (Ong 1982). Comprehensive English dictionaries contain several
hundred thousand words.
ability to copy memos and reports and thus distribute them to everyone in
parallel instead of circulating hand-written originals. This makes group work a
lot more efficient. The real revolution, however, is found in the way writing lets
us distribute large tasks among a large number of persons, synchronize and
coordinate their activities, and communicate intermediate results between them.
The ability to collect, systematize, and store information that writing confers
upon us, and the way it allows us to distribute and synthesize problem-solving
efforts, has totally revolutionized our capabilities and has made it possible to
organize massive undertakings. 6 A literate society can therefore routinely tackle
tasks that would completely overwhelm any illiterate society.
For really important decision-making groups, far more expensive solutions
than whiteboards and overhead projectors have been devised. The mission
control centers for manned space expeditions, for instance, bristle with
technology and support personnel. The same is the case with the so-called
situation rooms for military high commands. These rooms are equipped to
handle the real-time conduct of major wars, when the chiefs of staff must
communicate almost constantly among themselves and with their units, and at
the same time be able to receive and have displayed vital information about
their own and enemy movements. A few large, private corporations, where the
leaders deem themselves to be in the economic and market parallel to war (the
slogan “marketing is war” is an indication of this sentiment), have invested in
their own “situation rooms,” where top management can receive high-tech,
graphics-filled briefings.
The situation room is an obvious advantage when it is necessary to monitor
and respond in real time to complicated events unfolding rapidly. Whether the
situation room is actually of any use in a corporate top management
environment, or whether it just serves to enhance the prestige of the users, is
open to conjecture. The closest thing to real, functioning situation rooms in
commercial organizations are probably the broker rooms in banks and
brokerages dealing in currency, stocks, or raw materials, and the control rooms
in industrial processing plants and power stations.
6An interesting example of a really massive, pre-computer project is the Allied invasion of
Normandy in 1944. Participating in the operation were hundreds of ships, thousands of aircraft, and
more than a million soldiers, in addition to a large number of tanks and artillery units. This
enormous force needed not only to be equipped and readied in the ports of southern England; when
the attack had started, it also had to be managed. The efforts of all the services and fighting units
had to be coordinated in real time, and there was no room for time-outs to collect one’s thoughts. In
addition, the invasion force had to be supplied with food, fuel, and ordnance, over provisional
bridgeheads, and with constantly changing front lines. Over just one of the bridgeheads (on Omaha
Beach), 15 000 tons of supplies and 15 000 soldiers were brought ashore per day in the most hectic
period.
A more cumbersome operation is problem solving or decision making aimed

at expressing the collective views of very large numbers of people. The basic
technique developed over the millennia is the method of representation,
whereby a small number of representatives are elected or appointed and are
conferred the power of deciding on behalf of the electorate. For political
purposes, the basis of this method was developed by the ancient Greeks, and has
since only been refined and modified to suit much larger electorates—where a
recourse to direct voting in the town square has not been a viable alternative.
In organizations, the organizational hierarchy itself is supposed to support
this process, but it is seldom adequate to serve all the different purposes and
views present in a large organization. Often, organized labor takes over parts of
the function (also using the method of representation), and almost always we
find informal leaders and mediators who perform simply because a number of
people trust them and see them as voices for their concerns. Shareholder
democracy is also a graft of the old political metaphor into the realm of
organizations.
We noted above that the development of democracy in Europe and North
America was closely related to the invention of printing, the spreading of mass
literacy and the growth of the newspaper industry. Mass media both informed
and homogenized, and they provided the public, common event space for a
national identity, a national agenda, and national leaders. They are also the
channel for the representatives’ communication with their constituents. The
development of sampled polling provides another strong feedback mechanism,
and although most would agree with the truism that an opinion poll is not an
election, we can see that politicians are getting increasingly sensitive to results
of such surveys—thereby acknowledging the fact that they do indeed represent
a very effective short-term feedback mechanism.
In large organizations, some of the same technologies are brought into
service to create a corporate identity and a corporate event space. Even a
technique such as polling has gained a solid foothold in the organizational
world. It is presently used by a number of consultancy firms to assess
organizational health and provide a basis for proposing remedies.
Our inability to handle more than one conscious thought process at a time has
of course not been modified by technology. However, the art of writing makes it
possible to extend time slicing considerably and thus get a much better hold on
complex matters. When intermediate results are committed to paper, we can
leave the subject, do something else for a while, and then return to pick up
where we left. We can thus keep many more parallel processes going than a
person in an oral culture can, and with much higher precision.
Whether or not our speed of deliberation has been improved is really a matter
of definition. There is no doubt that both writing and modern communications
technology (both physical transportation and telecommunications) have speeded
up parts of the total process. Information can be collected much more quickly,
reading allows for more efficient information absorption, and consultations are
much easier when one has access to telephones and rapid personal
transportation. The reaction time for large organizations has therefore been
considerably reduced, with notable consequences for innovation rate, the
average lifetime for products, and so on. Strictly defined, however, that is not
part of the deliberation process; it is a consequence of better communications,
improved information retrieval, and more powerful tools for analysis.
Deliberation proper is a process that is internal to our mind, and, as such, it has
so far not been noticeably affected by technology.
The Development of an Analytical, Literate Mind

By liberating the mind from the task of remembering a numbing load of existing
information, writing also set the mind free to work on contemporary problems.
Its use slowly spread from the economic sphere to the recording of religious
material; to the accumulation of knowledge; the creation of more widespread,
detailed, and lasting systems of law; and to immortalizing verbal art and historic
accounts such as poems, songs, and plays. It therefore had a profound effects on
the human mind and the way people thought about things, effects that were just
as important as the more immediate economic and political consequences, and a
vital precondition for the Industrial Revolution and the developments that
followed. “More than any other single invention,” says Ong (1982, p. 78),
“writing has transformed human consciousness.” It is, however, only quite
recently that we have become conscious about the magnitude of the difference
between the oral and the literate mind. This new awareness really started in the
1920s with Milman Parry’s groundbreaking theories about Homer’s epic poems
the Iliad and the Odyssey, but more widespread understanding was only
established by the works of Walter J. Ong (esp., 1982) and Eric Havelock (esp.,
1986).
What characterizes the oral mindset? First of all, people in oral cultures are
predominantly concrete in their thinking. They think in terms of physical
objects, actions, and events. Interpretations of things are always tied to their
context of events and actions. A striking illustration of this is provided by A. R.
Luria (Ong 1982) in his studies of illiterate, literate, and partly literate people in
Uzbekistan and Kirghizia in 1931–32. In one of
his experiments, he showed his subjects drawings of geometrical objects, such

as circles and rectangles. The illiterate peasants would invariably identify them
as representations of objects they knew—plates, buckets, mirrors, doors—never
as abstract categories. Students from the same communities (under training to
become teachers), on the other hand, immediately identified them by their
abstract classifications. Likewise, when asked to eliminate one of the four
objects hammer, saw, log, hatchet, that did not belong in the group, the fully
literate subjects would immediately eliminate the log (which did not belong to
the abstract class tools), whereas the illiterates would protest, saying that all the
objects belonged together. Both the hammer, the saw, and the hatchet belonged
to the situation of working with logs, which was their frame of reference. Ong
refers to a particular peasant, who, when pressed, eliminated the hatchet,
because “it doesn’t do as good a job as a saw” (p. 51). Asked to explain what a
tree is, a respondent answered: “Why should I? Everyone knows what a tree is,
they don’t need me telling them.” The abstract definition of a class of objects
named “trees” was simply not part of his mindset.
To us, this concrete, contextual orientation seems backward and even
childish—a patronizing view that is a product of our literate bias. That we
compare oral people to children is not surprising, since young children are
themselves oral people, living in a partly oral community with other children.
The process of acquiring a capacity for abstract thinking and symbol
manipulation, which we normally regard as part of the natural maturing process
of the child and the young adult, in reality represents a forced, culturally
determined transition from an oral to a literate mindset, a transition necessary to
become a fully functional member of a modern society.
The action- and context-centered state of the oral mind is easier to understand
if we consider the nature of the spoken word and the way it differs from the
written word. The spoken word is an event—it happens when spoken and heard;
immediately disappears and lingers on only as a fleeting trace in the memories
of the people present. It is also a social event, since an exchange of words
requires at least two people. Talking is action; it happens in a social context. To
be remembered, the spoken word must be forceful; it must tie meaning to
actions and events that are easy to remember. Rhyme and rhythm is often added.
The bard in an oral society is not primarily an entertainer (although he is also
that): he is even more a living memory bank, storing both the chronicles of
important events and the social code of moral conduct, all embedded in the
stories that make up his repertory. His memory, however, is neither infallible
nor incorruptible; he subtly edits his songs to suit the audience, especially
people of power and riches. Oral memory is
therefore unstable, and accounts of past events are often changed to suit the
present. 7
Because the volume of information that can be kept in memory and wilfully
retrieved is limited, accumulation of knowledge in our sense of the term is not
possible. An oral society will therefore be hesitant to pursue new information
and will stick to tradition, to proven knowledge. Since knowledge and
experience accumulate only in the living memories of the relatively few people
who survive to an old age, oral societies normally also hold their old members
in great respect: They become—like our libraries—repositories for society’s
accumulated knowledge. If they die before they have transferred their
knowledge to others, that knowledge is irretrievably lost.
It is probably no coincidence, therefore, that old people love to tell the same
stories again and again, and that small children are equally enthralled by the
hundreth repetition of a favorite tale. It is not unreasonable to claim that these
complementary characteristics of children and old people constitute an
important evolutionary advantage for oral societies, and that natural selection
has favored families with this trait. 8
The written word, on the other hand, is an unchanging thing; it has an
existence of its own, and reading and writing it is something one does in
isolation—if not physically isolated, at least mentally so. It is normally not part
of a collective experience. There is a literally a world of difference between 300
people gathered in the village féte grounds to hear a visiting bard relate the
latest news of the king’s exploits, and 300 students sitting in a large reading
room, all of them absorbed by their own particular book.
By eliminating memory as the prime storage for information, writing also
eliminated the need for oral mnemonics. The heroic personalities of oral
narratives, the narrative itself; the bonding between pieces of
7The myths of the Yir Yoront also had some of this flexibility, although only in details, and Ong
(1982) provides further examples of the flexibility of religious doctrines in oral societies. It is
interesting to reflect upon this in view of how present-day religious conservatives get locked into
the particular wording of the scriptures—as the Vatican in its view of contraceptives. Mainstream
Protestants and other liberal Christians have solved this and similar problems by abandoning literal
interpretation of the scriptures in favor of a symbolic approach. Christian fundamentalists, however,
remain prisoners of the fixed scripture, and will appear increasingly archaic as the world changes
and our knowledge grows, just as their counterparts in other religions with immutable holy books
(such as Islam).
8Inthis connection, it is also interesting to note the speculations by deBeer, mentioned in Anderson
(1990), on the reasons for the extraordinarily long time a human needs to reach adult stature – about
15 years, or around one-fifth of a normal life span; deBeer argues that the reason for the slow
physical development of human offspring (lagging far behind that of the brain, which is almost
complete at the age of 5) is the evolutionary advantage of prolonged dependency on the parents,
guaranteeing that the children do not leave before they have had time to acquire all the knowledge
necessary to become a competent adult.
information and descriptions of actions or concrete properties; the rhymes, the

rhythm, and the reinforcing redundancies (copia) of oral accounts became
superfluous in the written text. A written text is its own memory, it is always
there for reference. It fosters linearity in presentation and argument, as well as
precision and clarity. Distancing itself from the boisterous vigour of oral
dialogues, the written discourse grew cool and analytic.
The surviving texts from the classic period in Greece provide an illustrative
example of the transition from an oral to a literate state of mind. After the first
spurts to record oral material, such as the Iliad and the Odyssey, the Greek mind
slowly started to explore the new tool. Its expressions became less and less epic,
more and more analytic. Language became separated from man, it acquired an
independent existence, and Greek philosophers started to exploit and study the
analytic and epistemological properties of the language itself. Even Socrates
(470–399 BC) demonstrated an analytic approach that was clearly marked by a
developing literacy (Havelock 1986), although his dialogues retained an oral
unaffectedness right through Plato’s writing.
It is noteworthy that Socrates did not write down his thoughts himself, even
if he lived more than two hundred years after the invention of the Greek
alphabet—we had to wait for Plato (427–347 BC) to write “the first extensive
and coherent body of speculative thought in the history of mankind,” to quote
Havelock (1986, p. 111). Plato himself struggled with the transition: Despite his
distinctly literate, analytical discourse, he laid out his text in the form of
dialogues; he extolled the virtues of the dialogue as the supreme pedagogical
instrument but banned the poets from his city-state in the Republic, because they
appealed to the emotions alone and not to reason (Ong 1982, p. 80). Plato’s
pupil Aristotle (384–322 BC) brought the transition to its conclusion in his lucid,
analytical prose and his foundations of formal logic.
After Aristotle, the literate program was firmly established: an analytic
approach, a linear account, a concise prose, and a context-free language
separated from its author and the collective listening experience. Free from the
oral mind’s constant load of memorization of precious private and collective
information, the literate mind could allow itself to collect new knowledge, to
compare new with old, and to speculate about new theories. With an infallible,
extended working memory, it could tackle much more complex problems than
the oral mind, present them for large audiences, and preserve them for posterity.
A truly literate society therefore tends to be much more oriented toward new
things than toward tradition, and is much more prone to invent and develop. It
takes fewer things for granted, and has great confidence in its abilities to
understand, change and improve.
Communication
The area of communication has been one of far-reaching innovations in most
aspects. If we look at physical transportation, travel time (and cost) has shrunk
enormously, especially in the last 150 years. Paleolithic man could travel at
most a few tens of kilometers per day and normally did not stray outside an area
he could traverse in a few days. Today, it will not take more than 48 hours to go
from almost any major city in the world to almost any other. Travel between
capitals seldom requires more than 24 hours. From Oslo, I can go to New York,
have a meeting, see the town, and get back in less time than it would have taken
my grandfather to make a return trip to Bergen on Norway’s west coast, and his
grandfather again to go one-way from Oslo to Lillehammer—the venue of the
1994 Winter Olympics, some 180 kilometers to the north.
The increase in speed for the transportation of goods is just as great if one is
willing to pay the price. Books, machine parts, flowers and fresh fish is speeded
around the globe in airplanes, and a salmon can be served in a Tokyo gourmet
restaurant as little as 48 hours after it is snatched from its enclosure off the coast
in the south, west, or north of Norway. 9 But even for less perishable and costly
goods, intercontinental transport takes a matter of weeks, or maybe a month or
two at the most. Mail services are universal, and even if the distribution speed
varies from country to country, two places in the world are seldom more than
two weeks apart. Courier services offer considerably faster delivery. These
improvements in transportation make global trade and global organizations
eminently feasible, especially when paired with the advances that have been
made in the communication of information.
For the developments in global transportation pale when compared to the
strides taken by the communication of information. Our range of
communication has been extended to interplanetary proportions, and the speed
to the speed of light. For most practical, earth-bound applications, we talk about
instant, real-time communication regardless of distance. This holds both for
one-to-one (telephone) and one-to-many communication (radio and television
broadcasting). In addition to the electric and electronic media, mass distribution
of books, newspapers, and magazines has made possible a massive exchange of
information. Return channels such as sampled opinion polls and “letters to the
editor” make leaders aware of prevailing opinion, and even make people aware
of what other people think—in itself a very important precondition for opinion
formation.
9Such rapid transit (with its concomitant high transport costs) is not the usual—most of the salmon
requires three to five days to reach its destination. Only the most experienced experts will note the
very slight difference, as salmon packed in ice keeps very well.
For organizational purposes, the telephone became of great importance, and

numerous effects have been noted (Pool 1983). One of the first of major
importance was the support of a physical separation of plant and office. Until
then, it had been common to house the office in a building adjacent to the plant
itself, to ensure sufficient communication between administration and
production. The telephone allowed the office functions not immediately
connected to production to move to the central part of the cities, closer to
customers and finance institutions.
By making the physical separation of different parts of the enterprise more
feasible, the telephone in fact supported centralization of decision-making while
allowing physical dispersal of organization units. The control provided by the
telephone convinced managers that they could locate major business units far
away from the main office, and put greater emphasis on proximity to major
markets or sources of workers, energy, or raw material.
The telephone also had great effect on the speed of many types of
transactions, from banking to the ordering of goods—especially perishable
goods. Pool cites a 1906 article in Scientific American on “The Sociological
Effects of the Telephone,” describing how oyster barge men were put out of
business because restaurateurs could phone their orders directly to the oyster
planters. “In general,” Pool notes, “the greatest business use of the telephone
has been in finance, commerce and where complex logistical coordination is
required.” Even the railroads, after much hesitation, converted from the
telegraph to the telephone for train operation. As Pool goes on to say (1983, p.
68),
It permits coordination of pieces of that complex clockwork which is the economic

system. It is used millions of times a day to control production, shipping,
recording, and selling. It permits the operation of a complex division of labor. All
of that was recognized from early on.
With the new transportation infrastructure for information, goods, and people in
place, it suddenly became feasible to build efficient distribution networks, to
trade reliably over great distances, and to exert a degree of control over
branches and subsidiaries in remote locations that had been impossible up until
then. In retrospect, it is evident that the telegraph and later the telephone went a
long way toward easing the constraints on organizations spanning great
distances, and the improved physical transport offered by the railroads took care
of much of what was left.
Among the predictions recorded by Pool, some are more notable than others.
One prediction in particular should have a familiar ring for proponents of
electronic mail and video conferencing: that the telephone would reduce travel.
This was attributed both to the use of normal telephone calls and to telephone
conferencing, which was easy to set up
in the first decades of telephone service because of the flexibility of the manual
operators. Then, as now, it was difficult to discern any reduction in travel as a
result of the improved telecommunications, but it is of course impossible to say
what the volume of travel would have been without them. After a long period
when mechanical exchanges made telephone conferences all but impossible,
they become a viable service again in the last decades of the twentieth century,
when computer-based switches were introduced. The jury on travel reduction is
still out.
The telephone has obviously made group work easier, has probably made a
number of meetings unnecessary, and has made it more feasible to organize
work groups without co-locating the members. Experience tells us, however,
that it cannot fully replace face-to-face meetings. The telephone is very useful
for questions, informal discussion, and general conversation between two
persons, but most people experience telephone meetings as awkward and rate
them as clearly inferior to “real” meetings. Reports about regular use of
telephone conferencing almost invariably involve small, close-knit groups of
people who have been working together for a long time and know each other
well. The most common exceptions are training and sales conferences, which
more have the character of broadcasting (Johansen 1988).
The videophone, or “picturephone” as it was originally called, was
envisioned as the next step in telephony, and it has prematurely been predicted
at regular intervals. It was also expected to reduce travel significantly. Pool
quotes a 1914 article in Scientific American in which it was argued that
something soon had to be done to check the congestion of the city, and that the
fundamental difficulty seemed to be the necessity for close proximity when
transacting business. The telephone and the picturephone were expected to
change that. Since the late 1960s, conference television has been an available
instrument in a number of countries, but it has met with limited enthusiasm. The
main reason may be the high cost or poor quality involved so far (generally, you
have had to choose one or the other), but people also feel that it is “artificial”
(Johansen 1988). Even here, the most successful use has been in training and
sales, where the broadcasting aspect is strong.
The telephone even eliminated jobs—most notably, the position as messenger
boy. This was far from inconsequential: there were considerable numbers of
messenger boys employed in every large city, giving many a poor family a
welcome extra income. The advent of the telephone left them bereft of their
jobs, but, on the brighter side, it therefore allowed them to stay in school.
It is irresistible to end this discussion of the telephone with another quotation
related by Pool, showing how established mental sets can prevent an otherwise
knowledgeable person from perceiving and
understanding the portents of a new technology: in 1879 (three years after its
invention) Sir William Preece, then the chief engineer of the British Post Office,
told a special committee of the House of Commons that the telephone had little
future in Britain, even if it seemed to be a success in the United States (Pool
1983, p. 65):
There are conditions in America which necessitate the use of such instruments
more than here. Here we have a superabundance of messengers . . . The absence of
servants has compelled Americans to adopt communication systems.
The Iron Constraint on Information Exchange

Unfortunately, the revolutions brought about by the externalization of memory
and the development of communication technology has not had any parallel
when it comes to our personal interfaces with the real world: our basic
capabilities for information input and output have not been changed much by
technology. Yes, we can now instantly connect to and talk with a person sitting
halfway around the globe, but our information exchange can still not exceed 250
words per minute—in fact, because of the slight deterioration that always
follows transmissions through the telephone, we cannot even talk as fast as
when sitting in the same room. True, we may now watch events taking place on
other continents on TV in full color and in real time, but we cannot absorb the
televised information any faster than information reaching the eye directly.
The invention of writing helped somewhat, because we can read a little faster
than we speak (and thereby listen). Written material is also more generally
accessible (it is physically separated from the originator) and thereby allows us
to devote more hours per day to information intake. The typewriter has likewise
made writing a little faster (and the result generally more legible), but the
increase in speed has not been dramatic and has not made any real difference for
our organizing abilities. The slow speed at which we can absorb and output
information continues to be a source of frustration, and it remains an iron-clad
constraint on our organizing abilities.
Much energy has been expended through the centuries trying to alleviate this
shortcoming, and we are still striving to cope: from the perspiring student
poring over his books the last weeks before the exam to the distressed CEO
spending even Saturday and Sunday reading reports, memos, and magazines,
desperately trying to catch up with the information constantly pouring in. No
breakthrough has yet been achieved, and the measures we employ are still the
tried and tested ones.
The first main remedy is selection—ferreting out the most relevant pieces of
information. It is typically a demand placed by managers on their subordinates.
Many top-level managers set a limit of one or two pages on memos, contending
that what cannot be presented on one page is not worth knowing. Selection is
also standard procedure in the news media. The problem with selection is that,
to obtain a good result, the selector(s) must know exactly what is relevant for
their masters/customers, which they of course do not—at least not fully.
Managers are normally aware of this, and seek out alternative information
sources as well. Nevertheless, we all have to depend on others doing selections
for us, and we never know in which way they are biased—and, too frequently,
we are not concerned. Only professional investigators and researchers routinely
question the reliability and completeness of the information they receive.
The next countermeasure against information overload is concentration—
presenting the information in as compact a form as possible. The copywriters in
large newspapers are infamous for this—reducing the lavish prose of a proud
journalist to a few close-cropped sentences. Television news and advertising are
also arenas for extreme compactness.
Concentration is not only achieved through expert editing and economy in
words, however. Numbers are more readable if presented in a systematic layout
such as a table, and the information they contain is even more accessible if
presented as graphics. The brain’s capacity for pattern recognition and visual
processing is massive, and any information that can be presented in pictorial
form will be grasped much quicker than the corresponding numbers. We must
still conclude, however, that our capacity for absorbing information has not
been dramatically enhanced in general, and is lagging far behind the huge
increase in our storage and communication abilities.
When it comes to information dissemination, the picture is mixed. In face-to-
face communication, our capacity has not increased. If we extend the concept to
letters, typewriters (and later, word processing computers) have marginally
increased our rate of output, but it is nothing to brag about. Our capacity has
increased beyond all measure, however, when it comes to one-to-many
communication. The printing press, radio, television, records and tape of
various kinds have totally revolutionized the human capacity for addressing
others. Of course, it is an intrinsic property of the concept of mass media that
channels are not open to everyone—but in principle (at least in an open society)
they are open to anyone. The powerful nature of the mass media’s
communication capacity is reflected in the exertions made by totalitarian
regimes to control them, and in political parties’ efforts to use them.
For organizations, the new media mainly allow for easier communication
from organization leaders to their subordinates and for more efficient
information distribution to customers or other organizations in the environment.
Even fairly small organizations have their internal circulars, and larger
organizations often have elaborate internal newspapers or magazines. Mailing
lists are kept for many purposes, and almost all commercial (and many public)
organizations advertise. Indeed, mass media advertising is one of the
prerequisites for the formation of large companies building their business on
national or international brand names. By dramatically enhancing our capacity
for one-to-many communication, mass media technology has also significantly
enhanced our possibilities for organizing and sustaining large organizations.
Serial Mind, Parallel Action

While our minds remain serial, and refuse us to divide attention, we have for
some purposes overcome our limited ability to do things in parallel: We have
built automatic machines. An automatic machine mimics the work of humans—
either directly, as with mechanical arms gripping objects and moving them from
one place to another, or indirectly, as when car bodies are painted by
electrophoresis as they move through large chemical baths.
Automation has to a certain extent made it possible for us to transcend both
our innate problems of coordination and of attending to more than one task at a
time. Pure mechanical automation, which started in earnest in the middle of the
eighteenth century, has made great strides, and quite complex products can be
manufactured with a minimum of human intervention. Through the
programming laid down in the mechanical design of machines and tools, a
variety of tasks can be carried out in perfect coordination, and with a good
number of parallel sessions per human operator. In addition to greater speed and
precision, an automated system wastes no time making decisions—the decisions
have been made once and for all through its design.
Automation, then, can be viewed essentially as canned action, as the
enactment of previous design. The machine is, so to speak, a set of crystallized
decisions, the result of an extensive information-processing undertaking, ending
with a carefully choreographed set of movements and work operations. It
represents a total externalization of a plan for a specific production process.
Once forged in steel and powered by steam or electricity, the automatic machine
can repeat its designed actions again and again. And once designed, it can also
easily be replicated, and our canned actions can finally be carried out in parallel
by a large number of similar machines.
This is quite different from using power to increase our physical might, as we
do, for instance, with bulldozers. A bulldozer allows a single individual to move
hundreds of times more soil than he or she could with a shovel only, and thus
increase his or hers productivity hundredfold. In principle, however, it is no
more than a power shovel. The automatic machine, on the other hand, replicates
on its own certain productive aspects of the human organism. We cannot really
say that such a machine is processing information, since all it does is to repeat a
sequence of movements. But, just as much of the human work in an
organization, those machine movements are carefully planned and designed.
Even pure, “old fashioned” mechanical automation is thus a powerful
expression of collective information processing, and it can be viewed as a part
of the organization—just as humans are. This aspect of automation becomes
even more evident when we cross the threshold into the world of information
processing, which opens up new areas for automation. Before the advent of
computers, for instance, there was scant automation in the realm of
administrative work. The most advanced examples were mechanical calculators,
bookkeeping machines, and punch card equipment.
By harnessing external sources of physical power and multiplying our
physical operations, automation has become the main basis for our phenomenal
growth in material wealth. It has made it possible for us to produce goods in
volumes that are many orders of magnitude greater than before. At the same
time, however, it has forced standardization. For even if automation of physical
operations is very efficient, mechanical automation also results in a rather
inflexible production apparatus. Machines all have their very definite purposes
and ways of operating, and their repertoire can normally only be changed by
physical modification. It takes considerable time and effort to externalize
decisions in the form of a machine. Mechanical automation is thus conducive to
mass production of standardized products, but generally extremely sluggish in
its response to changes in consumer preferences.
Emotions
Can technology “augment” emotions? It may seem preposterous to ask such a
question, but we must consider the evidence. Control of raw emotions has
always been of great importance in human societies, and it still is. We should
therefore expect efforts to improve it by tool development. As Wilson notes
(1988), even a hunter/gatherer society such as the bushmen of the !Kung San
“dread the prospect of tempers flaring out of control,” and place much emphasis
on the control and management of emotions. This is also the case for the Innuit
Eskimos. In general, it seems
that societies in which people are forced to live very close to each other develop
strong norms demanding tight control of emotions and encourage mental
training and the development of techniques for this purpose. Both China and
Japan are famous for this, and India is the home of yoga and other mental
techniques.
In industrialized societies, handling of emotional problems and aberrations
has typically enough been professionalized and has become the subject of
scientific research (psychology and psychiatry). And, as wizards of the material
world, we have not been content with techniques and “empty talk.” The
development of psychopharmaca has proceeded from the narcotic herbs of tribal
society to a broad array of modern medications. Even if we are not counting the
more exotic (and even illegal) drugs, both sedatives and stimulants are routinely
used by a large number of people every day. Hochschild (1983, p. 54) even
reports that nurses in the medical department of AT&T “gave out Valium,
Darvon, codeine, and other drugs free and without prescription” to help
employees cope with stress and boredom on the job.
If we follow Morris (1967), we may also include much literature,
photographs, and movies/TV as emotion-controlling technology. Morris notes
that our innate interest in sexual activities outside the pair-bond is strong but
socially denounced, and that the solution is sex by proxy, from the innocent
romantic to the hard-core pornographic. A similar case may be made for our
bent toward adventure and heroism, especially when young—it feeds the movie
industry like nothing else. The movies and the corresponding books may
provide a much-needed cathartic effect for postmodern youth trapped in a
society where the challenges of life are increasingly abstract, and where
physical excitement and adventure is channeled into sports that are themselves
ever more regulated and loaded with safety precautions.
Our basic emotions and desires have not changed, then, but we have learned
to control them to a certain degree both through mental techniques, the use of
social norms, projection (the use of proxies through literature, pictures, or
movies) and (in some instances) medication. The basic techniques of self-
control and the instrument of social norms are nevertheless very ancient indeed,
and it is doubtful whether (for the purpose of organization) the differences
between modern man and his Paleolithic ancestors are really significant in this
respect. Our emotions are still among the major sources of organizational
conflicts, disturbances, and failures.
Conclusions
The discussion in this chapter has shown how our technological achievements
from the advent of literacy and up until the invention of the
electronic, digital computer has greatly improved our potential for organization.
We may point to three periods in particular: the slow development of literacy
in the great Eastern and Western civilizations in the long period from about
3500 to 600 B.C., the development of modern numerals and mathematics from
about 900 to 1600 A.D., and the development of automation, physical transport,
the telegraph, and the telephone from about 1800 to 1945. We can safely
conclude that this pre-computer technology vastly improved our storage and
communication capabilities, and provided a solid augmentation of our basic
problem-solving and
Memory Unlimited amounts of information can be stored outside the brain for
indefinite periods without loss of content.
Main constraints: Large amounts of information require
considerable physical storage space, retrieval becomes problematic
when volume increases.
Processing Greatly improved by better preprocessing and storage of intermediate

results. Far better monitoring of complex events. Much better
possibilities for distributing tasks over time and between many
persons, as well as for coordination and cooperation over distance. A
literate mindset that is more analytical and more interested in change
and improvement. Vastly improved capacity for parallel actions
through unlimited replication of “canned” processes with mechanical
automation.
Main constraints: Externalization of information processing still not
possible. Processing capacity per se not significantly increased.
Communication Physical transport revolutionized, communication of information

doubly so—information can be transmitted instantly regardless of
distance. Mass media allow information dissemination on a massive
scale. For individual information absorption and dissemination,
however, there are only minute improvements in speed, although
accessibility is greatly improved.
Main constraints: High cost of large volume point-to-point electronic
communication and low social “bandwidth” of the affordable
channels, the iron constraints of our own input/output limitations.
Emotions Some improvements in control.
Table 6-1: Main technology-based changes in preconditions.

decision-making abilities—especially when it came to groups and large

undertakings. However, its effect on our “processing” was almost totally
indirect—it supported problem-solving and decision making only by storing,
arranging, presenting, and communicating information, and could not augment
our processing capacity directly. The key points are summarized in Table 6-1.
These were the main, technologically based changes in the preconditions for
organizing that allowed the great changes in organizational appearance and
functioning that took place in the nineteenth and twentieth centuries.
147
7 The Modern Organization
“Everywhere in the world the industrial regime tends to make the

unorganized or unorganizable individual, the pauper, into the victim
of a kind of human sacrifice offered to the gods of civilization.”
Jaques Maritain, Reflections on America, 1958.
Into the Modern Age

It took a long time to discover and exploit the new organizational possibilities
opened up by the evolving technology. The state continued to be the chief
domain for organization on a significant level, and the feudal structure remained
the main organization type. Religious hierarchies were also modeled on the
feudal state, with divinely sanctioned offices corresponding to the nobility’s
secular positions based on lineage and inherited rights to land. The
administrative technologies provided by literacy were only used to support
already existing organization practices. Literacy was limited to a very small part
of society—the ruling elites, a number of their servants, the religious
establishment, and a few others—and so the information economy of the feudal
structure was still necessary to manage large states and large religious or
military organizations. Economic organization remained small-scale, mostly of
the Simple Structure type—either with one owner/manager, or with a small
number of partners dividing the managing role between them. (A partnership
may also be viewed as the commercial variant of the Councilcracy.) The
structure provided by family ties was almost always there as the dominant pillar.
This situation lasted for a long period of time—more than 2000 years, if we
count the time from the more widespread development of literacy in
the Mediterranean until the start of the Industrial Revolution in England. There
were of course important developments taking place in that interval, but by and
large they were all variations and refinements within the scope of the
organizational forms discussed in Chapter 5. Not even the considerable growth
in literacy that followed the development of printing in Europe and the United
States from the sixteenth century onward produced any significant
developments in organization—with the possible exception of the slow but
steady growth of representative democracy in England. The revolutionary
changes in the preconditions for organizing that were provided by the invention
of writing were simply not exploited. This fact can serve to remind us that new
capabilities do not force development in themselves—they remain potentials
until they are actually discovered and explored.
The best explanation we can provide for this failure to take advantage of
available administrative technology is that the material needs for large
organizations besides the state, the armed forces, and the Church were simply
not there. The vast majority of people all over the world still worked the land,
the traditional organizational structures were quite adequate, and the
preeminence of the land-holding nobility in the body politic of the feudal states
was in harmony with this state of affairs. The craft-based production of material
goods still did not achieve a volume where production needed more
organization than the direct supervision provided by a master or a “foreman.”
The same was the case with trade—as long as transportation was slow and fairly
expensive, volume was low and could easily be handled by the traditional
merchant and his few assistants. Chandler (1977) describes how this was the
case in the United States right up to the 1840s. Even in Europe, where both the
scientific revolution and industrialization started, there was little innovation in
organization before the middle of the nineteenth century.
But, as we now know, a revolution was brewing in Europe and the United
States, where a long and slow accumulation of knowledge and development of
new tools now accelerated. The application of external power (especially steam)
and the construction of machines for manufacturing boosted the output of the
growing factories, whereas the advent of trains, swift sailing vessels, and later
steam ships provided cheap, rapid transport. These developments cleared the
field for business ventures many orders of magnitude larger than before.
The new opportunities, however, could not be realized through the
traditional, small-scale business organization. Building commercial
organizations capable of handling much larger numbers of people and spanning
much greater distances than before thus became one of the major challenges of
the new entrepreneurs.
7 The Modern Organization 149
The Growth of Complexity

The Starting Point
Prior to the Industrial Revolution, work tended to be holistic in character—one
person usually carried out a complete set of tasks. This does not mean that there
was no specialization at all—even hunter/gatherer societies show some
rudimentary differentiation of roles, and medieval society displayed a rich set of
specialized occupations, notably in the crafts. Even so, work was usually quite
varied. A craftsman, for instance, would do almost all the work on an object,
from obtaining raw materials to the finishing touches and even the delivery to
the customer. Apprentices and journeymen would often take care of the more
tedious chores, but the craftsman was always in control of what happened—
coordination took place by means of informal communication and mutual
adjustment, intuitively and without significant formalized structure.
Farmers likewise carried out all the different tasks associated with their
position, and merchants would usually have a very personal relationship not
only with their customers and suppliers, but also with the different work
processes required. They might hire laborers to handle the goods, and clerks to
do various administrative work, but there was not any great degree of
specialization. By and large, business was a family affair or a partnership—
where the partners worked largely in parallel with similar tasks, rather than
dividing work according to functional specialization (Chandler 1977). You may
well say that this is the spontaneously natural way to work for a human—it has
dominated all the way from hunter/gatherer society up until the spread of
industrialization.
If there was a hierarchy in the business, it would typically be of the task-
continuous type (Clegg 1990), in which a person at a certain level would master
all the activities of lower levels. The owner, mastering all the tasks in the
business himself, would usually be very competent to coordinate all of the
activities. The importance attached to such mastery is illustrated by the fact that
it was usually considered obligatory for the owner’s heirs to work in
subordinate positions for many years, with all the different tasks, before being
considered ready to take part in the direction of the business.
Scaling Efforts
The first steps toward industrial production and modern organizational forms
consisted of a more intensive exploitation of traditional approaches. Entrepre-
neurial artisans increased the number of apprentices and journeymen in their
shops; in some trades, such as building and
shipbuilding, masters increasingly took on total contracts and then put together
teams of the necessary craftsmen to complete the job. But even this extended
approach was kept within the Simple Structure model—the workforce or team
was usually not larger than the number of people the entrepreneur himself could
oversee.
Another approach to expanding the production of goods was the “putting
out” method, whereby a master or a merchant contracted out work to
households or independent craftsmen. This method of production was
extensively used in Europe, and one merchant could have more than a hundred
such contractors working for him. None of these practices were new, however—
according to Goody (1986), they were already routine in Assyria more than
3500 years earlier—in 1900 BC!
People working under the putting-out system had considerable freedom in
their daily life; they could to a large extent decide their own working hours. To
illustrate this, Chandler quotes the historian Blanche Hazard (Chandler 1977, p.
54):
The domestic worker had enjoyed all the latitude that he needed or wished. He
sowed his fields and cut his hay when he was ready. He locked up his ten footer 1
and went fishing when he pleased, or sat in his kitchen reading when it was too
cold to work in his little shop.
Descriptions such as this make us reflect on what we have lost through

industrialization—and it is no wonder that the first factories had great trouble
getting skilled persons to show up regularly for work every day. They still
tended to go fishing or swimming when the weather was good, and to be absent
the day after a particularly intense celebration. Such privileges are not easily
surrendered, and we may ask if not some of them are now in the process of
being restored under the banner of teleworking.
In the struggle to find suitable new ways of working, old practices were
stretched in yet another way, even as the trend toward specialization and factory
organization was becoming more pronounced. Building on traditions from craft
shops and contracting, internal contracting became a fairly widespread way of
organizing production—especially where specialization was not too extensive
(Chandler 1977, Clegg 1990). In this type of arrangement, the factory owner
negotiates a contract with a number of subcontractors specifying the quantity of
goods to be delivered over a certain time period, usually a year. The factory
owner would typically provide floor space, tools, raw materials, and so on, and
pay the subcontractor a lump sum, possibly with the addition of a minimum
1 A common term for the small workshops such workers usually put up as annexes to their homes.
A length of about ten feet was the usual size of these shops.
foreman wage. The subcontractor would then hire his own people to do the
work, pay their wages from his own contract money, and supervise their work
himself.
Another form of internal contracting developed later as a result of the merger
waves in the last half of the nineteenth century (Chandler 1977, Clegg 1990).
The new amalgamated companies were too large to be managed by traditional
means, and administrative techniques for large-scale, task-discontinuous
organizations were still not fully developed. The former owners were therefore
often asked to continue, but now as internal contractors, with lump-sum
payments for providing agreed volumes of their goods or services within the
larger whole. As Clegg says, such arrangements to a large degree reconstituted
the task-continuous style of management in pockets within the larger
companies.
Internal contracting had the same advantages in information economy as the
feudal structure: It effectively encapsulated the production process within the
work group or the acquired firm and relieved the top management from
worrying about the details of the daily work process. They only had to manage
their contractual interface to the subcontractors. These advantages have helped
such arrangements to survive to the present day—even in advanced industries in
the leading countries of the world. For instance, you will find that Japanese car
factories are surrounded by entire districts composed of small, family-owned
workshops (Clegg 1990), barely surviving in a harsh contracting system that can
be viewed as a combination of internal contracting and putting-out.
The growing intensity in the use of practices such as contracting and putting-out
that marked the immediate preindustrial period in Europe and the United States
can definitely be interpreted as a sign of stress on the old order. Technological
progress was accelerating, and the craftsmen and businessmen of the day were
feverishly trying to accommodate the changes by scaling up their traditional
work processes and organizations.
However, their traditional organizational practices could not be scaled up to
exploit the fine-grained specialization that was now developing, with its
accompanying needs for rigorous planning and detailed control of activities. In
the internal contracting system, coordination between the subcontractors was
mostly handled on an informal basis, and it was often less than optimal. Top
managers had little information about real costs and waste in production, for
instance. A closer coordination of the entire production process and more direct
supervision and control of costs and quality was difficult to achieve with the
traditional approaches, and the concept of centralized planning and total
coordination of production gained ground. Even among traders and transporters,
the increasing volumes of goods and raw materials required new approaches.
The Birth of the Machine Organization

The New Needs
As we have already noted, the most pronounced features of industrialization
were the use of increasingly sophisticated fabricating machinery and the
mechanization of transportation, both accompanied by growing functional
specialization. In their turn, these developments invited factory production on
an ever larger scale. To coordinate the efforts of large numbers of specialized
workers, new work practices and organizations were needed.
There were probably two main reasons for the development of functional
specialization: one was the growing pressure toward increased productivity, the
other the scarcity of skilled workers (craftsmen). The limited availability of
craftsmen and the several years needed for training one represented a serious
bottleneck for industrial growth. Extensive specialization thus became a
prerequisite for the rapid growth the industrialists pursued: it allowed them to
hire unskilled laborers and train them only in their particular narrow tasks—a
process requiring maybe only days or a few weeks at the most. In addition, such
workers were much easier to control and command than the traditionally quite
independent-minded craftsmen.
The decisive advantage of specialization, however, was its impact on
productivity and the way it supported mechanization. It was specialization and
mechanization together that brought us the productivity revolution of the
nineteenth century. The roots of specialization are certainly older and can be
found in early attempts to organize craft production in larger units, as in the
tenth century English textile “industry” mentioned by Mintzberg (1979).
However, until the advent of the mechanized factory, it played no important
role.
One of the first descriptions of thorough specialization was Adam Smith’s
famous example of the trade of the pin maker, presented in The Wealth of
Nations (published in 1776), where he identified 18 different operations
involved in making pins. He also observed that this specialization resulted in a
productivity far superior to that of a traditional, holistic approach. Although the
pin-making process primarily represented an elaboration of the craft approach, it
clearly pointed toward a new era. It was not until the nineteenth century,
however, that functional specialization became widespread, was supported by
mechanization, and ushered in a new type of organization.
In the United States, extensive specialization was first implemented in the
manufacture of small arms early in the nineteenth century, coinciding with an
advancement in the precision in manufacturing to the point where parts became
interchangeable, allowing products to be assembled without
the extensive adjustments of the preindustrial and early industrial era. Eli
Whitney, the inventor of the cotton gin, was the first (in 1801) to demonstrate
publicly the assembly of guns from piles of standard parts (Morgan 1986); the
American Springfield Armory is generally reckoned to be the first factory to
achieve such production on a large scale (Chandler 1977). In Britain there was
evidence of specialization in the organization of Boulton and Watt (the steam
engine manufacturers) as early as 1830 (Hollway 1991).
Specialization subsequently spread to other industries, but the development
did not really accelerate until after 1850, developing into what was known
through the latter part of that century as “the American System of
Manufactures” (Pine 1993). This approach to work organization requires much
more emphasis on coordination and control than the craft shop approach and its
derivatives. As Koolhaas remarks (1982), it entails splitting all the problems of
design and production—an integral part of the craftsman’s work—away from
the worker. Those tasks must now be carried out by specialists on design and
planning, and the workers are only required to carry out their ordained tasks,
which become more and more specialized and narrow as mechanization and
automation progress. This process was accelerated when industry made the
transition to the second generation of mass producing systems, pioneered by
Henry Ford.
The responsibility for coordination is now removed from the workers and
shared between the central planners and the plant supervisors. To ensure that the
throughput at each step in the production process matches the total process, and
that the quality of each worker’s output satisfies the standards required for the
assembly of parts into working products, stringent measures for quality and
production volume become necessary. Specialization in production therefore
also calls for a much more sophisticated approach to information processing and
communication and, consequently, to organization. It cannot be fully realized
within the framework of a Simple Structure, and it is also easy to see that it is
impossible (above a very modest scale) without writing. Indeed, the analysis of
the 18 operations of pin production presented by Adam Smith is a typical
example of the literate mind at work—you would not find this kind of analysis
in an oral society.
Clegg (1990) is talking about the same processes when he points out that the
growth of large organizations with extensive functional specialization
constituted a decisive break with the task-continuous organization. It was no
longer possible for any one person to master all the specialized tasks in the
organization, and the direction and supervision of activities on lower levels had
to be indirect and based more on formal standards—also dependent on writing.
Even if writing provided the necessary tool for handling large amounts of
information and building complex organizations, however, literacy had to be
widespread to be really practical for large-scale administrative purposes. That
was exactly what happened in the eighteenth, and nineteenth centuries, which
saw the development of mass literacy and an unprecedented spread of
knowledge through printed books and newspapers. In addition, the new
communication technologies—the rapid physical transport, the telegraph, and
the telephone—made it possible to build not only complex and large
organizations, but organizations that spanned great geographical distances.
Within a short period of time, then, many of the traditional constraints of
human physiology were considerably amended by new tools. The simultaneous
changes in the preconditions for production and organization combined to open
vast new territories for human industriousness and ingenuity, and development
finally surged ahead. It changed forever not only the commercial sector of
society but also the political one, at least in Europe—the fast-growing European
bourgeoisie did not in the long run accept the political monopoly held by the
king and the land-owning nobles.
The Transition to a New Organizational Form

Nowhere did the new developments take stronger hold than in the United States.
This is reflected in the rapid growth of the mass-producing American industry
and in the fact that both the most influential theoretician of functional
specialization and the man who applied it most successfully were Americans.
The man who developed this line of thought to its natural conclusion was
Frederick W. Taylor, and the man with the greatest practical success was Henry
Ford—who, after the introduction of the moving assembly line, managed to
manufacture cars at close to half the cost of his nearest competitor, all while
paying his workers the highest wages in the industry and getting immensely rich
himself (Chandler 1977). When he introduced the assembly line in his Highland
Park plant, the amount of labor expended to make a car dropped from 12 hours,
8 minutes to 2 hours, 35 minutes. Six months later it was down to 1 hour, 33
minutes. This breakthrough inaugurated the transit from the first generation of
mass production systems, such as the American System of Manufactures, to the
second—the Fordist systems (Pine 1993). The first-generation systems still
incorporated a lot of the qualities of craft production, and primarely achieved
higher productivity by moderate specialization backed by tools that augmented
the workers’ efforts. They thereby retained much of the flexibility of craft
production and could turn out quite varied products
with small retooling costs. The second-generation systems developed

specialization further, increased the dependence on machines, and introduced
automation to a much greater degree. Productivity increased dramatically, but
flexibility was correspondingly reduced and the cost of retooling for new
products increased.
As both corporations and public institutions grew to staggering new
dimensions, the necessary administrative workload also grew, and the sheer
volume of it became much more than one or a few individuals working in a
fairly unstructured manner could handle. According to Chandler (1977), the
need for extensive administrations first arose in the rapidly expanding railroad
companies, where the administrative tasks grew with each new line, each new
car, and each new locomotive.
The men who faced the challenge of establishing the first major, private
administrative apparatuses had few models to learn from. Their major source of
inspiration must have been the fantastically successful new methods for material
production—whether it was in mass production of industrial goods or in the
large construction works of the time. These new leaders were all civil engineers,
and there is reason to believe that they (as most people will do) tackled the new
problems with methods from their existing repertory, rather than from, for
instance, the military model (Chandler 1977). There is no doubt, however, that
at least some of them were acquainted with military organization, since the
military academy educated some of the best civil engineers. Morgan (1986)
thinks the military did indeed provide an important model for organization; he
gives special mention to Frederick the Great’s reorganization of the Prussian
army in the middle of the eighteenth century—which preceded the great railroad
and manufacturing companies by about 100 years. But Frederick the Great and
the first large-scale industrialists may have had a common source of inspiration:
Frederick the Great was especially fascinated by automatons and mechanical
toys, and through elaborate drills, increased specialization of tasks, and
standardization of weapons and uniforms, he wanted to shape his soldiers into
the human equivalents of mechanical toy men (Morgan 1986). Behavior and
equipment should be standardized to allow easy replacement and
interchangeability in war, and the men should learn to fear their officers more
than the enemy. As an eigtheenth century Danish-Norwegian regulation for
officers stipulated for the attack: “In the rear follows the non-commissioned
officer, with drawn sabre, driving his men forward with blows and harsh
words.”
But even if Frederick the Great increased specialization in his army, a
military force at that time did not have very extensive specialization. Planning
capacity was also limited and information economy was still very important.
We should not be confused by the fact that there were great numbers of soldiers,
or that the hierarchy of command was very
elaborate, because when they battled, they all did more or less the same in
parallel. Rather than being a model of the emerging industrial corporation, the
military still built on feudal roots and was in many ways more like a massive
replica of the preindustrial artisan shops.
The enormous success of the mass-producing factory and of engineering
must have provided both a more immediate and a more powerful model for the
organization of all kinds of activities than military organization. Taylor himself
believed that his principles were equally valid for clerical work, and he was not
the only one. Henry Ford, for instance, also believed in the general applicability
of the principles of mass manufacturing. It was no wonder, then, that the
functionally specialized, procedural work model was adapted even for routine
administrative work and clerical production from the very beginning. The trend
toward imitation of the factory grew through the 1920s and 1930s, as new office
buildings were constructed with the explicit purpose of facilitating the flow of
paper among office workers, who were no longer given separate offices, but
housed in factory-like halls (Sundstrom 1986). Some offices even used
conveyor belts to carry papers from one operation to the next!
At about the same time, Max Weber delivered another strong impetus in this
direction through his analysis of bureaucracy and his deep conviction that it
represented the ultimate in rational information processing. And the crucial tool
was writing—to Weber, it was the key that made everything else possible,
coupled with the superior skill developed by the well-educated, specialist
bureaucrats: Specialization and hierarchical supervision (documents could be
passed on), impartiality (decisions could be audited, and they could be contested
and appealed to higher authorities) and the application of rules and regulations
(they could be written down in unequivocal form).
However, writing was not only a tool for increasing efficiency and
impartiality, it had its greatest potential as a tool for managing complex work
and large organizations. It was the information storage and communication
capacities becoming available in the wake of the growing literacy that opened
for the decisive transition from the Simple Structure to the archetypal large
organization of the modern era—the Machine Bureaucracy.
Writing also led to a depersonalization of coordination. In nonliterate society,
complexity had to be kept to a minimum, and authority relationships were
strictly personal—building on recognized power relationships between
particular individuals. Enlarging a Simple Structure necessitated extensive
delegation of power from one individual to another, bound together by personal
loyalties and dependencies—often reinforced by family ties. Information flows
had to be kept to a minimum.
In the new paper-based, functionally specialized organizations, a large part of

the practical coordination effort was shifted away from the direct, personal
relationships of nonliterate organizations toward files, written plans,
instructions, rules, and regulations. One still had real persons as superiors, of
course; their presence was undoubtedly very important, and in some parts of the
organization (notably in the bottom layer of manufacturing or construction
organizations) authority and coordination would still be very personal. But, as
complexity and organization size increased, the growing and strongly regulated
flows of work, control information and staff information (Mintzberg 1979)
carried a larger and larger part of day-to-day coordination. For many employees,
the human face of direct, personal authority was to a great extent replaced by
the rule of written plans and regulations.
The principle of functional specialization that was the hallmark of these new
organizations was further reinforced by the nature of paper-based information
storage itself. A filing cabinet—perhaps the single most important element of
pre-computer administrative technology—had the very constraining property of
being accessible in only one physical location, and if a person did not work on
the premises, there would be a number of tasks that he or she could not easily
carry out. The account of a bank customer, for instance, could only be read or
updated in the main office or (if the bank had decentralized account
administration) in the branch office that kept the account.
Another constraint of paper-based files is that they normally have only one
index—if you have a file of persons, for instance, you must choose whether you
want to organize it by name, by date of birth, or by address (to mention the three
most common keys). Cross-referencing paper-based files is extremely time-
consuming, and really only viable for historical (and thereby unchanging)
information. When paper-based files grow really large, their monodimensional
nature tends to favor a procedural, specialized mode of administration. Files
kept in the form of punch cards and processed with the help of card counters
and sorters were of course more flexible, but not sufficiently so as to create a
fundamentally different situation. These practices of organization and
information processing formed in the nineteenth and early twentieth century still
have a very strong influence on the way we work.
The Limits of Monolithic Bureaucracy

As they grew, the functionally specialized organizations became increasingly
unwieldy, since they did not have the mutual coordination of the total work flow
that is inherent in the holistic approach. Coordination had to be handled through
planning, written communication, and hierarchic
management—tier upon tier of managers, finally converging in the president’s

office. The larger the organizations became, the more energy had to be devoted
to the coordination of their various functional departments.
Of course, the formal structure would always be somewhat alleviated by
informal horizontal links facilitating daily operations, but functionally
specialized organizations are nevertheless inherently difficult to manage—they
require large management resources and are slow in responding to changes in
the environment. As Williamson noted (1975), reports (upward) and instructions
(downward) are liable to interpretation at each organizational level, and
therefore tend to become more inaccurate for each level they pass through. If
there are too many levels in the hierarchy, this “control loss,” as Williamson
terms it, can isolate top management from reality. When a certain size is
reached, such organizations simply threaten to atrophy.
The most monumental example of the failure of large-scale functional
specialization is perhaps the collapse of the communist economies in eastern
Europe. It was not only the absence of competition that made those societies
rust out, it was also the serious breakdown of coordination that was a
consequence of the attempt to organize whole societies as monolithic,
functionally specialized corporations. In the former Soviet Union, for instance, a
country with about 290 million inhabitants, there were numerous examples of
important product classes where production had been allocated to one large,
specialized facility only.
One is reminded by this of the fact that Frederick W. Taylor’s ideas were
well received by the Bolsheviks in the young Soviet state (Morgan 1986).
According to Clegg (1990), they were in fact introduced by Lenin himself.
Braverman also notes the enthusiasm and quotes (1974 p. 12) Lenin as saying
that “We must organise in Russia the study and teaching of the Taylor system
and systematically try it out and adapt it to our ends.” 2 It is quite evident that
Taylor’s extreme emphasis on planning, control and rational behavior
corresponded very well with central Marxist-Leninist dogmas as they were
practiced in the Soviet Union under Lenin and his successors. It is in fact
tempting to suggest that the Soviet Union in many ways represented a
monstrous attempt to create the largest Tayloristic factory organization ever.
What we can learn from this experiment is that monolithic, functionally
specialized organizations do not scale well—they may work quite perfectly up
to a limit, but then gradually crumble under the sheer weight of the required
coordination. The total work flow of a whole society is orders of
2Braverman’s reference is V. I. Lenin, “The Immediate Tasks of the Soviet Government” (1918),
Collected Works, vol. 27 (Moscow, 1965), p. 259.
magnitude too great to be coordinated within a single hierarchy—it is indeed

much to complex to be deliberately coordinated at all. All the successful
economies of the modern world have in common a large private sector evolving
according to principles resembling Darwin’s “survival of the fittest.” As
organization ecologists have pointed out (see for instance Hannan and Freeman
1977 and 1984), there are many parallels between an open economy with
independent actors and nature’s ecological systems—today’s free-market
economies have achieved effective large-scale mutual adjustment (increasingly
at a global level) only through each unit’s independent exploitation of its
immediate environment.
There is of course considerable disagreement about how unfettered the actors
in such an economy should be, and judging from the relative successes of, for
instance, the American, Japanese, German, and Scandinavian approaches (both
in economic and social terms), the optimum degrees of freedom in the economy
are by no means obvious—but few people dispute the basic soundness of the
principles.
It is also interesting to view this in the perspective of information economy.
In the free-market economy, the complexities of operation are encapsulated
within independent companies, minimizing the amount of information that has
to cross organization boundaries. The resulting simple interface to the world
(mainly product properties and prices) makes organizations interchangeable and
permits the competition and dynamic, continuous adjustments that are the
hallmarks of an open economy.
But let us return to the corporate dimension. As we just noted, the growing
organizations of the early twentieth century were coming up against the limits
of the available administrative solutions. Remedies had to be found. In the
corporate world the honor for creating the major new model is most often
bestowed upon Alfred P. Sloan Jr, one of the managers of General Motors
during its turnaround in the early 1920s (Chandler 1977, Williamson 1985).
Pierre du Pont took over a controlling position in GM after the company’s near
bankruptcy during the collapse of the automobile market in September 1920
(Chandler 1977), and he brought Sloan in to help him with the cleanup.
They quickly realized that the sprawling empire of companies assembled by
William C. Durant needed much closer attention than Durant had given it.
However, they decided against creating a centralized company organized in a
single tier of functional departments. The company’s activities were, in
Chandler’s words (Chandler 1977, p. 460), “too large, too numerous, too varied,
and too scattered to be so controlled.” Such a configuration would also have
swamped the top managers with daily administrative tasks and prevented them
from devoting their time and energy to the tasks du Pont saw as the most
important ones for top management: strategic planning and business

development.
Du Pont and Sloan’s solution was to establish within the company
autonomous operating units, called divisions. Each division was given the
responsibility for a particular price bracket, and was given complete control
over all the resources and functions necessary to manufacture and sell its own
cars. The total work flow of General Motors was thus divided into several self-
contained work flow domains. Separate financial and advisory staffs kept close
tabs on the development in the line divisions, constantly reviewing their
performance according to plans and forecasts and continually revising budgets
and forecasts not only according to past performance, but also with an eye on
the national income, the state of the business cycle, seasonal variations, and the
expected market share for each line of products.
Top management in General Motors was consequently relieved from its
position as the crowning apex of day-to-day administrative chores, and could
concentrate on long-term development. With fewer levels of coordination, the
divisions also became more nimble actors in their respective markets than GM
could have been if managed as one integrated company. This represents the
Divisionalized Form in Mintzberg’s classification.
In our perspective, however, divisionalization was not an innovation at all. It
simply represented a recourse to the fundamental administrative principle of
feudal type societies—simplification by encapsulation of complexity—and for
the same reason: to achieve the information economy necessary to manage
within the constraints of the available administrative technology. From their
vantage point, Sloan and du Pont converted the organization from a vast array
of tiered departments into a small number of operating and staff divisions,
which they controlled chiefly through sales targets, budgets, and profit rates,
just as the feudal king used tributes and quotas for military contribution as his
main instruments of control. We may say that Sloan and du Pont simplified GM
by encapsulating the complexity of car manufacture within the divisions,
“hiding” it from their view as general managers, thus reducing the information
flow between the divisions and company headquarters to a trickle. Each division
was of course still a complex, hierarchical, procedural organization full of
functionally specialized departments. But, being much smaller than GM as a
whole, and with simpler objectives, the divisions were easier to manage.
In this connection, it is also interesting to note that the railroad company
generally considered to be the best run in the United States in the last part of the
nineteenth century—the Pennsylvania Railroad—deviated markedly from the
monolithic organization of its competitors. Even more
of a structural parallel to a feudal type state, it was organized into five self-
contained, geographically delimited units (Chandler 1977). They were in their
turn composed of smaller geographical units with a great degree of
independence in operations, but with the same kind of tight, central control of
key performance parameters later developed at GM. The central management
and staff handled external strategies for expansion and relations with connecting
roads; they determined and supervised technical standards, and closely
monitored the financial performance of the different units. There was also a
centralized purchasing department.
So the final verdict may not only be that divisionalization was no more than a
revival of one of our oldest administrative techniques, but also that it was
Pennsylvania’s president from 1852 to 1874, Edgar J. Thomson, and not Alfred
P. Sloan Jr, who should be awarded the honor for reinventing it as an
organizational tool for large corporations.
The principle of encapsulation can become the basis for modifications at
lower levels in the organization as well. Instead of a functionally specialized
organization covering several markets and/or delivering many products or
services, one can for instance organize a department or organizational unit
responsible for a specific market or a product/service, or even for a certain
product in a certain market. This ensures increased responsiveness to the
environment, by reducing the number of organizational layers that need to be
activated to arrive at decisions concerning product strategies, customer service,
or manufacturing methods. 3
Very few organizations are consequent in following only one pattern,
however. Different structures will often be applied at different levels, and, even
within one main level, there may be numerous exceptions—usually as a result
of ad hoc responses to pressing challenges from the environment. A bank
handling loan applications by passing them along through a functionally
specialized organization, for instance, might react to a sudden surge in
application volume or increased competition by forming a separate loan
department to process applications faster and more efficiently.
As already indicated, the development just described was both strongest and
most consequent in the United States. But not even there did the organizational
forms portrayed here—in Mintzberg’s terminology the
3While facilitating coordination and responsiveness, market-based, or product-based organizations

may suffer from disadvantages when it comes to economy of scale and sustaining necessary
expertise, however. Creating several independent production units requires the duplication of many
functions, which can lead to higher overall costs. If the decentralized specialist groups become too
small, specialists may find them less attractive places to work, since they think a certain number of
like-minded colleagues is a prerequisite for maintaining and developing their skills and knowledge.
Machine Bureaucracy and the Divisionalized Form—replace all others. Even as

late as in the 1990s, 90% of the six million businesses in the United States had
less than 20 employees, and it goes without saying that they were not organized
like Machine Bureaucracies. The situation is just the same in other
industrialized countries. In Norway, for instance, about 80% of even the
industrial firms have less than 20 employees. If we include companies between
20 and 50 employees, the figure rises to 90%, employing close to 35% of the
total workforce in Norwegian industry.
Consequently, the older organizational forms, such as the Simple Structure
and the Adhocracy, are obviously thriving. The Simple Structure is, among
other things, still a natural form for the small shop and for an entrepreneurial
startup driven by one individual’s vision. The Adhocracy has survived as the
preferred structure for many knowledge-based companies (such as consultancy
and law firms) and organizations dominated by research and development—
even organizations so large that they would otherwise be candidates for
bureaucratic organization.
For the organization of really small companies, the technological innovations
of the Industrial Revolution have not meant too much, since specialization is
generally limited and coordination anyway depends on close, informal contact.
Moreover, most small firms have local markets and relatively simple logistics.
The dramatic effect for craft-type firms came first and foremost from the
competition many of them suddenly faced from standardized, mass-produced
goods marketed on a national or even international scale. That was the change
that drove scores of them out of business.
The new preconditions, however, as we have just seen, made it possible for
entrepreneuring people to build much larger and more complex organizations
than before. For those larger companies, which tried much harder to routinize
tasks, the effects of functional specialization and the limitations of available
administrative technology combined to make the Machine Bureaucracy the
dominating organizational structure and divisionalization the main remedy for
handling complexity too great for a monolithic structure.
A New Concept for Coordination

The Bureaucratic Advantage
We have already concluded that the bureaucratic organization could both grow
larger and operate more efficiently than earlier organizational forms, and we
have also said something about the reasons for this. To fully understand the
nature of the change, however, it is necessary to take a closer look. And it is all
the more worthwhile to do so, since the
development of the bureaucratic organization also contains the seeds of a new

intellectual tool—the explicit, conceptual model—that will not fully come into
its own until our use of computers matures in the twenty-first century.
As noted earlier, Weber’s main explanation for the bureaucracy’s
effectiveness was the superior skill of its clerks. They are well educated and
highly specialized, and they continuously polish their proficiency through their
work, following the guidelines laid down by their superiors. He compares the
bureaucracy to a machine—it functions in much the same way as a modern
factory producing goods in a very efficient, partly automatic manner—and
attributes its efficiency to the increased productivity and quality at each step in
the production process (see, for instance, Weber 1968, pp. 973-75). If we
generalize this argument for both manufacturing and clerical organizations, we
may say that it is specialization, the concomitant superior skills of the
employees and their improved tools that do the trick. And the arsenal of tools
includes not only the “hard” tools and machinery of material production, but
also office implements such as files—extremely important through their
capacity for implicit coordination.
This is not a sufficient explanation, however. Specialization and improved
skills may indeed increase quality and efficiency at each step in the process, but
there is still the challenge of coordinating the work of the multiplying ranks of
specialists—making it possible to build and run large organizations while
preserving the advantages achieved for each individual task.
A more comprehensive explanation is provided by March and Simon (1958).
They recognize two basic methods for the coordination of large organizations
with high internal interdependence among tasks—that is, organizations with a
high degree of internal specialization, requiring careful and extensive
coordination to operate efficiently. The first one is coordination by plan, which
is based on preestablished schedules, the second coordination by feedback,
which relies on continuous transmission of information about the workings of
the different parts of the organization.
Coordination by feedback requires open lines and fairly intensive
communication between the coordinator and the coordinated. It corresponds to
(and encompass) Mintzberg’s direct supervision and mutual adjustment, which
began as the two basic (intuitive) coordination methods used in small-scale, oral
societies. However, the heavy communication load of coordination by feedback
becomes a severe penalty when the organization grows. Relying on coordination
by feedback alone, the effort needed to coordinate an organization will grow
much faster than the organization itself, and, without some kind of
simplification scheme, an
organization would not have to become very large before coordination would
break down and confusion reign.
There are two main ways to solve this problem. One is to abolish the need for
coordination as far as possible, which was the essence of the encapsulation of
complexity inherent in the feudal system. This “evasion tactic” was the only
available method in preliterate societies, and it was later revived in the form of
divisionalization. It can only work, however, when there is no need to
coordinate persons or processes across different subunits.
The other solution is what March and Simon (1958) terms coordination by
plan (termed coordination by program in Figure 3-1 on page 51), which
requires much less communication and thus emerges as strikingly more efficient
(March and Simon, 1958, p. 162):
As we noted earlier, it is possible under some conditions to reduce the volume of

communication required from day to day by substituting coordination by plan for
coordination by feedback. By virtue of this substitution, organizations can tolerate
very complex interrelations among their component parts in the performance of
repetitive activities. The coordination of parts is incorporated in the program when
it is established, and the need for continuing communication is correspondingly
reduced. Each specific situation, as it arises, is largely covered by the standard
operating procedure.
The efficiency of the bureaucracy, then, both in its blue collar and white collar
versions, is also to a large degree based on the fact that work is standardized and
the coordination of work is preprogrammed. The various tasks are first analyzed
in considerable detail, and prescriptions for carrying out work and solving the
most common problems are specified. Once they have learnt those
prescriptions, the workers and clerks are able to execute most of their work
without further instructions.
We can clearly see how this is a direct continuation of central principles
behind coordination in the Simple Structure (discussed in Chapter 5). The focus
is exactly the same: both in the Machine Bureaucracy and the Simple Structure
the necessary coordination is achieved through directing work, as opposed to
the information sharing of the Adhocracy. But whereas the Simple Structure
relies on direct supervision for its coordination, the Machine Bureaucracy relies
on indirect supervision: The role of the physical supervisor is assumed by the
standardized work rules (the program). Mintzberg (1979) clearly builds on the
passage from March and Simon quoted above when he describes the
coordinating mechanism of the Machine Bureaucracy—in fact, he quotes briefly
from it himself (Mintzberg 1979, p. 5)—and his standardization of work
processes is roughly equivalent to March and Simon’s coordination by plan.
The ultimate in preprogrammed work is of course the automatic machine,
which represents a carefully designed program forged in steel,
repeating its designed actions again and again without further human guidance.
It is also worthwhile to note that the impressive efficiency of the automatic
production line does not only reside in the speed of each particular operation,
but just as much in the perfect, automatic coordination of those operations.
We can then extend the taxonomy of coordinating mechanisms presented in
Chapter 3 with two new variants, both dependent on technology (Figure 7-1).
The first one I propose to call explicit routine, the other simply automation. The
explicit routine is the “program” you end up with when you consciously design
an organization. In larger organizations it will usually contain both organization
charts, overall process descriptions, and job descriptions. It will normally be
based on a planning process involving at least a basic level of explicit modeling
and design.
In addition to these two extensions, the era of organizational tools also
opened up for a development of mutual adjustment, where considerable
extension is possible if the adjustment is mediated not by direct information
exchange, but by indirect communication through a common information
repository. To be practical, this common repository needs technology for
externalizing memory: Although it is conceivable that implicit coordination
could be used with a person’s memory as repository, it is hard to see how it
could be of real importance. The written record, however, created exactly the
kind of information repository needed. Records kept together in a file made it
possible for many persons to base their work and decisions on the same
information, and changes introduced by one would apply for the work of all the
others, without

Tacit Explicit
Skills Skills
Technology Dependence
Implicit Explicit Automation

Coordination Routines
Figure 7-1: Taxonomy of coordinating mechanisms extended by pre-

computer technology.
any need for meetings or other forms of personal communication. This made
possible what I prefer to term an implicit coordination of the people who
worked with it: a coordination that is an implicit, automatic effect of working
from a common base of information, eliminating the need for extra
communication and direct supervising efforts.
At least within a narrow area, then, a fairly large number of people can
coordinate at least part of their activities by working with the same files. By
relying on this continuously updated, common kernel of information, they
directly modify each other’s actions by mutual adjustment without ever meeting
face to face, extending the use of this mode of coordination to a significantly
larger number of people than was earlier possible. For a bank clerk, for instance,
it is not necessary to notify all the other clerks about a change in a customer’s
account balance; it is only necessary to post it in the books.
Although implicit coordination works by effecting mutual adjustment
through common information, it is sufficiently differentiated to be considered a
separate coordination mechanism. As such, it became the first technology-based
coordination mechanism, and thus marks a watershed in organizational history.
Its use, however, was limited before the development of large private and public
organizations in the nineteenth century.
What about direct supervision, then—did it, too, receive a boost from the
new technologies? Some will perhaps argue that the developments of
communications technology in the nineteenth century, especially the telegraph
and telephone, have indeed made a difference and allowed direct supervision to
be used over much greater geographical distances. This is of course true: The
geographical reach of direct supervision was greatly extended by pre-computer
communication technology, and larger organizations spanning greater distances
could be kept under close control. The space of constructible organizations was
thus extended, but I will maintain that this did not imply the creation of any new
coordinating mechanisms: It was still one person giving orders to others. While
implicit coordination represents something genuinely new (coordination not via
direct communication, but indirectly via a common, external information
repository), direct supervision via telephone or letter represents little more than
an amplification of the principal’s voice.
I have not included here any extensions to the two mechanisms subsumed
under standardization of skills. Tacit skills are by definition unchanged in their
nature, although modern mass media have greatly expanded the sources for the
kind of information that contributes to the formation of tacit skills. This is
precisely the reason for many parents' dislike of certain kinds of TV-programs,
films and records: that they may tend to instil unwanted norms and tacit skills in
their children. The teaching
of explicit skills has of course been strongly enhanced by the development of

literacy, but although the textbook certainly made it easier to teach the same
skills to many people, it did not change the mechanism per se: The way
coordination is effected is not related to the medium for the original knowledge
transfer.
Explicit Design and The Emergence of the Conceptual Model

Both methods of coordination by standardization (of work and of skills) have
old roots and have been used in nonliterate societies, for example, in craft and
trade. In their old versions, however, these coordination methods were largely
implicit in tradition and customary ways of working and organizing. The
circumstances of their use did not involve formal planning or written
documentation. Consequently, they did not contain explicitly designed work
programs; rather, they grew out of customary practice and were transferred from
generation to generation as part of the continuation of a craft, a trade, or the
social order.
As we have already seen, however, there is a definite limit to the level of
complexity that the unaided human mind can handle. The new organizations,
with their elaborate interdependencies, were far too complex to be conceived
and run within the framework of an oral tradition, and by unaided memory
alone. Both their manufacturing and clerical parts required detailed and explicit
analyses of the operational requirements and the interdependencies of the
different steps and levels in the process, to be followed by careful and detailed
design and planning of operations. Writing was an indispensable tool for this
work, as well as for the design and construction of the new tools and machinery
that were so important for the new developments.
Following Ong (1982) and Havelock (1986), there is also good reason to
believe that a mature literate tradition, a developed literate mindset, was a
necessary prerequisite to this new analytical approach to work and organization.
People from oral cultures seem to have trouble using and manipulating the
symbols and abstract categories used for complex analysis and planning. The
oral mindset is concrete and person oriented, and it correlates with the basic
organizational structures (relying on personal authority) that we explored in
Chapter 5. The literate mindset is abstract and role oriented and correlates well
with bureaucracy, where work is specialized and authority is tied to positions,
not particular persons.
In the modern organization, then, work is no longer organized in accordance
with custom and tradition, but according to a conscious design based on an
explicit analysis of the desired outcome and the available
means. In my view, this represents the decisive break with the past and marks
the transition to a new paradigm for the organization of human work. The old
paradigm, developed in an oral world, was characterized by a reliance on
tradition, tacit knowledge (Polanyi 1967) and theories-in-use (Argyris 1980),
and was focused on personal relationships, family ties and holism in work. The
new one, born in the first fully literate societies, builds on conscious analysis
and explicit design and focuses on the coordination of interdependent,
specialized tasks. It will almost gleefully break with tradition when that is
instrumental to improvements in effectiveness and efficiency.
If we look at this from the perspective of action-oriented organizational
theory (as discussed in Chapter 2), we can perhaps express it more clearly.
Within the oral paradigm, the organizational patterns of action more or less
emerged. Because they were based on tacit knowledge and theories-in-use, they
were seen as an inevitable part of the social fabric, as a part of the natural order
of things. However, the explication of design that forms the basis of the
Machine Bureaucracy meant that patterns of action were consciously
constructed for a specific purpose, partly separated from the larger society and
constantly open to inspection and improvement. The road to greater efficiency
seemed always to go through greater sophistication and thoroughness in design.
Indeed, the most efficient production is achieved through automation, which
builds squarely on a total, conscious design of recurring patterns of action, or—
to comply with the tradition of reserving the term “action” for human activity
(Silverman 1970)—recurring patterns of machine movements.
The Machine Bureaucracy, as we have described it, then, is a production and
coordination “machine” with a specific purpose—it is consciously designed to
accomplish certain tasks or solve certain problems. It is designed on the basis of
a detailed analysis of a set of purposes, tasks, and relevant environmental
factors that its creators deem relevant to its success. This set we may call its
problem domain. The designers need to have not only a knowledge of the
features and events in this problem domain, but also a set of postulates—a
theory—about how those important features and events relate to each other. To
use a systems term, we may say that the design is based on an at least partly
explicit conceptual model of the problem domain.
What is a conceptual model? It is a representation of a part of reality, just like
the physical models we use to comprehend and test the behavior of complex
artifacts and phenomenons—by, for instance, placing an airplane model in a
wind tunnel. Conceptual models are used for the same reason: to establish an
understanding of reality that is sufficient for initiating sensible, effective actions
toward the part of reality represented by the model. The only difference is that
the representation of reality is
conceptual, not physical—it consists of words and drawings on paper, on a

screen or in the mind. An organization chart, for instance, is a simple conceptual
model of an organization, representing its formal decision-making hierarchy.
In general, we can define a conceptual model as a conceptual representation
of a limited (bounded) part of reality, that part which we are interested in for our
particular purpose. The model is of course a simplification of reality and will
most often only describe the features that the designers judge to be sufficiently
important, and not self-evident. It is, however, crucial that all the aspects of
reality that are important for its purpose are represented in the model—objects,
phenomenons, the relations between them, and their static and dynamic
properties. The organization designers can then use this model—which
represents their best understanding of the problem domain—to work out the
details of the division of work, organizational structure, task formation, work
instructions, and the like.
I do not mean to say that the pioneers in organizational design used the term
“model” or were aware of the concept of modeling as a tool. Even today, the use
of this term in connection with organizations and organizational design is by
and large limited to the fraternity of systems analysts and designers. However,
even if the people who work out the designs and task structures of organizations
do not use the term and are unaware of the concept, the descriptions and plans
that form the basis for their designs are conceptual models nonetheless.
We may well say that even the traditional organizations of oral cultures built
on conceptual models of their problem domains. But those models were not
made explicit—they existed only as theories-in-use (Argyris 1980). They were
therefore not open to inspection or conscious elaboration and could not serve as
a basis for innovative design. To develop and design the modern organization—
with its intricate interdependencies and its new approach to coordination, with
its drastically reduced control and communication needs—an explicit model
was needed: A model developed on the basis of a conscious analysis of the
problem domain and documented on paper.
In contrast with models implicit in theories-in-use, the new, explicit model
was wide open to inspection and improvement and could therefore support the
steady improvements in operational planning, automation, and work procedures
that characterize the modern organization. It could also be used to establish a
necessary minimum of consensus throughout the organization regarding
important goals and operating principles, another condition for making large
organizations work. It was thus the combined development of programmed
coordination and explixit modeling that constituted the foundation of the new
organizational
paradigm, which we may simply call the literate paradigm—after the

technology that made it possible.
The Constraints of Standardization

Compared to older organizational forms, the Machine Bureaucracy offers great
economy in coordination and information transfer and is very efficient in
turning standardized inputs into standardized outputs. However, the penalty is
inflexibility: it can only handle inputs and deliver outcomes that are defined in
the underlying conceptual model. Because the actual formulation of both the
model and the standardization rules necessitates analyses and planning
procedures that are exceedingly laborious, changes in the model and the
“machine” are very expensive and take a long time to accomplish.
Therefore, Machine Bureaucracies are slow in adjusting their behavior to
changes in their problem domain. In very dynamic environments, they simply
cannot keep up. That is why one has to revert to more flexible coordinative
schemes for highly unstable problem domains: if the environment is simple,
direct supervision will be close at hand; if it is complex and especially if it
requires knowledge in multiple fields, mutual adjustment will be preferred, at
least at the most crucial levels. In war, for instance, the chiefs of the different
services will work very closely together, and during large and important
combined operations they will usually spend most of their time in the same
room, conferring continuously while adjusting the actions of their respective
services in real time. (That is what war rooms are all about.)
What this means is really that to use standardization of work as a
coordination method, and to reap the great rewards it offers, it is necessary to
have models and a modeling capacity that can keep up with changes in the
environment and operating conditions. In other words, there is a need for a
professional staff to analyze change requirements; design the new routines,
rules, and/or machine combinations; and then manage their implementation. The
great savings offered by automation and the new organizational form thus also
carries a penalty, which is greater cost for maintenance and change.
If the cost of updating the model becomes too great, or if there is a need for
change too often, the organization is forced to adapt a coordinating mechanism
that needs less analysis and planning but craves more resources in daily
operation. There are very real tradeoffs to be made here, and organizations that
straddle the crossover point or experience significant changes in the dynamics
of their environment very often run into huge problems trying to adapt their
structures and coordination methods to the new realities of life.
Culture Revisited
In Chapter 2 I concluded that the influence of social and cultural factors on the
local constructible spaces was too varied to incorporate into my analyses of the
enabling powers of technology. I still hold that they are—but the natural
question then is if this variation represents a problem for the validity my
analysis.
The rather neat account of the growth of modern organizations given in this
chapter is to a large extent in accordance with Chandler’s (1977) interpretation.
Chandler explains the rise of the modern corporation (and thereby the Machine
Bureaucracy) by the driving forces of the technological developments of
industry and transport and the concomitant development of a national mass
market for industrial products in the United States. Companies grew because
their internal coordination was more efficient than the market-based
coordination of small, independent firms, and because the larger firms had much
more power for market penetration and domination. It was the large,
homogenous national market in the United States, Chandler argues, that caused
the large, multiunit firm to flourish in the United States before it became a
decisive factor in European business. He acknowledges that legal and cultural
differences also had a role in delaying the development in other countries, but
he does not doubt that the American experience will sooner or later repeat itself
everywhere else, when the local economies reach the proper development stage.
This is a view shared by Williamson (1975, 1985), who argues forcefully that
the logic of transaction costs—the cost of exchanging goods or services
between people and across organization boundaries—eventually will prevail,
and foster similar organizational solutions everywhere.
However, many other scholars have pointed to the fact that the Industrial
Revolution originated in what is now loosely referred to as the Western
industrialized world, and the first large, private organizations were accordingly
products of those societies. Many of the organizational traits we used to take for
general principles may thus be no more than artifacts of our own particular
culture. Chandler and Williamson, who concentrate most of their arguments
around developments in the United States, may easily be unduly influenced by
American peculiarities.
Does this criticism, then, undermine the validity of the analysis we have
made in this chapter about the ways technology has extended the space of
constructible organizations?
Important instances of the criticism against the convergence theories of
Chandler and Williamson are Granovetter (1985), Hamilton and Biggart (1988),
and Clegg (1990). Granovetter discusses the tendency of theories of economic
action to offer explanations that are either undersocialized (fail to
take into account social factors, as the classical and neoclassical “rational man”
theories of economics) or oversocialized (picture humans as more or less
passively succumbing to the prevailing social forces, as in orthodox Marxism,
where class properties and class distinctions take preeminence over individual
characteristics). For instance, while Granovetter acknowledges that
Williamson’s focus on institutional and transactional considerations differs from
neoclassical theories, he maintains that Williamson’s theories are still clearly on
the undersocialized side, paying too little attention to sociological, historical,
and legal arguments.
As Granovetter points out, undersocialized and oversocialized theories
ironically both end up by robbing us of most of our humanity and discretion—
the undersocialized theories by making us mere slaves of a rather narrow logic,
the oversocialized by making us robots programmed by our environment to
merely enact prevailing norms. Granovetter argues that economic action is
thoroughly embedded in both the actor’s social environment and his or her
personal values and goals. Rationality cannot be judged on the basis of narrow
slices of a person’s life only. Organizational politics may for instance make it
subjectively rational for an individual to behave in ways that are economically
irrational for the firm. And history and traditional authority structures may
heavily influence company organization without determining it: Many aspects
of organization may be imported from abroad, or may even come as a result of
the idiosyncrasies of powerful organization members.
Hamilton and Biggart (1988) attack both the market-based theories of
Williamson and Chandler and the theories that explain national differences in
organizational structure on the basis of culture. To test different approaches,
they look at firm structures in three successful countries in East Asia: Japan,
South Korea, and Taiwan. If economic factors alone determined organizational
structure and practice, then Asian enterprises would be quite similar to Western
(especially American) enterprises, and they would not show great internal
differences. Similarly, since the three countries are strongly related culturally
(all of them drawing heavily on Chinese culture and tradition and having
intertwined histories), their organizations should not differ to much among
themselves, even if they differ from Western firms. Economic organization,
however, differ markedly in the three countries, and all of them are different
from the United States and Europe. To explain these substantial differences in
organization, Hamilton and Biggart conclude that the preexisting authority
structure (which was distinctly dissimilar in the three Asian countries) is the key
variable.
The arguments presented by Granovetter as well as Hamilton and Biggart
agrees with that of Clegg (1990)—and Clegg also draws on their work. His key
issue is precisely the paramount importance of embeddedness, institutional
frameworks, and modes of rationality, and his criticism
of Williamson (and a number of others, including Chandler) goes along the

same lines. Drawing on a broad selection of empirical material, he shows both
how structures differ between cultures and within them, and how they may vary
considerably even within countries, markets, or enclaves that are homogenous
in most other aspects. Clegg concludes that the diversities offered by
contemporary organization forms cannot be interpreted in terms of any single,
decisive factor—be it economy, culture or authority relationships. He mandates
a more complex explanation, encompassing a wide variety of contingencies, and
in this he seems to be much more in tune with the central theme emerging from
the last 30 years of organization research: that there are a large number of
contingency factors which have been shown to influence organization structure.
Summing up the arguments, he says (1990, pp. 162–63),
. . . organization forms are human fabrications which agencies 4 will structure using
whatever discursive rationalities 5 they can secure. These rationalities will vary in
their institutional location, drawing not only from occupational identities, or from
the regulatory framework of law, accounting conventions and so on. In addition,
they will also draw on whatever resources find expression in a particular context,
local resources which are particular for that context.
I believe we can conclude that the structure of modern organizations vary, and
that there are many important factors on different analytical levels that
contribute to that variation. It only serves to underscore one of the basic tenets
of general systems theory: the concepts of equifinality and multifinality. As
Crozier (1964) remarks, when analyzing the reasons for the fact that the French
economy lagged behind the British from the start of the Industrial Revolution
until after the World War II, the most baffling fact is not that some countries in
the industrialized world lag a little behind others, but that the differences are not
greater.
Of course, the examples of the communist countries in Europe, of Argentina
(one of the most developed and economically advanced nations in the world
around 1900) as well as the divergent development trends in Asian nations show
that there are indeed limits to this equifinality—you cannot succeed by any mix
of means—but the fact remains that there are no single prescription for success,
and no single pattern of development and organization that is destined to
percolate through the world and gradually make all organizations and societies
similar. And there is no such thing as technological determinism—a particular
technology or set of
4In Clegg's sense, an agency is an entity that makes thing happen. It can be an organization, a part
of an organization, or an individual.
5Rationality as it appears to the agent under the full (and dynamic) set of circumstances under which
he operates.
technologies does not invariably lead to the same organizational solution. Using
whatever material we humans have found in our environment and in our own
minds, we have together created a panopticon of practical solutions that shows
great leeway for variation.
Nevertheless, the iron constraints of our biology and our tools are still there,
and no member of the human race can operate outside them. Complex
technology and large-scale production of goods and services require
organizations much larger than the artisan shop, and, in this sense, the
technology of the nineteenth and twentieth centuries had a deterministic streak:
to exploit it for the creation of increased material wealth, we had to build large
organizations. Further, all large organizations have to tackle the challenges
inherent in the coordination of large numbers of people. If we look at large
organizations around the world, we will therefore find a lot of common ground.
However different the authority structures in such organizations may be, for
instance, they all have an authority structure. Despite major variations in job
definitions, in the proportion of workers that are skilled, in job rotation schemes
and distribution of authority, all firms above a certain (rather small) size still
have job specialization. And, because of this, they all have planning functions,
coordination needs, and extensive internal communication. Even the light-
footed Taiwanese manufacturing firms, relying more on their ability to adapt
fast to changes than on forecasting future market trends, have to plan their
production at least a month or two ahead.
In oral societies the constraints of human physiology will keep organization
at a fairly simple level, and people will depend to a very large degree on their
own immediate work and actions for their survival and well-being. In a
developed, literate society, on the other hand, one of the key aspects of human
life is the extent to which the citizens as individuals are constantly dependent on
the extensive collective information processing taking place in innumerable
large and small organizations. This information processing is really pervasive.
Even in organizations established to produce material goods, information
processing is often the major activity if measured in work hours. Its level of
complexity is also orders of magnitude greater than the collective processing of
even the most advanced oral societies, and is extended further through the use
of advanced automation.
I discussed this at some length in the preceding chapters and concluded that
the constraints on human communication and memory are the basic problems in
the control and coordination of organizations, and that control and coordination
are the most pressing operational problems of all collective human
undertakings. The subject of organized, collective information processing has
therefore risen to a much more prominent position
in modern societies than in preindustrial civilizations. The preoccupation in both

public and private enterprises in all cultures and societies with subjects such as
formal and informal organization structures, lines of authority, communication
channels, job designs, and management styles, as well as the fervor of the
discussion, is a clear expression of the universally perceived gravity—and the
universal validity—of the problem.
So far, the development of the Divisionalized Form represents our best effort
to harness this complexity within the bounds set by the modified preconditions
that has emerged from the technological development of the last 3500 years. It
is reasonable to assume that the possibilities awarded by these preconditions
have largely been exhausted in this period of time, as millions of attempts to
create and run successful organizations throughout the nations of the world have
employed a great breadth in innovation and angles of attack. To evolve
distinctly new organization schemes, we will therefore need new technological
developments, such as the emerging information technology that is the ultimate
subject of this book. Only they can modify the preconditions further and thereby
enlarge the realm of the possible. It is, however, up to us to explore the new
frontiers—it is people who discover, invent, and act, new developments do not
come about on their own just because they are feasible.
This is in good accordance with the framework of the present study, which
builds on the notion of physical and cognitive preconditions for organization
building, defining the limits of the possible in the organization domain, and how
the development of tools has changed them. Within the limits set by these
preconditions, which amounts to the total space of constructible organizations,
other constraints will also operate—cultural constraints, the traditions of power
arrangements, of markets and competition—to define the many local spaces of
constructible organizations one will find around the world.
Within the spaces defined by these constraints, within their innumerable
nooks and crannies, human beings maneuver, motivated by a diverse mixture of
basic drives, dreams, and emotions, as well as more elevated considerations.
And, as we all know, such individual mixtures vary enormously, and they are
not determined by the environment alone (as the oversocialized theories imply).
Not infrequently, they will come into conflict with established social values,
leading to breaches both of trust, custom, and law. That is what makes the study
of human action both so frustrating and so fascinating, and why theories
explaining human action and social evolution from just one perspective always
remain inadequate.
177
III IT and the

Preconditions for
Organizing
The four chapters in Part II have tried to show how human limitations have
constrained the development of organizations and how we have developed a
succession of tools to alleviate or circumvent these limitations. The foundation
is now in place for the analysis of information technology and its contributions.
Before we move on, however, it could be useful to sum up the conclusions from
Part II.
Chapter 4 (Confined by Physiology) began by looking at the six basic human
preconditions in more detail. The fickleness of our memory, our limited
information processing capacity, and the very short range and limited channel
capacity of our natural means of communication are the main factors delimiting
our natural capabilities for organizing. The chapter also noted that we are only
partly rational beings and that our actions are strongly influenced by emotions,
rooted in the deeper and more archaic parts of our brains.
Chapter 5 (The Dawn of Organization) explored the problems of
organization building in societies without significant tools for organizational
purposes, and tried to determine the extent of the space of constructible
organizations in such societies. I suggested that there were two basic structural
configurations, the Adhocracy and the Simple Structure, building on the two
primeval coordinating mechanisms—mutual adjustment and direct supervision.
For larger structures, where one ruler or one council could not manage the
complexity, the iron constraints on human memory, communication, and
information processing-capabilities forced a reliance on two principles: the
delegation of authority and the encapsulation of information.
Adhocracies do not scale well, but the Simple Structure can easily be scaled
up by encapsulation and delegation, preferably with geography
178 III IT and the Preconditions for Organizing
and lineage as structuring elements. Such a system provides an extreme

economy with respect to information processing, communication, and
memorizing. Based on land rights and family lineage, this feudal type
organization contains its own structuring information; that information is
constantly enacted in everyday life and thereby reinforced in everyone’s
memory. At any level, the number of people the ruler must deal with is thus
kept within manageable limits. It can be viewed as a forerunner of the
Divisionalized Form.
Chapter 6 (The Power of Technology) discussed the nature of tools and how
the most important precomputer technologies alleviated our original constraints,
gradually allowing for extensions of the space of constructible organizations.
The single most important innovation was undoubtedly the art of writing, which
made it possible to externalize memory and thus lifted many of the constraints
placed upon us by our limited recall. Even more important for the development
of organizations was the accessibility of written information for large numbers
of persons. The emergence of implicit coordination of people who work with an
active file marked a watershed in organization history. There was also a gradual
development of a literate mindset, characterized by abstract and analytical
thinking and extensive use of symbols, finally extended to the vast majority of
the population through printing and mass education.
The other great field of organization-relevant innovation concerned
communication—which quickly became the bottleneck for organization
building when the memory barrier was lifted. Communication has two aspects,
which for a long time were one and the same: physical transportation of people
and goods, and communication of information. Even if we know of regular
courier services as early as 2000 BC, communication technology capable of
serving as an infrastructure for mainstream organization building had to wait for
the Industrial Revolution. However, the bandwidth problem remains: regardless
of channel capacity, we can still only absorb 250 words per minute, and output
even less than that.
When it came to information processing, the ability to write down
intermediary results and collect written information made it possible both to
process much more complex problems and to time-slice (work on many
problems more or less in parallel) much more easily. The real revolution,
however, was the way writing let us distribute large tasks among a vast number
of persons, synchronizing and coordinating their activities and communicating
intermediate results between them. A literate society can therefore organize
massive undertakings and routinely tackle tasks that would completely
overwhelm any illiterate society.
Finally, mechanical automation helped us overcome our limited ability to
carry out physical operations in parallel. The machine is, so to speak, a set of
crystallized decisions, and it represents a total externalization of a
179
plan for a specific production process. Thus, even coordination is automatic in

an automated production line—that is one important reason for its phenomenal
efficiency. However, even if a steady succession of tools has enhanced our
capabilities, important parts of the basic limitations have prevailed—notably in
our abilities to communicate and process information.
In Chapter 7 (The Modern Organization), I then tried to assess the
relationship between the development of these tools and the emergence of the
modern organization. The most significant developments did not appear until
the needs of the growing firms during the Industrial Revolution outgrew the
capacities of traditional organization. The key concepts here were specialization
and mechanization, which required much more emphasis on coordination and
control and entailed splitting the problems of design and production methods—
an integral part of the craftsman’s work—away from the worker. This called for
a much more sophisticated approach to information processing and
communication, and, consequently, to organization. The Machine Bureaucracy
was born.
The development of the Machine Bureaucracy depended on the emergence of
a new concept for coordination—indirect supervision by the means of
standardization of work processes—which resulted in two new coordinating
mechanisms: explicit routines and automation. Both these new coordinating
mechanisms depended on writing; automation required additional technological
advances. With the addition of implicit coordination, there were now three new,
technology-dependent coordinating mechanisms available that supported the
development of very large and efficient organizations.
The new organizations also represented another decisive break with the past:
They required detailed and explicit analyses of both the operational
requirements and the interdependencies of the different steps and levels in the
process, to be followed by careful and detailed design and planning of
operations. The patterns of action constituting the new organizations were thus
consciously constructed according to a conscious design based on an explicit
analysis of the desired outcome and the available means. Explicating analyses
and design and committing them to paper, the new organizers also created
(unknowingly) the first explicit conceptual models of organizations. By lifting
the models out of the subconscious world of tacit knowledge, and literally
spelling them out, they also opened them up for conscious inspection and
improvement. This is the foundation of the modern organization. The chapter
concluded that it is reasonable to assume that the possibilities awarded by these
preconditions have largely been exhausted by countless trials and errors, and
that we will need new technological developments to evolve distinctly new
organizational schemes.
It is now time to proceed to analyze these developments. The approach from

now on will be more detailed, and divided into three parts (Part III, IV, and V).
Part III will delve into information technology itself and the way it helps us
relieve main limitations above and beyond what earlier tools have done. Part IV
will move on to discuss the new organizational opportunities that information
technology opens up. Part V will return to the subject of structural
configurations, now with information technology as a prerequisite.
Part III starts with Chapter 8 (Information Technology Characteristics), in
which I try to assess the state of the art of the technology and the likely
achievements in basic performance improvements during the next decade.
Chapter 9 (The IT-Based Preconditions) will proceed to analyze how
information technology can improve the capabilities of the individual over and
beyond the contributions of earlier technology. Following the conclusions in
Chapter 2, that organizations are constructed and that their system properties
derive from the qualities of the actions performed by individual organization
members, this discussion really represents the foundation for the analysis of
possible new organization forms and practices. To balance the fairly
technocentric discussion in Chapter 9, which mainly explores the basis for the
technical space of organizations, Chapter 10 (Emotional Barriers and Defenses)
will end this part by discussing emotional barriers and defenses against
technology-based changes—problems which are, in my view, generally
underestimated and ignored by the industry.
181
8 Information Technology
Characteristics
“O that a man might know

The end of this day’s business ere it come!”
Shakespeare, Julius Caesar, 1599–1600
Needed: A Realistic Assessment

One of my more cherished computer memorabilia is a graph produced in 1986
by IDC (one of the big players in the market analysis and forecast arena),
forecasting the developments in market shares for the “Primary Operating
System Environments” of the next ten years (through 1996). Among a number
of lesser mistakes, one stands out: Microsoft Windows, the increasingly
dominating environment in 1996, is not mentioned at all, and the top slot is
awarded to OS/2. This is not so strange, since it was not really introduced until
1987, when Microsoft’s own Excel spreadsheet was the first application to take
advantage of it. However, it should teach us to be humble before the task of
predicting developments in this industry, where the achievements have been so
impressive since the first experimental steps were taken in Britain, Germany,
and the US during and just after World War II. The basic price/performance
level of computers has improved more rapidly than for any other technology we
have seen, and the rate of adoption has been very high, especially since the PC
made computers affordable for almost any budget. Already, much has been
achieved that has irrevocably changed the preconditions for human work and
organizing.
Even if we know this, it can easily fall into technological myopia’s double
trap: at the same time becoming too conservative in short term judgements and
wildly futuristic in medium and long term judgements. When thinking about
one’s own business, where the details are well known—the installed
technological base, the budget for upgrades and new systems,
the people in the organization, the products, the competition—it is easy to get
caught by the present level of technology and the present practices, and fail to
take into account the dramatic and steady improvements in basic technological
capabilities and price/performance levels. Stepping outside the immediacy of
our everyday frame of reference, however, it is just as easy to get swept away
by bold predictions about how a galloping information technology will soon
stand society on its head and totally transform our lives.
This is not a new phenomenon. All great new technologies have had their
heralds and their prophets promising all-embracing changes, while countless
businesses have simultaneously been swept aside because the proprietors did
not see the short-term changes wrought in the basic preconditions for their
existence. It is tempting to stick to the telephone as an example—as we have
already noted how the chief engineer of the British Post Office in 1879
explained to the House of Commons why the telephone had little future in
Britain. About 20 years later, General John J. Carty, then chief engineer of
AT&T, was more bold, extolling the peace-making qualities of the telephone
(Pool 1983, p. 89):
Some day we will build up a world telephone system making necessary to all
peoples the use of a common language, or common understanding of languages,
which will join all the people of the earth into one brotherhood.
There will be heard, throughout the earth, a great voice coming out of the ether,
which will proclaim, “Peace on earth, good will towards men.”
This was written at about the same time as an eminently practical man, the
Japanese general Oyama, pioneered the use of telephones in warfare—his troops
strung telephone wires behind them as they advanced against the Russians in the
Russo-Japanese war in 1905, connecting all the regiments along a 100-mile
front to fifteen regional headquarters, three group headquarters, and finally to
the general himself, sitting in his headquarters ten miles behind the front line
with an excellent grasp of the unfolding events—in contrast to his Russian
counterpart, who had to rely on orderlies. Oyama’s victory was squarely
attributed to this innovative use of the telephone (Pool 1983).
Some of the visions are even recurring ones—presented anew for succeeding
generations of technology. A persistent vision in this class is the idea that we all
will end up as high-tech couch potatoes, working, living, and entertaining
almost solely by means of the wall-sized screens (or even three-dimensional
holograhic display units) in our living rooms-cum-offices. Now, drawings of
such rooms were presented as early as a hundred years ago (Dahl 1984)—the
only difference was that movie screens, telephones, and printing telegraph
receivers (for continuous news services) took the place nowadays reserved for
computers, videophones,
8 Information Technology Characteristics 183
and fax machines. And, come to think of it, has not the Internet also been
heralded as a a peacemaker, because it will connect, on an equal basis, people of
all nations, creeds and political convictions? Repression is no longer possible,
we hear, because the Internet is impossible to censor. Indeed it is, but physical
access is still needed, and sufficiently determined regimes will have little
problem controlling physical access for the masses. Let us not underestimate the
ruthless ruler—history shows him always able to turn any new technology of
value to his enemies into a tool for himself as well. As we shall briefly explore
in the last pages of Chapter 14, information technology is actually an eminent
tool for gifted dictators.
What we need for our purpose in this book is an assessment that avoids this
double trap—avoids visions of indeterminable future states, and avoids getting
caught up with contemporary products. We need an analysis that uncovers the
basic characteristics of the technology while preserving a realistic view of its
deep development trends.
This is not an easy proposition. The properties of computer systems are
multifaceted, and the products that are brought to market represent a
bewildering array of tools and gadgets, with market lives of a few years at the
most—some as short as a few months. Often, they are so complex that the
average user never utilizes more than a fraction of the functions available. This
profusion of products, often presented through high-strung marketing blitzes,
and extensively covered in the media, makes it difficult to distinguish the
important from the insignificant, and the truly revolutionary from the
superficially sensational. We must nevertheless try to do precisely that—to
ferret out the most important development trends without being led astray by
marketing hype and general excitement.
Since the virtues and deficiencies of specific products are of no consequence
for our purpose, I believe it is possible to make such an assessment. What are
important are the general capabilities of the technology, the capabilities that
allow the products to be created in the first place. These developments
constitute the deep trends of the industry, and they have fortunately proved to
be extraordinarily stable over a period of several decades. As long as we keep to
them, we stand a much better chance of being largely correct even for
predictions stretching a decade or two ahead. As an example, we may return to
the IDC forecast just mentioned: IDC was wrong in predicting that OS/2 would
lead in the market in 1996, but it was wrong for the right reasons. Their forecast
built on the belief that rapidly increasing price/performance ratio of PCs would
make them proliferate (which they did); that a multitasking, graphically
oriented, windowing environment would win in this new market (which is what
happened); and that this environment would thereby rise to the top slot in
market share (which it did). However, it was not OS/2, the obvious candidate at
the moment, that succeeded,
but the soon-to-come Microsoft Windows. This substitution, however important

to business analysts, is only of marginal interest in the greater picture of
computer utilization. The interesting point in that perspective are precisely those
where IDC hit the bull’s eye.
Three Basic Characteristics

Before we go on, it is necessary to establish just what constitutes the basic
characteristics of computers and computer-based systems. And, even if the
multitude of computers and computer-related products on the market exhibit
great variations in performance and capabilities, a closer look at systems past
and present will reveal that all their capabilities and functions can be related to
three basic properties or characteristics.
First of all, computer systems of all kinds process information—they operate
on it in some way or other. This capability has two aspects of equal importance:
The first is the actual operations on information (for instance, adding numbers
together); the second is the program—the specific instructions deciding the way
the physical logic of the processor itself shall operate. Next, computers also
store information, usually both programs and data, but at least a program of
some sort. Third, they communicate—data and program must be put into the
computer in the first place, the results must be presented to the user, and
information is often transmitted to other systems, either for further processing,
storage, or presentation.
The decisive underlying technology here is of course the representation of
information and programs alike in digital form—in 1s and 0s. All digital
information is thus represented by absolute values that can be copied,
manipulated and transformed indefinitely without degradation, unlike the
progressively attenuating amplitudes of analog technology. Without digital
representation of information, the computers and applications described in this
and later chapters would largely be impossible. (For an account of the
ramifications of the principle of digital information representation that is both
entertaining and enlightening, see Negroponte 1995).
The Externalization of Processing

In many ways, ENIAC—traditionally recognized as the first electronic
computer—was not a genuine, multi-purpose computer. It was a specialized
calculator, more like an electronic version of the Jacquard loom, optimized not
for weaving, but for solving mathematical equations of a certain kind (ballistic
trajectories for firing tables). However, in important
respects, it was also very different from the loom. It could punch intermediate
results on cards that could later be fed into it for new rounds of calculations, it
could loop, repeat subroutines, and do conditional jumps—that is, branch
the execution of the program in one of several directions based upon the results
of previous calculations. It already exhibited the tree basic characteristics of
computers, and it was more adept at solving equations than any other machine
before it. But, it was not very flexible—it was programmed by setting physical
switches and plugging cables in something that resembled old-fashioned,
manual, telephone switchboards, a maddening task that took considerable time.
It was the stored-program computer (of which the first one—the Manchester
Mark 1—was built at Manchester University in England) that unleashed the real
power of digital computing.
When the program escaped physical wiring and could be entered in a
rewriteable, electronic memory, it could both become more complex, it could
modify itself while running, and it became easily interchangeable. It is this
almost unlimited programmability that makes the modern computer so different
from a traditional machine, with its extremely limited programming logic (that
quite literally has to be forged in steel). However, in one very important respect
even the computer remains a classical machine: It cannot go beyond its set of
design objectives; it cannot do anything that has not been spelled out in
painstaking detail in a program on beforehand. Even with self-modifying
programs, the rules for the modifications are given by the programmer, and
programs that can “learn” from “experience” also obey preordained rules. The
instructions may not always be conscious—there will be errors and
ramifications the programmer was unaware of—but instructions they are.
Since programs are immaterial, the room for complexity is almost infinitely
large compared to physical automation, which is limited by material restrictions.
The result is that even small computer programs (and even the programming
that is inherent in the circuit designs) are immensely more complex than
mechanical automation can ever aspire to. The computer therefore allows us to
build logical “machines” that are many orders of magnitude more complex than
any physical machine we could conceive of. Electronic processing is also
infinitely faster than the movements of wheels, levers, pistons, valves, and other
actuators can ever be.
What does this really mean, then? What is the main contribution of the
electronic, digital computer when it comes to human work and organizing? I
have argued earlier that the main gift of writing was the externalization of
memory. Looking back over about 5000 years, this conclusion is
uncontroversial today. Some may think it premature to proclaim already the
main contribution of the digital computer, but I
believe that our experience during the decades since its first appearance
demonstrate this beyond reasonable doubt: it is no less than the externalization
of information processing—the possibility, for the first time in history, to
process information outside the human head. As such, it represents the first
tecnological innovation that matches the importance of writing. It might of
course be said that externalization of processing has some modest roots in
mechanical automation and computation devices, but the new frontier was not
opened up in earnest until the advent of the programmable, electronic, digital
computer—the first truly Universal Machine.
Just as the externalization of memory, the externalization of processing is of
course not in any way comprehensive—it does not cover the whole, broad range
of human information processing, with its rich web of logic and emotions.
Computers can only mimic certain aspects of the human mind. But this must not
fool anyone into thinking that those aspects are inconsequential or unimportant.
Experience has already shown us that computers excel in performing logical
operations; in information retrieval, selection, sorting and monitoring; and in
number crunching and quantitative conversions (such as converting information
from numbers to pictures)—types of processing that are extremely important in
administrative work, material production, science, and in all kinds of
information analyses.
This is a kind of processing, then, that we can externalize—that we can
offload to computers, and it is when we exploit them to multiply our capacity in
these areas that we really experience dramatic changes in productivity. Indeed,
computers already allow us to perform tasks that would be impossible even in
theory with human brainpower alone—no matter how many persons the project
employed. There are simply not enough people on earth to carry out all the
calculations involved in, for example, the modeling of the athmosphere that lets
us study global warming or produce ten-day weather forecasts—and if there
were, you would surely not be able to organize them (let alone pay them!). And,
as the processing power of computers increases, the multiplication factor for our
externalized processing only grows.
The development of the digital computer was incremental, with a lot of
different persons contributing with a large number of small steps. It is debatable
whether any of those steps involved a shift of paradigmatic magnitude; a
transition from one kind of machine or tool to a fundamentally different one. It
may even be debated whether the computer indeed is different in principle from
previous machines—or if it just a very powerful kind of self-regulating machine
(Wiener 1954).
However, even if the computer at the outset did not constitute something
principally new, the quantitative changes it has undergone since its
inception add up to a decidedly qualitative difference in relation to other kinds

of machines. To draw a parallel, we may well say that both the earth and the
smaller asteroids are lumps of rock circling the sun. But there is, literally, a
world of difference between the earth and a small asteroid, rising from their
different size and relative positions. The modern computer has a complexity,
flexibility, information-processing capability and storage capacity so immense
compared to any other machine that it constitutes a totally new class of devices.
The possibilities it opens up are in many areas profoundly different from those
that arise from traditional machinery, and they are all rooted in its paramount
characteristic: its programmability.
Trends in Processing Power

When it comes to processing, speed is obviously important, and it has grown to
become more important than we originally thought. If you had described the
processing power of the modern PC to a computer engineer or computer user in
1955, he (it would surely have been a he) would have wondered how a single
person could possibly utilize more than a fraction of that capacity. The users of
today, with PCs strained by the demands of the latest version of their office
suites, know better. In fact, as we enter the new world of pervasive IT, of
graphical user interfaces, of multimedia and giant databases, our thirst for
increased processing power is more acute than ever before. Will we never be
satisfied? There is no reason to believe that the annual increases in power that
we have grown accustomed to will slow down in the foreseeable future. The
predictions for the years through 2010 are fairly safe, since we can even
maintain the present rate of progress simply by the gradual refinement of
existing technology. There is also every reason to believe that we will be able to
continue the improvements even after the present approaches come up against
final physical limits (like quantum effects when chip details become too small).
If we look at the most familiar processors for the average user at the time this
book was written, the Intel 80x86 processor family (up to and including the
Pentium II/III 1), Intel was able to increase the processing power by a staggering
50% every year in the 20 year period 1978-1997 2, while at the same time
reducing the price for one MIPS (unit of processing power) by 25% per year.
This adds up to a 2000 times improvement in
1 The Pentium III is only a “media upgrade,” it is not a new processor at all. It is a Pentium II with a
new set of multimedia instructions, produced by an improved production process.
2 Figures are based on data compiled from articles and advertisments in numerous issues of Byte,
PC Magazine, and Scientific American.
processing power, and a 300 times improvement in price/performance in just 20

years. And it still goes on.
Today, data processing has come to mean more than numerical calculations.
Computers process not only figures, but text and graphics as well (including live
video). They store and retrieve vast amounts of information. At the bottom,
however, it is still just binary number crunching, and the speed of the processor
is of course directly dependent upon the speed of these basic calculations. In
modern, complex computers the speed of the primary and secondary memory
(mass storage devices), the capacity of the transfer channels between the
processor and the memory, and a host of other factors are of course also of great
importance for the total performance of a particular computer.
The key to the incredible increase in speed and the impressing reduction in
price we have seen, decade after decade, lies in the constant refinements of the
integrated circuit. The number of discrete components that can be put on a
single chip has increased tremendously, both because the size of components
have shrunk and because chip size has increased. The first integrated, one-chip
microprocessor (Intel’s 4004, launched in 1971) had a mere 2300 transistors.
The Intel Pentium II/IIIs of 1997-99 had 7.5 million (including on-chip Level 1
cache memory). That is an increase in the number of transistors per chip of
roughly 35% every year for 27 years—almost a doubling every second year. We
can expect this trend to continue for some time yet, and remain confident that
new approaches will be found when needed.
The increase of power at the chip level does not automatically translate into
corresponding gains in power at the system level. But, rather than being less, the
gains in system power are in fact likely to be even greater than the gains in
processor speed, as new, parallel architectures are perfected. The gains will
come for systems on all levels, from the most humble PC through database
servers and high-end transaction-processing machines, as well as workstations
and supercomputers for technical and scientific applications. Moreover, an
increasing number of specialized processors are added—such as graphics
processors, communication processors, I/O processors and sound processors. An
average user will have a multitude of superfast processors working for him or
her—allowing for much more sophisticated software and greatly improved user
interfaces.
If some of the basic problems of parallel processing are solved, standard
processors may become even more commodity-like than today and fall into a
pattern resembling the one we now have for memory chips: extremely high
volumes, low prices, and liberal use. This will allow not only for new
qualitative jumps in user friendliness, but also for the use of substantial
processing power in the most trivial circumstances. Not that
many years ago, people joked that even toasters would have processors in the
future. Today, some toasters have processors, which help them toast the bread to
the same degree of crispness regardless of whether the slice comes from a
freshly baked loaf or from the freezer.
Together with the advances in storage media and communication, the
increases in available processing power will also usher in the inclusion of high-
quality sound and pictures, including live video. Multimedia PCs are already
swamping the marketplace, but their capabilities will increase dramatically in
the future (without significant price hikes), and many forecasters (beware!)
predict the demise of the “dumb” TV.
It is tempting here to stray a little from the realistic assessments we are after
in this chapter—for more than any other tool, the computer is becoming an
expression of the human mind rather than of the human hand. Indeed, the
hardware can be viewed as a kind of materialized spirit—its power coming not
from physical force, but from its speed and accuracy in carrying out logical
operations. And the logic content of computers is increasing all the time, in step
with the miniaturization of electronic circuits. Whereas ENIAC was a 30-ton
agglomeration of rather crude matter, with a very modest logic content, a
modern microprocessor, several orders of magnitude more powerful, is almost
immaterial—weighing less than one gram without its protective coating. Even a
complete computer, with screen, mass storage, power supply, and keyboard, can
now weigh less than 300 grams, and the weight of these “palmtops” is still
going down.
Today, microelectronic chips are the medium for this logic, but there is no
necessity in this—it only means that such chips are the most economical and
convenient carriers at our present technological level. Other technologies will
take over later, and, in a not too distant future, logical operations may be carried
out by the manipulation of single electrons and photons. With intangible logic
thus contained in almost immaterial quanta of energy, one realizes that the old
philosophical debate about mind and matter is not nearly over yet.
In fact, if we look further ahead than the timeframe we otherwise adhere to,
the evolvement of the computer may give this debate a new fervor, a new
perspective, and a whole new set of arguments in the twenty-first century. I
have always been a skeptic when it comes to the question of computers
eventually attaining the same level of complexity as the human brain—partly
because the brain is so exceedingly complex, and partly because our knowledge
of its intricacies and operation is so limited, that we could not even use this
processor complexity if we could produce it. However, if we consider the
strides made in the few decades since the completion of ENIAC, versus the
more than 300 million years that have passed since mammals and reptiles
evolved from their common ancestor
(and the development of the brain started even long before that), I am not so
sure anymore. We have just discussed the power of computers in the near
future. But how powerful will they become further down along the road? In the
February 1996 issue of Internet World, science fiction writer Vernon Vinge
points to the fact that if the current trend in processing power improvements
continues, computer hardware processing capacity will actually reach what we
now estimate to be the level of the human brain already between the years 2030
and 2040. What about the situation a hundred years after that? Five hundred?
Two thousand? Will we be able to develop software that can take advantage of
this tremendous processing capacity and create entities with higher total
processing capacity than ourselves? If so, what will the consequences be? Who
knows—but one thing is for sure: the computers of the more distant future are
going to be incredible indeed when judged by today’s standards. As Arthur C.
Clarke used to say, any technology sufficiently more advanced than the one we
know will always look like magic. So would the computers of the distant future
seem to us, if we could catch a glimpse of them today.
The Future of Software

Processing power is useless without programs. In many ways, the development
of computers, or at least the use of computers, is a matter of software
improvement. To say something about the future of software, however, is the
most difficult task of all. The number of companies and people engaged in
software development is so large and the latitude for creativity in that area so
great that almost anything can happen—as far as it satisfies a need (real or
imagined) in the user community. What we can say is that, as more powerful
computers can take on ever heavier software loads, programs become more
comprehensive, more complex and increasingly incorporate very computing-
intensive components, not least at the user interface (such as graphics, video and
sound).
Standard packages are moreover becoming more and more flexible, due to an
increasing number of functions, options, and open parameters. Programming
languages, on their hand, are increasingly equipped with libraries of subroutines
and other basic program elements that increase programmer productivity.
Organizations therefore have a rich set of options to choose from when they
need new systems.
Working against the variety in main application areas is an increasing
globalization of software. Since the costs of changing software is considerable
and generally increasing, people will put great weight on the business prospects
of their vendors—especially for mission critical applications. To ensure the best
possible compatibility and interoperability between software packages, there is
also a strong tendency
toward choosing the market leader even if smaller competitors offer better
solutions. In some product classes, notably word processing and a few other
office products, we are actually coming close to a de facto monopoly on a
global basis. The number of serious players in the database market is also
rapidly dwindling. However, the major products are now so rich in functions
that they already represent a significant degree of overkill for the average user,
and most organizations are not anywhere near of getting full mileage out of the
systems they already have.
In fact, the bottlenecks of software development are rapidly becoming not the
programming itself, but the processes coming before and after. The mounting
complexity of the analyses and planning needed to create really large software
systems is now taxing the skills and intellectual capacity of both users and
analysts, and is not infrequently defeating them—leading to aborted projects or
software with major deficiencies. After a system has been completed, the
intricacies of the new software and its organizational ramifications often require
such skill and knowledge to really take advantage of them that the software is
poorly employed at best, and a prolonged period of trial and error is needed
before operations stabilize.
Storage
My first job after I finished my MA in sociology was in the personnel
department of a shipyard in Oslo. The shipyard was old; it was started as an
engineering workshop in 1841, and moved to the sea front and turned into a
shipyard in 1854. It eventually evolved into a modern yard, building some of the
first semi-submersible oil drilling platforms used in the North Sea. Countless
workers had passed through its gates since its inception, and a succession of
devoted personnel clerks had scrupulously kept the files for all who left,
probably in case they should return later (which was quite common—many
alternated as sailors). In the 1970s, when I worked there, the personnel office
had complete files for more than 100 years of employees. What a treasure trove
for a sociologist! But, alas, the information was written on thousands of
individual cards, the only indexing scheme was alphabetical by name, and the
effort required to extract even a fraction of the data these files contained was
prohibitive. So, although the information was physically there, it was not
accessible in practice. Had it only been in a database! Then, analysis would
have been comparatively easy and cheap—and of great interest—not only for
scientific purposes, but also for the company.
The database, however, is an even more recent phenomenon than the
computer. Computers were invented to make calculations, not to store
information, and storage was not exactly their strong point during
their first decades. ENIAC could store only twenty 10-digit numbers in its
internal accumulators while running—a meager 800 bits—apart from the
program, which was “stored” in wiring and switches. Data was fetched number
by number from punch cards. Because of its limited ability to store intermediate
results internally, more complicated calculations usually meant that intermediate
results had to be output to punch cards, which then had to be loaded for the next
sequence of computations. The first stored-program computer, the Manchester
Mark 1, had a CRT-based internal memory of between 6144 and 12 288 bits,
but still had to rely on paper tape for secondary storage. The first UNIVAC
computer represented something of a breakthrough with its internal memory
capacity of 84 000 bits (10.5 Kb in today’s language) and its magnetic tape
secondary storage with megabyte capacity (up to ten tape units storing more
than one megabyte each). After a period with quite exotic memory devices, like
CRTs and glass tubes filled with mercury, the magnetic core memory in the
1950s became the first practical way of equipping computers with a reliable and
comparatively large memory. However, we had to wait for the semiconductor
and the integrated circuit to make it both really large and affordable—and allow
computers to shrink to desktop size and price. The first practical
microcomputer, the Altair, appeared in 1975 with a basic capacity of a tiny 256
bytes (2048 bits), but soon 4 Kb add-in boards were available. When Apple II
was launched in 1977, it had 16 Kb of memory as standard (expandable to 64
Kb, as the Altair).
Still, limited memory capacity for a long time represented a serious
bottleneck for computer performance, since the size of the memory decides both
the size of the program modules running at any particular time, as well as how
often the computer must access its secondary (and much slower) storage
medium for reading and writing data. Especially when sorting and indexing
(common database operations) the size of the available memory has a very
decisive influence on execution time.
The low capacity and high price of secondary storage likewise limited the
computer’s role as an archival device for a long time. The first mass storage
devices were punch cards and paper tape. Then magnetic tape came along, but
even if it represented a great improvement in speed and capacity, it was still a
sequential medium—to get at a piece of information in the far end of the tape,
one had to spool it from one reel to another. The first random-access medium
was the magnetic drum memory of the late 1940s and early 1950s, but a
satisfactory solution was not found until magnetic disk memory was introduced
by IBM in 1956 and used in the IBM 305 RAMAC computer. The disk in
IBM’s first unit was 24” (61 cm) in diameter, had 50 platters stacked on top of
each other on the one shaft, and stored 5 Mb of data.
Today, the situation has totally changed, and it is still changing fairly
dramatically from year to year. Storage capacity, both in terms of primary
memory and secondary storage, is becoming ever cheaper, even when measured
in relative terms against the increasing demands from new hardware and
software. Although there are still some problems at the extremes, especially in
high-end graphics processing, ample storage is now increasingly taken for
granted.
Trends in Storage
Since the early 1970s, semiconductor memory has ruled the market for primary
memory, and the price performance ratio has improved steadily—even for
memory chips, the number of components per chip has increased by about 50%
per year, and still shows no sign of leveling out. The cost per megabyte has been
reduced by 35% every year since 1975, when it was $20 480 (chips only). 3
Because of the geometrical regularity of the design, memory chips can be much
more densely packed, and have a much larger number of components than
processors. At the time of writing 64-million cell chips (each cell consisting of
one capacitor and one transistor) are in volume sale, and one-billion cell chips
are planned for. Such a gigabit chip will be able to store (with today’s 8-bit
character standard) about 134 million characters, the equivalent of about 65 000
pages of text like this one—more than some people will read in their whole life.
With the gigabit chip, we will approach the limits for further improvement of
the venerable silicon memory chip. Larger chips may be manufactured by
increasing chip area, but continued shrinking of transistor size will come up
against the emergence of quantum phenomena—the chance jumping of
electrons across the insulating barriers. Because of the unpredictable nature of
these quantum jumps, they will destroy the reliability that is so important for
computer memory. It would be foolish, however, to suppose that this signifies
any permanent barrier to further improvement in the price/performance of
computer memory. Other technologies are already on the horizon, and even
newer ones are bound to appear further down the road.
For several years now, the increase in memory price/performance has been
faster than the increase in need for memory capacity. This trend seems to
continue, and we have already started to use memory freely, without bothering
too much about the cost. This will have significant
3Figures are based on data compiled from articles and advertisements in numerous issues of Byte,
PC Magazine, and Scientific American.
consequences for how computers will look and operate in the years to come.
When it comes to mass storage, disk development has relegated magnetic
tape to a position as a medium for backup and archival purposes. An increasing
number of gigabytes can be stored in surprisingly cheap miniature disk drives
(3.5” and smaller). The development has followed the same pattern as for
processors and memory, and has been just as stable. I have only seen data from
1983 onward, but for the 12 years through 1994, the price per megabyte of
magnetic disk storage fell by roughly 38% per year—from $100 to 50¢.2 After
that, it has fallen even faster—in 1997 storage costs on cost-effective magnetic
disks fell below 5¢ per megabyte, only a fraction of the cost for storing text on
paper, if printing and storage shelf/cabinet costs are counted (around $15 per
megabyte). With the present development, paper is even losing its cost
advantage for graphics storage. Disk storage is actually becoming “free” for all
but the most extreme storage needs, and even better performers for very high-
volume storage are in development—among them holographic storage of
information in light-sensitive crystals. This holds the promise of a new
revolution in storage with both higher densities and far higher speed than
conventional media, although progress has proved to be slower than anticipated
(Parish 1990, Baran 1991, Psaltis and Mok 1995, Thompson 1996).
The biggest drawback of magnetic media is that they lose their data gradually
over the years and must be refreshed from time to time (tapes in archives are
usually refreshed every two or three years). For archival purposes, certain types
of optical disks are therefore preferred, which—according to conservative
estimates—will retain their data uncorrupted for at least 60 to 100 years
(Harvey 1990).
Whatever mass storage technology wins in the future, however, we can be
quite confident that we will have available abundant capacity at very low prices,
enough even for the storage and real-time playback of high-definition video
movies. This will mean that digital storage media will become the most
economical and compact alternative for all types of information storage. As
Nicholas Negroponte has pointed out (Negroponte 1995), this will put a
significant pressure on the traditional media and possibly reduce their roles
considerably.
Pillars of the Memory Revolution

However, the large available volume is not the most significant aspect of
computer storage. Indeed, the theoretical space available for information storage
does not even increase as we make the transition to digital storage. This may
seem surprising, since data already takes up so much less space
on disks than on paper—but the earth is pretty large, and there is always a blank
sheet of paper or a new file card available. We should also remember that
microfilm is not a child of the electronic computer, and microfilm density can
be quite respectable—especially for ultrastrips (which is, literally, microfilmed
microfilm). Even if the new digital media now have overtaken microfilm as
well, it is only when their extremely compact and cheap storage is combined
with the access, search and retrieval capabilities of the computer that we
achieve something truly new.
Like written files, digitally stored information is also in principle available to
all and anyone. But, compared to the written file, which requires that you (or
someone else on your behalf) physically walk up to it to retrieve information,
digitally stored information is so very much more accessible. Anyone with a
terminal connected to the system, directly or through a communications
network, can access it, regardless of geographical location or time of day. It is,
moreover, simultaneously available to a large number of users—how many is
determined solely by technical factors such as system capacity and the number
of communication ports, and the limits here are steadily and briskly being
pushed upward.
If we envision future computer systems—of, say, several decades into the
twenty-first century—with extremely powerful parallel processors, holographic
mass storage and direct fiber connection to a fiber-based telecommunication
network, their transaction capacity will be several orders of magnitude greater
than the most powerful systems available today, and they may serve global
communities numbering hundreds of thousands, even millions, of simultaneous
on-line users. Even the largest multinational corporations would then easily be
able to consolidate their operational databases into either one unified virtual
base (distributed among many physical sites) or one central physical base. The
actual solution chosen would depend on the level of integration required in
running the organization and its business activities. The idea of a central
database for a multinational corporation may seem preposterous at this point in
time, but I am not so sure it will look that way in a couple of decades.
Vast amounts of information are of no use, however, if you cannot retrieve
the items you need when you need them—and to retrieve them you first have to
find them, just as with written information. Luckily, the fact that digitally stored
information can be read by a computer also means that the computer can search
for us, as long as we can provide relevant search criteria. The computer can also
index, cross-index, sort, and compare with enormous speed. It can retrieve one
record from among millions in a small fraction of a second, and, just as quickly,
store it back again after it has been changed. It can select groups of records on
the basis of certain properties and sort them according to various other
properties,
it can count them, and so on. The computer’s outstanding ability to search vast
amounts of information in an incredibly short time, and extract, combine, and
concentrate data, makes for a momentous difference between computerized and
paper-based files.
When we talk about records, we mostly mean information that is precisely
defined and put into a strict format, like accounting data for a bank or customer
information for an insurance company. The items of information and the form
they are going to be stored in are decided on beforehand, and for each item a
corresponding field is defined in the database. The field is normally designated
at least as numerical or alphanumerical, and will usually have a maximum
number of positions, or even a mandatory number of positions (as with dates
and article numbers). There are also usually a number of other design options
for each field.
The advantage of storing information in this highly structured form is that it
can be easily retrieved, counted, and classified; numbers can be used in
calculations; names and addresses can be used to produce mass mailings; and so
on. It is simply a prerequisite to automatic processing: The programs must
“know” exactly what kind of information they shall retrieve, where it is stored,
on what form, and exactly what to do with it and where to put it afterward. It is
structured databases such as these that lie at the bottom of almost all the familiar
success stories about profitable use of IT that circulate in the business world
(and in the realm of public administration, for that matter).
There is quite a lot of information that is impossible to accommodate in a
structured database, however. In fact, even these databases will often contain a
“comment” field, wherein unstructured, textual information can be entered—
information that is too important to be left out but too special to be defined in
advance, or simply too varied to be included in a classification scheme.
The first attempts to use computers to store and retrieve more “soft”
information were just a few years behind the applications focusing on structured
data. As early as the second half of the 1950s, 20 years of headnotes of design
patent law cases had been entered into an IBM 305 RAMAC (the machine with
the world’s first magnetic disks). At the same time, what was probably the first
full-text “database” came into being when Professor John F. Horty of the Health
Law Center of the University of Pittsburgh used the university’s computing
center to solve a practical problem: the actual implementation of a bill passed in
the Pennsylvanian legislature to replace the term “retarded child” (and all its
permutations) with “exceptional child” (and all the corresponding permutations)
in the state’s health statutes (Bing and Harvold, 1977).
After two consecutive tries with groups of students reading the statutes and
substituting terms, and with too many errors still remaining, the
complete text was registered on punch cards—and the substitutions left to the
computer. Horty then found that the machine-readable text could be exploited in
much more exciting ways as well, and he went on to develop what was probably
the world’s first full-text search and retrieval system.
The benefits are here already—just think about the full-text databases now
offered by many leading newspapers around the world, or the improvement in
literature searches provided by computerized book and journal catalogues—not
to mention the rapidly expanding jungle of both serious and oddball databases
accessible through Internet. Of course, no search is perfect, as everyone with
some experience in database search will agree, and will probably never be, but
with some experience, computer searches are already vastly more effective than
anything before. Although they will not find all the relevant information, and
often not even the majority of it, the catch is always much more complete than
with manual searches, and exceedingly fast in comparison. The performance of
computer-based systems is simply so much better than the old, paper-based ones
that they can only be compared in principle.
In real life we may often encounter practical obstacles to information
retrieval—such as incompatible storage formats, inflexible database structures,
inadequate application software, and the like. However, they all represent
temporary technological shortcomings or are the results of the vendors’
commercial considerations. They are not consequences of the technology’s
inherent properties, and can thus always be overcome—even if it may cost a lot
of money sometimes.
Communication
From Artifacts to Waves and Currents
Communication has always involved both physical transportation of goods and
people and transfer of information. Except for marginal technologies such as
semaphores, drums and the proverbial smoke signals, transfer of information
over almost any distance before the telegraph was equivalent to physical
transportation, since it invariably involved people (messengers), tablets, paper
or paper-like materials. The telegraph, and, later, the telephone, radio, and
television, changed this and established information transfer as a separate
category—the symbols of human communication escaped from the world of
paper and parchment and became embodied by radio waves and the current in
telegraph and telephone wires.
Computers do not change this in any basic, physical sense. Electronic mail
and telefax are much more efficient than the telegraph, but they still rely mostly
on electrical signals traveling along a wire—in some instances
probably even the same physical wires that earlier carried the telegraph traffic.
Even the introduction of satellite communication and optical fiber do not change
anything in principle—it is still a matter of transmitting symbols instead of
physical objects, even if the capacity of the carriers probably has increased
beyond the wildest dreams of both Bell and Marconi.
The context of communication has also broadened to include not only
communication between human beings, as was the case with all the pre-
computer communication technologies, but also direct communication between
computer systems. Of course, one may always claim that it is only a matter of
mediated communication between humans, since there is always a human
somewhere downstream and somewhere upstream. Both are often so far
removed from the direct effects of the communication, however, that I would
maintain that system-to-system communication is a separate category that
merits its own considerations.
When it comes to physical transport, the consequences of information
technology have been mainly indirect: The ship, the airplane, the train, and the
automobile are not inventions of the computer age; neither are the steam engine,
the diesel engine, the petrol engine, the jet engine or the electrical motor. But, of
course, there are important contributions that improve the performance of our
physical transportation systems, both with respect to design, operation and
administrative support. Modern jet liners, for instance, could not have been
designed and built without computers, they could hardly have been flown
without computers, and computer-based navigation and air traffic control
systems allow regularity under almost all weather conditions. Routing systems
for railroad cars are computer-based, trucks are directed with the help of
computer-based systems, and computer systems keep track of each single
package transported by express freight companies and courier services.
Electronic customs systems, as Norway’s TVINN, 4 speed the transport of goods
further by eliminating delays at the border. The contributions of information
technology to physical transport are marginal, however, compared to the
improvements originally brought by the development of the prime movers of
goods and persons: ships, trains, cars and airplanes.
Basic Input and Output

Communication is a many-sided thing, however. Most basic is the
communication between the computer itself and its users and programmers.
4TVINN was the second or third such system in the world when it went into regular production in
1988. (New Zealand was first, and Singapore came along about the same time as TVINN.)
Some may puzzle over the fact that I include both input and output of data under
the heading “communication,” but both constitute information transfer, and will
increasingly include direct data capture from sensors, data interchange through
common database access and messaging between computers. That way, they
become inseparable from what we normally think about as computer
communication. But let us start with the basics.
For any piece of data to be processed, and any program to be run, it must of
course first be loaded into the computer. As we have seen, loading data and
programs into the first computers could be quite demanding. Being able to load
programs from fast disks or over networks, we are far better off today. To enter
data we still often have to use keyboards, however—even if it is increasingly
captured directly from sensors, bar code readers and other automatic means of
data capture. Data can of course also be generated internally in the computer as
a result of transformations or processing of original data. Once registered, data
is stored in a mass storage device like those discussed earlier.
Keyboards have not changed much since the qwerty keyboard was devised a
hundred years ago, proving the enormous inertia of established standards.
Although more efficient keyboard layouts have been devised (for instance, the
Dvorak keyboard, with the most frequently used letters in the middle), it seems
that the old standard is going to keep its dominating position. Keyboards are
still our main instrument for communicating with computers (complemented by
mice and tablets of various kinds) and will remain so for many years yet. Both
speech and handwriting recognition have repeatedly been prematurely
announced, and all commercially available systems so far have had serious
limitations.
The problem is that both speech and handwriting recognition belong to the
difficult field of pattern recognition, where humans excel and machines are so
far ineffectual. Reliable recognition of continuous speech with normal
vocabularies from arbitrary persons is exceedingly difficult for a computer, and
requires both more sophisticated software and a lot more powerful hardware
than we have had available so far. Recognition of natural, flowing handwriting
from arbitrary persons is even more difficult, and will probably take more time
to solve than speech recognition. However, with the rate of improvement we
have grown used to in processor power, there is little doubt that both will
become available in the foreseeable future. Whether speech recognition will
succeed in the marketplace is another question, but those who prefer to talk to
their PCs will eventually have the opportunity of doing it for a very small extra
cost (if any). When general handwriting recognition eventually comes, it is
difficult to say if it will meet with success outside a number of niche markets—
especially since it will only appear after general speech
recognition has become an affordable reality. However, it may be a preferable

interface for taking notes and for editing and making corrections in text entered
by dictation or by keyboard.
Direct data capture by way of sensors is rapidly becoming more important,
however. It is no longer only a question of lab data or temperature and pressure
in processing plants. Increasingly, our shopping is registered automatically by
light pens or label scanners, payments (for goods, bus tickets, and pay phones)
effected by card readers communicating with smartcards, toll road passage
certified by machines reading chips glued to a car’s windshield, and so on.
Because of the huge savings in labor hours it normally represents, there is a very
strong impetus for increasing the extent of automatic data capture, and we will
see even more of it in the future.
After input and processing follows output. When ENIAC produced an
answer, it communicated it to its users by punching cards. Its successors rapidly
learned to address their users through screens and teletype printers. Printing
rapidly became the method of choice, as it allowed the user to read at his leisure
without tying up precious processor capacity. When processors became less
costly and more powerful, and users more craving for the direct responses of
interactive computing, screens became increasingly important, and are today the
predominant medium.
Printing is still very important, however, and paper is easily the preferred
medium for final output and presentation. It got a real boost during the 1980s by
the development of the low-cost laser printer—an incredible improvement over
the clattering one-page-per-minute typewheel printers of yesteryear. At first, it
lacked color, but there is no doubt that the price/performance equation even in
this field will improve to the point where high-quality laser color printing
becomes economically attractive for “the rest of us.” For ink jet printing, it has
already happened. Color is of course intrinsically more complicated than black
and white and should therefore invariably cost more, but in a mass market, it
does not always turn out that way. Consider photography: Even if color film and
prints are intrinsically more expensive than black and white, color costs (a lot)
less for the average consumer—simply because all the big consumer-oriented
labs as well as the small automatic developers only do color, thereby reducing
black and white prints to handicraft work produced by your local photographer.
Market penetration decides the price.
Printing also has a wider role in communication. The fax machine has been a
runaway success, and the reasons are obvious: It is low-cost and very easy to
use; utilizes existing telephone connections (and thereby addressing
conventions); transmits the output from any program and printer, including
one’s pen; and accepts graphics as well as text. Electronic mail will definitely
win in the long run, but the fax machine
has, at our present level of technological sophistication, provided a very simple

and elegant solution for rapid communications. With a fax-capable modem and
appropriate software, you computer can use fax machines as remote printers.
With cheap color printing and faster modems (or ISDN), an improved fax
standard could allow you to output directly to a printer halfway around the
globe with decent speed, normal print quality and faithful reproduction of any
fancy letterhead at the receiving end.
One of the main reasons for the heavy reliance we still place on paper—
perhaps the main reason—is the shortcomings of present screen technology.
Whereas all the other vital parts of a computer system have enjoyed a very rapid
and sustained increase in performance, screen development has been sluggish.
To do away with paper, we need screens that are large enough to show us a lot
of information simultaneously, so we can work the way we are used to, with
several information sources available concurrently. The screens must have good
contrast, high resolution and provide good reading comfort. Preferably, the
viewing position should be easy to change, to let us keep our normal habit of
shifting back and forth between positions when reading a report or a book. None
of the commercially available screen technologies today can fulfill this.
Really large displays are extremely bulky as well as expensive, and you
simply cannot buy the big, high resolution screens necessary to really do away
with all the paper on that desktop. To do so, you need full square meter screens
or larger—flat screens that are part of our desktops or even constitute the
desktop, preferably tiltable, and with a detachable panel for comfortable
reading. With adequate resolution and contrast, screens like that would really
take a bite out of the paper market—we would not need printouts for that final
control, documents could be distributed electronically, and the receiver might
prefer to view the information on-screen instead of printing it out. Even the hour
of the CD-ROM–based magazine might finally come, and the fax machine
would at last feel the competition of screen-fax and global electronic mail.
Unfortunately, such screens will not be available in the short run. There are
potential technologies under development, but they will need considerable time
to achieve the sizes and prices necessary (Chinnock 1997, Sobel 1998).
However, I believe we can be confident that such screens will appear in the not
too distant future.
Electronic Mail
Not long after computers got screens and text editors, entrepreneuring
programmers and users found a way of connecting computers over ordinary
telephone lines. The first first crude standards for message exchange between
different types of computers was probably the protocols developed in the
ARPANET project around 1970 (the forerunner for
Internet). This computer-mediated messaging was based on the computer’s

ability to store data and to forward and receive messages automatically. Thus, a
new medium was born with the almost instant transfer of the telephone coupled
with the asynchronous nature of the letter. It has proved to be a great
combination, and its double nature shows in the fact that email language tends
to be much more informal than the language in old-fashioned letters—it is
almost oral in character.
Up to quite recently, email suffered because of a lack of standardization
regarding address formats, message formats, character encoding, and
mechanisms for attaching files to messages. Even if there were pioneering
standards available quite early, most of the different computer companies had
their own proprietary mail systems, and there were even competing international
standardization schemes. As late as in the early 1990s, the process of
standardization seemed slow. Then the sudden and phenomenal growth in
Internet use rapidly established de facto standards in all these areas. There are
still problems, but the market pressure on the vendors has increased
dramatically, and those who want to survive must converge toward a common
set of solutions fairly rapidly—since the users will flock to the solution that
ensures them the most painless communication. As a result of this, we are
experiencing a dramatic increase in the use of email, and the use of file
attachments has already become a viable alternative to fax and remote printing.
An offshoot of email is computer conferencing—where electronic mail is put in
a storage area where all the conference participants can read it and respond to it.
It then remains in storage for later reference or is deleted at the discretion of the
conference moderator (more about conferencing later).
Telephones and Videophones

In an earlier discussion of the way technology has enhanced communication, we
touched upon both the telephone and video-based communication. The further
development of both has for many years depended on microelectronics and
computer technology. Telephone switching has already become a task for
specialized computers (digital switches), telephones themselves are increasingly
chip-based, cellular phones are crammed with microelectronics, and the whole
transmission system is now in the midst of a change from analog to digital
signaling (ISDN and ATM 5). There is also a movement toward one or a few
digital
5ISDN stands for Integrated Digital Services Network, ATM for Asynchronous Transfer Mode.
ATM is a second-generation standard for digital communication that will allow transfer rates
several orders of magnitude greater than ISDN. ATM is already available on a limited basis in many
countries and is expected to eventually supersede ISDN.
standards for cellular telephones, which means that we will eventually get truly
international networks allowing the owners of standard cellular phones to place
and receive calls from their own sets regardless of which country they are in. In
the high end, there will even be fully global satellite-based services with
handheld receivers about the size of normal cellular phones. Other notable
developments in telephony are voice mail—really only an auditive parallel to
email—and the self-service switchboard, which represents an attempt to
automate switching, access control, and simple direction-giving.
As we have noted earlier, the videophone has been one of the chief recurring
sensations of the twentieth century. The limiting factor has always been the
excessive cost associated with the high-capacity lines needed for the
transmission of live video. Considerable progress has been made in compressing
video signals, however, and the prices of channel capacity have also been
falling at a steady rate, not least because computer-based technology has made it
possible to transmit far more information over existing copper wiring than
anybody thought possible as late as in the 1980s. This means that we can finally
see the day when this turkey will mutate into a bird with a more pronounced
ability to fly. With the present development, high-quality video telephony and
conferencing is bound to become available at affordable prices, even before the
public telephone networks are fully rewired with optical fiber into every home
and organization.
The days when we seriously considered the need versus the cost before
placing an intercontinental telephone are by and large gone, and the same thing
will happen with video. Combined video and computer conferencing—with
simultaneous viewing of screens and exchange of comments and data—will
become cheap enough to allow widespread use. And there are even more exotic
alternatives on the horizon: research has already started on the possibilities for
holographic displays, allowing for three-dimensional representations. Work
done at MIT’s Media Lab (among others) indicates that it will be feasible some
time in the future. If so, it will give an entirely new twist to video conferencing
and make simulated presence almost as good as being there. However, it is
definitely outside the frame of realistic assessment we adopted at the beginning
of this chapter.
System-to-System Communication
Email is really built on the ability of computers to communicate with each other
automatically and without direct human intervention. In this case, the messages
are directly originated by and intended for humans, but the mechanism behind
can also be used for other purposes—such as
communication between applications running on the same machine as well as

communication between systems residing on different computers. It was
pioneered already in the 1950s, and the SAGE computers (the core of the first
American early warning radar system) were linked when they were deployed in
the late 1950s.
Basically, information interchange between application or between systems
can be effected in two different ways: Either you can share the data by common
database access, or you can have the applications or systems send messages to
each other. The common database approach is really the computerized parallel
of the concept of the active file—a unified collection of organized information
for administrative purposes. It is often the natural solution inside an
organization, or at any rate for those parts of an organization that have to work
intimately together. The messaging concept will most likely be the right answer
for communication between organizations and between various parts of a large
organization, where shared databases are not a convenient solution.
Common database access means that information captured or keyed into one
application or task module 6 can be accessed, processed, and presented by other
applications or modules. Data registered through a production control system,
for instance, can thus be immediately available for the sales support system,
keeping sales representatives continually updated on the status of individual
orders. The same data can then be directly utilized by an inventory control
system (ordering replenishments for parts or raw materials that are running
low), and a transport scheduling system (supporting the shipment of finished
products). An order entry system may in its turn supply input data for the
production control system, possibly by way of a separate or integrated system
for production planning and scheduling. An executive information system may
cull data from all of the various systems, presenting a coherent and continuously
updated picture of the main activities in the organization.
This is not yet a description of the common situation in user organizations, 7
but it should gradually become so. Common database
6The boundaries of applications tend to follow the boundaries between task clusters in the
traditional organization—general ledger, for instance, or order entry, inventory control, or payroll.
The different applications in their turn consist of a number of task-oriented modules, usually
organized around specific screens. The modules can in many ways be viewed as applications within
the application, and in less traditional future systems, they will probably be grouped differently (and
quite a number of them may even be eliminated). The distinction is therefore somewhat arbitrary.
7In most organizations the various systems (and few organizations will have systems in all these
areas) will most likely be of different origin, use different databases and formats, and even run on
incompatible machines from different vendors. In some industries (notably, the automobile
industry), integration has become fairly advanced, but, even there, much remains to be done.
access is bound to become a cornerstone in administrative computing. Going

external, however, messaging becomes necessary, and message transmittal in
the form of automatic data transfer irrespective of application or computer type
is not trivial. It is not necessary here to describe the many layers of
communication protocols needed, but I will point out that communication on the
application level requires a considerable amount of standardization, both
between vendors (equipment and software) and between users (data formats). If
an order is to be dispatched automatically by a inventory control system and
directly received and processed by an order entry system in another company
(running on a computer from a different and incompatible vendor), even the
number and definition of the data fields, as well as their size and content must
follow strict standards—otherwise, the receiving computer may mistake an
order number for an amount to be shipped, and the name of an article for a
company address. Such standardized definitions must be established for all the
different types of “documents” required.
This standardization is really what such messaging, or electronic data
interchange (EDI) is all about. Several standards have been created through the
years—national standards, industry standards, and even company-based
standards (large companies are able to dictate their suppliers and strongly
influence their customers)—although none of them have been really
comprehensive. Today, there is a broad effort underway to create a truly
international standard—or, rather, set of standards—for commercial
applications. The work is carried out under the auspices of the United Nations.
The UN/EDIFACT 8 effort aims at establishing a definite international standard
for all main types of documents used in international business, regardless of
industry. It has been underway for a number of years; the first standards have
already emerged, and more will follow in the years to come. There is also work
underway on standardization of drawings and graphics, an important area for
the manufacturing industry. Creating standards involving so many nations,
agencies, and industry associations is a promethean effort, and it is destined to
take a long time. Indeed, it will never be completed—there will always crop up
needs for alterations and new standards. But the main groundwork and the
standardization of the main document groups are well underway, and the
rewards for its accomplishment are so great for all parties involved that it is also
an effort destined for ultimate success.
In addition to this basic standard, we can expect supplemental conventions to
develop at many levels—within organizations and between trading partners and
manufacturers and their suppliers. The net result
8United Nations Electronic Data Interchange for Administration, Commerce, and Transport.
will be a lot less keying in of information, with concomitant savings in labor

hours. The consequences may prove much more profound than simple savings,
however, a subject we will return to in later chapters.
An Escape from Paper?

Information Presentation
In Chapter 6 we discussed the transition from orality to literacy, and the
fundamental changes it wrought. In addition to all this, it also brought us a
physical format for information presentation—the written word is a material
product with both design and packaging, both with an interesting history we
unfortunately cannot detail here. Suffice it to say that after centuries and even
millennia of tablets, scrolls, and various other formats, the sheet and the book
gradually became the preferred solutions. Sheets are easier to manufacture than
scrolls, and books have the advantage of being random access devices—you can
open them on any page.
That sheet of paper, the page, has been with us ever since, and it is today the
principal setting for all kinds of textual information: almost everything has to fit
a page. Even as I write this, thin dotted lines appear from time to time on my
computer display, telling me that I have reached the end of the current page—
yet there is no such thing as a “page” in a computer text file. The creators of my
word processor, however, knew full well that the intended product of my
keyboard efforts would be stacks of pages from the laser printer down the hall,
and therefore provided me with that unobtrusive cue to help me with my
formatting.
In the beginning, computers were not at all slaves of this paper paradigm, but
as soon as printers entered the scene, their output was brought under page
control. Even if it has an interactive nature and can display varied information,
the traditional computer screen is treated largely as “reusable paper”—
information is displayed in an orderly, serial format that resembles a paper-
based presentation as much as possible. For office support systems the ideal has
always been to make the screen look identical to the prospective paper output,
even to the extent that black letters on a white background are preferred—in
spite of the fact that this combination is not necessarily ergonomically
preferable.
The paper paradigm can be traced further in software design, from the on-
screen, visual index cards of archiving systems and simple databases, to the
drawers, folders and document icons presented by the latest in office support
systems. The reasons are twofold: first, most of what we compose on the screen
is finally destined for output on paper, and must therefore be designed to fit the
page format produced by the printer. Second, the software designers all seem to
think that the friendliest computer is the one
that provides the user with an emulation of his or her paper-based past, and thus
attunes itself to the user’s established mental set for office work. How could it
be otherwise? As both Ong (1982) and Havelock (1986) note, it takes an oral
culture generations to pass from an oral form of expression (with formulaic
style for mnemonic purposes) to a truly native, literate (chirographic, written)
style. We have more than 5000 years of chirographic tradition behind us, and
with only a few decades of computer experience, we are doomed to mimic the
past—we need time to adjust, to enter into a working relationship with the new
technology, to iterate our way toward more computerate manners. As the
systems become more powerful, however, and (not least) the screens bigger and
more comfortable to read from, things are bound to change.
The evolvement away from the paper paradigm can be traced in the
development of user interfaces. We have now largely left the green-phospor
character-based screens, and are immersed in paper-mimicking graphical
interfaces. A growing number of features point even further, though. Some of
the controls are distinctly nonpaper, for instance, buttons and sliding controls.
Even if they are borrowed from well-known mechanical and electrical
appliances and thus do not represent “native” computer innovations, they show
us that the computer can do more than simulate pen and paper.
Hypermedia
More exciting still are the developments in the direction of object orientation,
hypermedia, and “hot links” between applications. Such a higher-order object
consisting of chunks of text, spreadsheets, and graphics is nevertheless little
more than a compound document, a collection of pages, and it will still print
neatly. The next step is to have a document that merely contains pointers to
where the different information chunks are stored. When any of the chunks is
modified, this modification will then also automatically apply to all the
compound documents that contain pointers to it. Such documents can still be
printed without problems, but version control now becomes more important.
However, such links can also be made conditional—like an electronic
footnote. A word, a picture, or a table can be marked as a button, and when you
activate that button, the link will come to life and retrieve the supplementary
information. This is the basic idea of “hypermedia,” an idea that has also
inspired the principles behind the World Wide Web. A “document” with such
buttons would be harder to print, but it could still be done, if it were acceptable
to convert the buttons to footnote markers and the supplementary information to
footnotes.
But hypermedia (and the WWW) goes further than this—it can easily be
nested: The supplementary information can contain its own buttons, pointing to
even more information, where you will find still more buttons, and so on. This
is what hypermedia is really about—a three-dimensional information structure
where you can establish both factual (providing explanations or supplementary
information) and associative links, building an information structure that will be
totally unprintable. Some of the information chunks may consist of live video or
sound, which is even less representable on paper.
Hypermedia links can have many properties. They can be bidirectional,
meaning that you can not only point to a reference like in a footnote, but you
can also, from the reference, be able to trace all the works referring to it. The
links can also have different granularities: A link can point to a book, paragraph,
or word; a picture or a section of that picture (possibly even a single pixel); a
passage in a piece of music (or even a single beat); and so on. There can also be
different types of links, with different access levels, and you may be able to
establish filters, showing you only the types of links you are interested in. A
distributed hypermedia system could also have “document sensors” that
constantly monitor documents or parts of documents, carrying out certain
actions when triggered—for instance, notifying the original author if someone
else updated a piece of information.
Advanced hypermedia systems should make it possible to establish “living”
information bases, where revisions are automatically accessible to all links, and
where the revisions themselves can trigger messages to those who have referred
to that information. Full version control will be possible, and it should also be
possible to backtrack revisions and reconstruct the original document. If write-
only storage media are used, such a system should make it possible to operate
without paper even in highly formal environments, such as government offices,
because of the audit trail left in the system.
The idea of a hypermedia system was first presented by Vannevar Bush in his
famous 1945 article about the Memex (Bush 1945). However, since the
computer was yet in embryo, he was thinking about a microfilm system. The
hypermedia concept was brought into the computer world by Theodor H.
Nelson, whose brainchild, the Xanadu hypermedia system, was designed with
the idealistic goal in mind of providing an engine for a world-wide hypertext
publishing network (Nelson 1988). The network should store a particular piece
of information in only one location, while allowing it to be incorporated in any
“document” on any server in the network via links. Despite the somewhat
grandiose aim, Xanadu became a real piece of software, with 30 years of
development work behind it. It never succeeded in the marketplace, but it was
well-known
and admired in the inner circles of the software community and can
undoubtedly be nominated the mother of all hypermedia systems.
Today, the baton has been handed to the Internet’s World Wide Web. The
Web is still primitive in principle compared to Xanadu, but the size of the
amalgamated information bases you find there is already staggering, and is
growing every day. It represents the first real step down a very interesting road.
Multimedia and Animation

Multimedia, for its part, is only possible when we leave the paper paradigm. To
really come into its own, however, it, too, will depend on the advances in
display technology—even without paper, there will still be so much text around
that affordable screens with good reading comfort will be necessary. If we think
about the possibilities offered by holographic or other three-dimensional
displays, they are truly staggering.
Holographic displays would also add a new realism to the three-dimensional
(3D) CAD and modeling systems already in use in engineering, architecture,
design, and science. These systems are also leaving the paper paradigm behind.
While they once started out as drawing tools mimicking the original paper-
based drawing process, they are now increasingly able to generate full-bodied
three-dimensional models of the drawn objects—models that can be rotated,
exploded, and enlarged to reveal detail. If the object is a house, the system
allows you to do a walk-through on screen, studying how the light flows
through the windows and the effects of different forms of interior lighting
schemes and color options. A model of a processing plant can reveal any
conflict between the positions of various equipment. The models on screen can
now also be reproduced as physical models in what almost amounts to an
“object printer,” where a thin beam of UV light solidifies layer upon layer of
polymer.
Computer models can in turn be animated, to simulate, for instance,
production processes, thereby bringing the design process one step further.
Numerical simulation is nothing new, of course: even old ENIAC was involved
in that. But when we leave the printouts behind and couple simulation with
animated models showing the result of the simulations in real time, our
comprehension of the outcomes is brought to an entirely new, more
sophisticated level. Instead of struggling with reams of data, taxing our working
memory to the limits and beyond trying to visualize the effects, animations can
bring the vast capacity of our visual system into play, improving our
understanding many times over.
The windowing interface is thus a product with a Janus face, straddling the
fence between the old and the new. But there is no doubt which side is
growing fastest, and there is a large number of efforts underway to develop it

further. The Internet marketplace is perhaps the strongest force today, but
interesting work is also being done on totally new user interfaces in laboratories
such as Xerox PARC in Palo Alto—the birthplace of the original graphical,
windowing user interface (Clarkson 1991). When really large, affordable
screens with good reading comfort become available, the movement away from
the paper paradigm will experience a surge in development and demand.
Structuring Information
We noted above that the paper paradigm can be traced in the way information is
represented on the screen. But the matter runs deeper than that. If we look at
how information has been structured both in storage and processing, we will
find very strong influences from the administrative practices developed during
the growth of modern organizations since the middle of the nineteenth century.
The protocol and the file were the established hallmarks of administration, and
it was no wonder that the budding systems analysts and programmers were
caught up in the reigning paradigm. Just note the vocabulary: computers still
store their data in “files,” the collection of data belonging to one entity is called
a “record,” and the individual pieces of information are located in “fields.”
Information is located through “indexed” keys.
First of all, computer files came to mimic the paper-based files and punch
card equipment they replaced, and they were designed and used more or less as
electronic filing cabinets. Information items were put in fields collected in
records, which were direct descendants of the file card. One located a particular
record through a key (normally, a name, address, or some kind of unique
identification number), just as in a manual file. The database was really nothing
more than an elaboration of this scheme, allowing several indexes as well as
pointers relating records through connecting fields. Early databases only
allowed predetermined pointers that were part of the application program code,
and were thus extremely inflexible. In effect, they even required all the report
formats one would ever want to be part of the program specification—a
hopeless task in a changing business environment. To establish new relations
between data elements, you would have to modify the program code, and the
old pointers and definitions could well be hidden within algorithms deep down
in the program. That is why any functional change (including the creation of
new reports) used to cost so much and carry the risk of creating new errors and
inconsistencies. Later databases became more flexible, but at the price of more
complex software and a much heavier work load for the computer.
As electronic filing cabinets, the computers fitted quite nicely into the
existing work flow. And, as programs evolved to support existing routines, it
was just natural that they, too, were structured basically in the same way as
traditional office work. The main determinant for this structure is the way large
tasks decomposed into the basic tasks people actually perform when they work.
As we have noted earlier, the classical model for administration was molded on
the mass-producing factory and reinforced by the characteristics of large, paper-
based files. It prescribed functional specialization, fixed procedures, and
detailed rules and regulations. Routine tasks were decomposed into separate
steps or operations, similar to an assembly line, and often described in detailed
manuals. Job design was procedure oriented: It specified who would do what to
which piece of information and in which order.
The Functional Approach

Computer programs, following the well-trodden paths of 150 years of
administrative work, were also viewed as procedures, decomposed into single
steps of operation on the various data files. Systems people call this method of
decomposition functional (Coad and Yourdon 1991) or algorithmic (Booch
1991), and the resulting design and programs structured. Another vital
characteristic of this approach is the distinction between program and data. Data
are viewed as given, something you analyze to design a suitable database. The
program contains the operations you want to perform on the data, and to design
the program you analyze functions. Or, to use the language of systems theorists;
traditional, structural analysis is based on the assumption that reality is
composed of entities and their states on the one side, and functions on the other.
Functional or algorithmic decomposition and structured programming as a
method was also reinforced by the very nature of the computers themselves,
since they all complied with the basic scheme devised by John von Neumann in
his famous “Report on the EDVAC” in 1945. 9 A von Neumann machine (as all
computers following his basic principles are called) is characterized by serial
processing—it has one central processor, fetching and executing one instruction
at a time. Almost all computers
9EDVAC (Electronic Discrete Variable Computer) was the direct descendant of ENIAC and was
also conceived as a project for the Army Ordnance Department—the same organization that
financed the construction of ENIAC. The report was the first outline of a design for a stored-
program computer, but the project had many delays, and EDVAC was not completed until 1952.
to date have been von Neumann machines, even if computers based on parallel
processing are now becoming more common.
Because systems based on functional decomposition and structured design
are built like trees of specialized routines, there is a great degree of
interdependence between different parts of the system. Changes in one
subroutine may require changes higher up in the system, which in their turn
have consequences for many other subroutines. As such systems get larger, they
therefore become very complex, and the interdependencies very difficult to keep
track of—which means they also get increasingly difficult to change. However,
needs always change over time and force modifications of the system. Since the
cost for altering the main structure of a system based on functional
decomposition is usually very high (it is often tantamount to developing a new
system), changes tend to consist of ad hoc patches and additions crisscrossing
the original logical structure. This further increases complexity and makes the
system less and less comprehensible, until one reaches a point where it becomes
unstable, because its procedural labyrinths are no longer fathomable. Changes
may suddenly have unpredictable consequences in unexpected parts of the
system, and one is faced with the choice of using it as it is without further
changes, or discarding it completely to build a new one.
The Object-Oriented Approach

The complexity and inherent inflexibility of systems built on functionally
decomposed designs made reflective systems people look for ways to simplify
things, and from their work grew the object-oriented approach. It, too, has
interesting parallels in the organization domain—such as the feudal type state
and the divisionalized enterprise. The underlying principles of the object-
oriented approach to systems analysis and design are exactly the same as those
applied to achieve maximum information economy in organization, and which
we have analyzed earlier: the reduction of complexity by way of modularization
and encapsulation.
The basic metaphor for the object-oriented model is the modularity found in
human cognition and natural hierarchies—for instance, the hierarchy organism-
organ-cell (Booch 1991). A cell manages its own internal processes, and the
fact that all the cells in the organism do so in parallel relieves the central
coordinating organ (the brain in an animal) of an impossible burden of
coordination. Indeed, it also makes feasible organisms, such as plants, that are
without any central coordination at all. For organisms that do have central
coordination, such as mammals, the brain does not exert any direct control over
intracellular processes;
rather, it controls cell behavior by broadcasting chemical messages that trigger

local processes in the appropriate cells.
This is basically the same concept that lay behind Sloan’s and du Pont’s
management of GM’s divisions by sales targets and budgets—leaving the
internal workings and initiatives in the divisions for the division management.
The pure organizational example would be an organization consisting of self-
managed teams. It is also the principle behind the capitalist, free-market
economy, where independent companies (“objects”) chart their own courses and
cooperate through the exchange of “messages”—contracts, money, goods, and
services. As history has shown, this is a superior coordination method for very
complex organizations, such as large national states.
If we return to the organismic metaphor, a cell—for instance in the liver—is
an example of an object. It would belong to the class “liver cells.” The liver
itself as a complete organ would constitute a higher-order object, containing
both liver cells and other objects (e.g. the gall bladder and blood vessels), and
belong to the class “livers,” a subclass of the main class “organs.” The organism
would be our problem domain if we were only interested in its internal
composition. If the area for study included the organism’s interaction with its
environment, including other organisms, the complete organism would in itself
be an object of a still higher order than the liver, and a member of its own
class—for instance (if it was a mouse), the class “mice.”
Like an actual cell or team, an object (in the data-processing sense) has an
internal structure—it stores its own data, or states, as well as the rules pertaining
to those data and their representation. Its internal structure is thus encapsulated
and hidden from the environment—the environment only “sees” the object as
the messages it can receive and send and the behavior it can display. The class
defines the properties that are common for all the objects in it, and the objects in
their turn “inherit” them. Changes in the class description therefore instantly
apply to all the relevant objects. An object communicates with other objects
through messages, and the receiving object “knows” what to do when it receives
a message. No master program is thus needed to direct the detailed processes
within each object, and any object is interchangeable with any other object
having the same “message interface.”
The focus of object-oriented decomposition, then, is not the functions and
procedures found in the problem domain, but the items or entities of interest for
the system in question (Booch 1991, Coad and Yourdon 1991). Since those
entities are generally much more stable than particular procedures, object-
oriented systems are less susceptible to need major changes as business
requirements change. The changes that are required will also be easier to
implement. Object-oriented systems are moreover
able to handle greater complexity than functional systems, through their use of
encapsulation and hierarchies of objects and classes. They will also, in most
cases, offer better abstract representations of reality than functional systems.
Object orientation represents a break with the old link between the paper
paradigm, functional decomposition, structured systems, and functional
organization. It shakes information structuring loose from the file cabinet and
the procedure, and returns it to the main avenues of human cognition. For,
objects and object classes are indeed the pillars of human cognition: we cannot
even comprehend the world until we have established notions of object classes
(e.g. “houses”), their properties and their relationships to other object classes.
The object-oriented approach is therefore much better suited to analyze the
domains of human work and cognition and to create systems that are compatible
with human thinking and human work.
215
9 The IT-Based Preconditions
“Memory is like all other human powers, with which no man can
be satisfied who measures them by what he can conceive, or by what
he can desire.”
Samuel Johnson, The Idler, 1758–60
We have now established a reasonable understanding of the central properties of

information technology: Computer-based systems process information, they
store information, and they communicate. And, the key to the power of
computers, to all of their capabilities, is their programmability—the possibility
to have immensely complex sets of logical operations executed automatically.
Both hardware and software have undergone rapid developments, and new
classes of organizational tools have been developed, as we have just discussed
in Chapter 8.
We must now try to assess how these new tools enhance our own capabilities
over and beyond the contributions from earlier technology, since this is the key
to understanding possible extensions to the space of constructible organizations.
This follows from our analysis in Chapter 2 and the conclusion that
organizations are constructed: Only changes in the abilities or available options
of the individual organization members can give rise to fundamental new
possibilities for organizational design.
In this chapter, we shall only assess the basic enhancements. Their
organizational ramifications are too diverse, and will have to wait for a detailed
analysis in the four chapters in Part IV. The attempt at synthesis will then follow
in Part V.
Memory
The externalization of memory provided by the art of writing had profound
consequences for our administrative capability, our problem-solving ability, and
our capacity for knowledge accumulation. All of these
improvements were not only important for our ability to build organizations;
practically all our advances in knowledge and technology—that is, the very
foundation for our present material prosperity—rest squarely on the invention
and exploitation of writing.
However, except for a certain refinement of the material means for filing and
archiving (such as the invention of more elaborate classification schemes—e.g.
library systems—and, eventually, the punch card), very little happened to the
basic efficiency of information management up through the centuries—and our
biological memory has hardly improved noticeably either. The usability of
stored information (externalized memory) has also been severely constrained by
several factors: the need for physical access to the storage media (files, books,
etc.), the large amount of work involved in the search and retrieval of
information items, and the slowness of the human input/output process (reading
and writing).
The computer is now changing this—we are in the middle of a revolution in
memory tools that will still roll with considerable speed for several decades, and
perhaps bring changes as fundamental as those brought by the introduction of
writing itself. This new revolution is built on the fundamental improvements it
offers in storage economy and accessibility.
We can already hoard immense amounts of information almost anywhere,
and the prices are swiftly falling to a point where cost will be irrelevant as long
as the information is even remotely useful. This will allow us to accumulate all
kinds of information produced by our own computer-based information
systems, and have it instantly accessible on-line for reference, monitoring, and
analysis. As the price of the storage medium falls, the limiting factor will in the
end be the price of the information itself—either the price of purchasing it, as in
the case of commercially available information, or the cost of producing or
capturing it (and then organizing it for storage and retrieval), as with
information indigenous to the organization or available through partnerships or
other business relations.
Access to this information has moreover been separated from physical
access; an extremely large number of people can access the same piece of
information simultaneously, and it possible to locate, retrieve, sort and compare
information with great speed and accuracy. Access is of course not as swift as
the recall of something we remember clearly, and the ripples of association will
also still flow faster in the brain. But for the vast volumes of information we
cannot even hope to remember, and, even more significantly, all the information
we have never heard about at all before our new tools find it for us, we will
have an access that is many orders of magnitude faster and more exhaustive
than before. Less time will be spent in search activities, and the yield of relevant
information will be
9 The IT-Based Preconditions 217
much greater. It should allow our own information processing to draw on a

much larger set of facts and viewpoints than before. In fact, the limiting factor
will be our own innate capacity to absorb and digest the information.
This new memory technology is not like our own memory, it is an external
memory just like the traditional filing cabinet—but it is also much more than a
filing cabinet, or a bookshelf, or a library—so much more that it deserves a new
name. To keep in tune with the trend-setting term “artificial intelligence” (AI),
we may call it artificial memory (AM). This term is not entirely new, Simon
(1976) refers to it as already in use—but it has up to now been used to denote all
kinds of records outside human memory, chiefly libraries and paper-based files.
However, since it has been used so sparingly and since it matches the now
thoroughly established term “artificial intelligence” so well, it seems easier and
wiser to change its content rather than to coin an entirely new term. It will still
fit Simon’s (1981) definition of the trichotomy natural-synthetic-artificial.
Of course, not all information of interest to us is available in digital form.
Much is still found in books and journals, in looseleaf binders, archives, even
piles on desks and shelves. The all-encompassing information bases we have
been alluding to remain a potential, not a fact. Nevertheless, the information
used and produced in the organizations of the world is rapidly becoming
digitized, as computer-based systems are introduced to take over or support one
after the other of their work processes, and the structured database is becoming
the new linchpin of organizational coordination.
This is perhaps the core of the greatest revolution of digital storage—the
tremendous boost it gives to implicit coordination. No longer is the effect of this
extremely effective coordinating mechanism limited by the physical access
restrictions and puny volume of traffic associated with paper-based files—in the
comparatively near future, hundreds of thousands of simultaneous users all over
the globe can be served by a single database, and thereby achieve strong, instant
coordination in selected aspects of their work. In a more extended perspective,
volume restrictions will for most practical purposes cease to exist.
We shall explore this new revolution in more detail later (in Chapter 13). Let
us finish for now by noting that the number and quality of available information
bases are also rapidly growing, thanks to the standardizing effects of Internet.
Whereas the various database services up till quite recently displayed ample
variation in log-on procedures and command languages (and enjoyed limited
commercial success), Internet has in a few short years completely taken over as
a gateway, and the www browser has established the standard for information
retrieval and presentation. The chief obstacle today is a reliable and simple
mechanism
for pay-as-you-go information retrieval, but it will soon be cleared away. New
software will also come to assist in the process of finding what you seek—both
search engines that “learn” from your searches as well as “agents” who crawl
the net on your behalf, returning periodically with their catch, have been
proposed so far (Stein 1991, Negroponte 1995). Interesting products are bound
to come along here, even though it usually takes longer to perfect this kind of
AI-based software than the proponents initially believe.
This is not so important for our purpose, however, since there is little doubt
that it will only be a matter of time before most of the information we need in
the course of our daily work will be accessible through our workstations, with
very capable search-and-retrieval tools available—probably even tools that will
be able to continuously (and in the background) build and maintain links and
search profiles from the factual and associative jumps we make during our
normal search activities. Such tools will apply both to external and intra-
organizational information bases.
When that happens, our active information repository will become vastly
larger than it is today. Comfortable, speedy access is critical for information to
be used, and scarceness of time and economic resources will always tend to
severely limit our search. A lawyer I once met summed it all up in what he
called the law of arm’s length: “You know, 99% of the time, you make do with
the information you can reach out and grab without leaving your desk chair.”
Which is, of course, the reason why people keep private copies of central files,
why they buy books that are available in a library three blocks away, why they
make copies of everything that may come in handy at a later stage. It also means
that 99% of the time we make do with the information we remember that we
have and know where to find, and even the information in our offices is liable to
get lost within a meter or two of our desks, as long as we do not index anything
and everything. IT will place a vastly larger volume of information within
“arm’s length”—which is what much of the current excitement about the
Internet and the so-called “cyberspace” is about.
Processing and Capacity for Work

Before writing, our only available strategy for alleviating the limititations of our
processing capacity was simplification—singling out a few variables for
concern and forgetting the rest. Writing greatly enhanced our overall processing
capacity by providing a second-tier working memory with storage of
intermediate results. We accumulated knowledge, kept records, and expanded
our vocabulary to allow more precise expressions. We achieved a vastly
superior understanding of nature and of causal
relationships. Writing also made it much more feasible to decompose complex

tasks and distribute them both over time and between many persons. It greatly
improved the possibilities for coordination over distance. The telegraph and
telephone further reduced the significance of geography. Finally, the
development of mechanical automation made it possible for us to multiply our
capacity for physical labor. What do computers provide over and above this?
First of all, we should note that the time we need to think through a problem,
to mull over likely and unlikely consequences, to weigh the preferences of
possible solutions is still an internal process of our minds. As such, it is no more
augmented by information technology than by writing. We are still also serial
processors, and can only concentrate on one problem at a time, and information
technology does not change this any more than previous inventions have done.
No matter how advanced the workstation on your desk, when the telephone
rings, you will still lose the thread of your work, and the conversation will
effectively block any other serious mental activity. Combined with the
restrictions of our working memory, this one-trackedness of our mind will still
put a limit to the number of variables we can handle simultaneously.
What the computer does offer is the opportunity to unload some of the
information processing itself. Further, computer-based support tools may make
task switching even easier than written notes and records could, and increase
precision by organizing information better. At the personal level a properly
equipped workstation can keep parallel work processes going—such as sending
and receiving faxes in the background, sending and receiving voice messages,
calculating large spreadsheets—and even do database selections and sorts while
a person works on other tasks. There is also the prospect of “intelligent agents”
that can take care of more complex tasks, but, for the present, such agents
represent little more than a gimmick. What the distant future will bring is of
course impossible to foretell, but in my lifetime I do not expect to see computer
systems that can effectively emulate humans on an overall basis—research on
neural networks and artificial intelligence notwithstanding. Computers and
humans are simply too different to make that an early success. We should
reserve for our own minds the tasks where we excel, and exploit computers for
the tasks they master—and where we can profit from their great speed, accuracy
and untiring work. The computer’s potential for automation and processing of
quantitative information is thus in my view much more important than its role as
an office assistant.
However, if we look at the whole process involved in reaching decisions in
organizations, including the collection of relevant information and consultation
with others, considerable improvements are possible, mainly
in the area of information collection and communication. McKersie and Walton

quote an example from a high-tech organization (1991, p. 252):
During my last weekend in Washington, an important issue arose late Friday that
required an official written agency position Monday morning. A few phone calls
locked in the key experts (five different states) for an electronic brainstorming
session on Saturday. I got initial thoughts from everyone on electronic mail Friday
night (ideas were iterated once or twice) as well as access to information and
graphics from local databases with comments and proposed rewrites or
reorganizations with appropriate rationale. Three iterations were completed by
5:00 P.M. and a draft was electronically forwarded to three senior managers at their
homes for approval. After incorporating their revisions, the position paper was
approved and printed for an 8:00 A.M. meeting Monday morning with the head of
the agency.
There is no doubt that the position paper could have been produced by Monday
morning even without computers, relying on telephone and possibly telex or
fax. But it is equally clear that electronic mail and remotely accessible databases
made it much easier to produce the required document, and with a better result.
Further developments in software (discussed in more detail in Chapter 11) will
help to speed up the decision-making process even more. Just like writing,
computer-based systems thus provide a set of tools that make it possible for us
to exploit our innate processing power to a larger degree than before.
The Quantitative Revolution

However, the main contribution of information technology towards mastering
complexity is without doubt the way it allows us to manipulate quantitative
information—information that can be expressed in numbers and categories. This
sounds rather narrow at first, but when you look into it, you will find a vast
array of applications where the computer has greatly enhanced our ability to
handle complex tasks. The computer’s ability to handle numbers, and to present
them graphically on the screen, has meant a revolution that is, in my opinion, at
least as great as the original contribution of written numerals. The reason is
twofold.
First, even if it was possible to develop advanced theory in mathematics,
physics, and engineering without computers, much of that knowledge was
simply impossible to use in practice because of the enormous burden of
calculation. Cheap and powerful computers now allow almost any scientist and
engineer to routinely carry out calculations that were simply unthinkable 50
years ago. For instance, finite element analysis is now a practical, everyday tool
of engineering—not merely an exotic, theoretical possibility. Chaos theory was
not even discovered before the computer
was available, since the nature of the regularities in chaotic systems were too
complex to comprehend without it (Gleick 1988).
Second, the computer is capable of tracking causal relationships—and not
only between a few variables (as in a simple spreadsheet model), but also
between exceedingly large numbers of them (as in a weather model for a 10-day
forecast). The computer works as an automatic preprocessor, combining a large
number of predefined causal relationships into a few aggregate ones that we can
comprehend and manipulate in our own minds. It can even do so in real time, as
in trading systems, process control systems and flight and weapons control
systems on modern military aircraft—which let the pilot control and coordinate
tasks that would have demanded a sizable crew just a decade or two ago.
The advancement of artificial intelligence can also contribute. Heller (1991)
describes, for instance, a system for routing trucks developed by Carnegie
Mellon University and DEC. Incorporating rules developed by interviewing
experts in the trucking company, it has allowed the same experts to reduce the
company’s continuous-mileage transport costs by 10%—not because it “knows”
more than they do, but because the system is able to take all the rules into
account every time, even when time is scarce. AI also has the potential to
become an important tool for real-time monitoring of complex technical
installations, where it is essential to maintain a continuous overview of main
events.
Automation
As we have already noted, automation is our only way to achieve the capability
of working on many tasks in parallel. Simple automation can be seen as an
enhancement of the capacities of the individual operator only, more
sophisticated automation replacing scores of workers of many different trades
must be seen as a tool on the organizational level. We shall return to this last
(and most important) aspect of automation in Chapter 12. At this point, we shall
only discuss the main differences in principle between mechanical and
computer-based automation.
In the classical machine the “program” governing its movements is contained
in the physical shape of its different parts. Since one part can only have one
particular shape, its information content is low, and to make a machine with a
relatively high information content (able to do complex operations or different
types of operations), one must use a very large number of parts, making the
machine expensive to manufacture and less reliable (one of the main concerns
of mechanical engineers is always to reduce the number of parts as far as
possible). Mechanical automation has nevertheless been developed to a very
high level of sophistication,
where even highly delicate operations, such as the manufacture of light bulbs,
have been fully automated (Bright 1985).
Another area of importance to automation concerns linking sensor input to
operations. Some such links are simple to establish by mechanical means; the
best known is probably the classic thermostat with a bimetallic switch. But,
even in this field, the complexity achievable without the help of computers is
limited, and despite much ingenuity, automation could not proceed beyond
certain limits—witness the control rooms of precomputer power stations or
factories in process industries, where a considerable number of operators would
walk around, all the time reading dials, turning wheels, pulling levers, and
flipping switches.
Now, to automate by computers, we still have to plan and describe in
painstaking detail every action to be carried out. The material difference is that
the information, instead of being embedded in the physical shapes of parts, is
simply lodged in software and hardware logic. What was earlier impossible to
change or required new parts or rebuilding can now be done by changing
parameters or code lines in software—a much simpler and more economical
alternative. This difference translates into an enormous divergence in the level
of complexity we can operate with and the speed by which the embedded
information is processed.
As the processing power of the controlling computers is increased, the scope
for automation in production and control is therefore drastically widened, and
no final limits can be seen. Computer-controlled systems can collect and
analyze very diverse and sophisticated signals from a broad variety of sensors,
can direct all kinds of machinery, and can be equipped with an array of
responses covering almost any conceivable eventuality—also error conditions
and accidents. Any process that can be precisely defined can in principle be
automated. This is not to say that we do not experience limits today, but it is
very difficult to determine which ones are fundamental and related to basic
constraints in the nature of computers and computer-based systems, and which
ones are simply due to the present immaturity of our computers, our software
and our theories of computers and their use. I suspect that very few of the
constraints we have experienced so far are of the fundamental kind.
Computer-based systems are also very reliable when we consider their
enormous complexity. This may sound strange in the ears of the average
computer user, regularly frustrated by inexplicable error conditions and just as
indecipherable error messages. Considering the number of discrete electronic
components contained on the chips in an average PC, however, and the number
of code lines in the software employed, we might be more surprised by the fact
that it works at all than by the relatively few errors that occur. If we count each
separate component on the processor chip, the memory chips, and all the other
chips for a separate part (which it
really is, even if it is miniaturized beyond normal comprehension), and do

likewise for every line of source code for the programs normally run on the
average business PC, that PC consists of several hundred millions parts, and
will soon be in the billions. How reliable, for instance, would a machine with
300 million mechanical parts be? A modern airplane, such as the Boeing 777,
has approximately 3 million parts (excluding computer components!)—and
needs extensive, regular servicing to operate within acceptable safety limits.
Communication
All the great breakthroughs in the history of human communication—the
written word, printing, the telegraph, the telephone, radio, and television—were
developments of the pre-computer world. Even the telefax is an old invention
from the nineteenth century, although microelectronics was needed to make it
cheap enough and provide the necessary document quality to really make it
popular. Computer technology has so far not provided breakthroughs of a
similar magnitude, and its ability to do so in the near future is in my view
generally overestimated—despite the advent of several interesting technologies:
electronic mail and conferencing, cellular phones, videophones, and, of course,
the modernized fax.
Our Very Own I/O Bottlenecks

The reason for this is simply that communication is not only a matter of
transporting symbols from one person to another. There is also the problem
actually absorbing incoming information, and disseminating the information we
intend for others (the human input/output, so to speak)—a problem that
represents a far more formidable problem than increasing the bandwidth of
long-distance information transfer. Actually, there is even a third important
aspect of communication, which we shall not discuss here: the question of how
well meaning survives the encoding/decoding processes involved in human
communication.
Although bandwidth is an interesting issue, then, it is definitely not the only
issue. Electronic communication as well as digital storage and computer assisted
search and retrieval certainly allow us access to vast information resources. But
even if we can unearth mountains of relevant information, how can we absorb it
all and really use it? Information must still be read off the screen or off paper
printed out by the system. The cry is in fact already going up in corporations
worldwide, from overloaded managers and professionals: What we need is not
more information, but the key information. We do not even find time to absorb
the day-to-day
business data that is already there. As Long remarks (1987, p. 45), “By
providing a manager with the capacity to call up more and more analyses, there
will be an implicit pressure for him/her to do so.”
Simon (1976) reflects on a new computer system installed by the U.S. State
Department for receiving the 15 million words received per month from its 278
diplomatic missions throughout the world, and with the ability to print out 1200
lines per minute (the old teletypes could only manage 100 words per minute).
He remarks wryly (1976, p. 284): “A touching faith in more water as an antidote
to drowning! Let us hope that foreign ministers will not feel themselves obliged
to process those 1200 lines per minute just because they are there.”
The sad fact is that our innate capacity for information absorption remains
the same as before, as does our capacity for disseminating information. Our
eyes, ears, and mouths are the same as our forefathers’, and even if some
people from time to time speculate about the possibilities of interfacing
computers directly to the brain, we can safely rule that out as a useful option in
the foreseeable future.
In particular, the computer has not done very much to improve on our ability
to express ourselves. It has made it possible to compose text somewhat faster,
but that is about all. Of course, we can now produce much better looking
material than before; our presentations can be studded with graphics and nice
fonts. But it still takes the same old time and effort to present the result of our
thought processes to others. Where the computer can help is when it can
concentrate or transform information in such a way that it speeds our
perception. However, both the challenge and the remedy vary according to the
nature of the information—that is, if it is verbal, pictorial, or numerical.
Verbal and Pictorial Information

Information embedded in text or pictures is intrinsically resistant to automatic
concentration. Text can be condensed by rewriting and the creation of
summaries, but both are labor-intensive tasks that will require human processing
in the foreseeable future. For film, still pictures and sound the possibilities are
likewise meager. The only significant advantage computers offer is the fact that
computerized searches should return information with a higher content of
relevant material than we can obtain through manual searches. That is, even if it
will not help us to absorb more information, the information we ingest should
be more relevant.
Apart from the improvement provided by the basic capabilities of search
engines, work is also being done on other kinds of tools intended to concentrate
the information presented even more. Associative searches and hypermedia
links have been mentioned, and experiments are also
made on structuring tools (Winograd 1988) and programmable filters for

electronic mail and conference contributions (Robinson 1991).
In addition to this, new ways of information presentation and representation
should make it possible to get a somewhat better overview of complex textual
and pictorial information, especially with the advent of larger screens. However,
our absorption of this kind of information cannot be improved dramatically—at
the very most, we are probably talking about a doubling, not orders of
magnitude.
Numerical Information
Numerical information is quite another matter, however. Since numbers very
often lend themselves to graphical representations, we can tap into the very
powerful visual-processing capacity of the brain. There is no doubt, for
instance, that a pie, column, or line chart conveys information much more
quickly than the tables they are based on. Three-dimensional charts add even
more information (if used correctly).
Graphical representations are nothing new—scientists and economists plotted
graphs long before the advent of computers. What is new is the speed and ease
with which the conversion can now happen, and the forms the graphs may now
take. Graphs used to take a long time to produce, even simple ones, and only the
most important or complex information was the subject for such VIP treatment.
Today, with a modern spreadsheet or statistical package, graphs are almost a
free lunch to be had once your data are registered. Software packages for
administrative purposes also increasingly have capabilities graphical output, and
spreadsheets steadily improve their capacity for extracting data from other
applications for further analysis. Indeed, spreadsheets are now evolving into all-
purpose tools for reports and analysis.
A quite different example is found in the new naval navigation systems, in
which the combination of satellite navigation and electronic maps lets the
helmsman follow the ship’s position continuously on a screen showing both the
map and a representation of the ship. The small ship on the screen moves as the
ship moves, and the map rolls continuously in the direction of movement. For
fast crafts operating in narrow waters, such a system is much safer than
traditional navigational aids.
The most sophisticated visual representation systems today are found in
engineering and scientific data processing, where information is presented not
only as static graphs, but also as animations. Especially impressive are
simulations based on numerical models—be they of new airplanes, waves
generated by projected boat hulls, car suspensions, cloud development, or
cosmological events. Animation in particular can concentrate numerical
information to a very high degree and present us with clean and comprehensible
representations of enormously complex data. Such tools have up to now been
very expensive because of the large computational power required. But, because
of the ever decreasing cost of raw computing power, this fascinating class of
tools is now rapidly becoming available for almost any purpose.
Animated simulation is in my view a branch of software with a great future.
It will not only help in designing physical objects (e.g. factory production lines,
cars, or houses) but also aid in all kinds of data analysis. Even social science
survey data can conceivably be animated, with moving planes and shapes
visualizing the mapping of multivariate distributions and correlations. In
business, animation should be able to provide very interesting tools for
analyzing and monitoring key variables (budget and real)—in production, sales,
and accounting—and, not least, in combinations of these areas. I believe such
tools will be applied in all areas where there is a need to analyze or monitor
complex numerical relationships. The managers of a retail chain, for instance,
could get sales for various product groups or products presented as animated
columns on a map with all their locations—one column for each location. By
presenting 20 days per second, the whole year could be played through in 15-20
seconds. Interesting overall patterns—such as seasonal variations and
geographical variations—could be spotted at once, and greater resolution
applied to the graphics for more detailed analysis.
The New Channels

Although our innate bottlenecks for information absorption and dissemination
largely remain in place, the channels for information transfer have seen
significant development on all levels—from the physical (from copper to fiber,
from earthbound radio to satellites) via basic representation (from analogue to
digital) and bandwidth (including multiplexing) to presentation (application
level). What are the implications?
Electronic Mail and Conferencing

The most touted aspect of computer-mediated communication is electronic mail
(email). There has been (and still is) much excitement over this new medium,
and the growth of the Internet in particular has raised the spirits of many
journalists and salespersons to exuberant levels and fostered visions of a world
of unrestrained communication. Electronic mail is also much used as an
entrance point to computer use for managers and is therefore often somewhat
oversold as a productivity tool.
What electronic mail provides is simply the ability to transmit written

material instantaneously and have it stored for later presentation in case the
recipient is not present. Good email systems also make it very convenient to
answer, by automatically applying the address of the sender to the reply, posting
a reference to the original message, and so on. At its best, therefore, email
functions as something of a cross between letters/memos and telephone
conversations—it has the speed of the telephone but does not require the sender
and the receiver to be simultaneously available at their respective terminals. It
can therefore be very efficient for people who are away from their desks a lot
(especially frequent travellers) and can significantly reduce the number of
unanswered telephone calls. It is particularly useful for communication across
time zones—especially for intercontinental communication, where time
differences can be so large that there is no overlap of working hours—and for
cooperation on documents.
Developments are also under way for “screen sharing,” where two or more
people can both see the same screen picture and have access to it. There is work
going on both on systems for meetings where the participants are in the same
room (when the common screen typically will be projected onto the wall) and
for meetings/conferences where the participants are in different geographical
locations. It is too early to assess the impact of such systems, but they will
clearly facilitate cooperation over distance, particularly in small groups.
Ordinary conferencing systems, on the other hand, are really just an elaborate
form of email systems, where exchanges are open to all participants in that
particular conference. Conference systems are useful for spreading information
fast to many people (“bulletin board” function) and for conducting group-
oriented work.
Experience shows that we intuitively perceive email as a new medium with a
new set of properties: The casualness induced by the easy, instantaneous, and
paperless transmission of messages; the ease of replying; and the absence of the
formalism associated with paper combine to make email messages much more
“oral” in their form than ordinary letters. Because of its informal character,
email can also function as a valuable feedback channel for managers. It tends to
elicit comments more in line with what the manager would get through an
informal chat with a subordinate, while preserving the time-saving,
asynchronous nature of the written memo. Through conscious use of electronic
mail, it is therefore possible for managers to appear more accessible to a larger
number of their subordinates. The informal and private character of an email
system, however, also makes it conducive to gossip and slander—it is in many
ways a new and much more efficient office grapevine.
Electronic mail is not the panacea that some people seem to think, however,
and many of the predictions about email and conferencing have a lot in common
with the generally euphoric predictions made about the telephone around the
start of the twentieth century. We therefore have good reason to maintain a
relative calm.
The main point is simply that email does little to speed our comprehension.
We may transmit messages more easily, and, yes, it is somewhat easier to
compose them. Email messages also tend to be terser (and thereby more
efficient) than other written messages. Word for word, however, they still take
the same time to read as other written material, and the same time to ponder as
information in any other guise. Because it does not enhance our basic
communication capacity in any dramatic way, email will therefore not represent
the revolution in communication that many predict. We already spend a fair
amount of our time communicating, and to increase our use of email we must
spend less time on other channels. To fulfill the predictions of some of the most
eager proponents of email and conferencing, we would have to forgo all or most
other tasks, and that is just not going to happen. Many users in organizations
with electronic mail networks even now receive 100 messages or more per day,
which is probably already taking them to the limits of their capacity. “We
believed e-mail was a way of saving time,” begins an article about email in
Svenska Dagbladet, 1 one of the major Swedish broadsheet dailies. “But the truth
is that we are about to become slaves under the new communication medium.
Many people spend several hours every day answering electronic mail.” In the
article, Professor Jacob Palme of Stockholm University says his research shows
that it takes an average of 30 seconds to read a message and 4 minutes to answer
one (see also Palme 1995). If 100 messages is received, 20 of them answered
and 5 new ones written, a simple calculation indicates that this will on average
take 2,5 hours, or a third of a normal working day.
Mitigating remedies for overload will no doubt be found—indeed, some are
already available (such as filtering and automatic prioritizing based on
keywords) and more sophisticated schemes are under development. That is not
the point, however. The crux of the matter is that we have about the same limit
for output and input of verbal information as our forebears of hundreds and
thousands of years ago, and with the size of the nozzles thus being relatively
constant, we gain little by increasing the diameter of the hose, or by connecting
more hoses.
Although distance education can make teachers more effective by saving
them travel time, I always wonder when I hear proponents of
1 Svenska Dagbladet, October 16 1998. My translation.

distance education extol how PCs and modems will also make it possible for so
many more students to achieve direct contact not only with national experts, but
even with leading international academicians. I keep getting this vision of a
poor professor, already straining under the effort required to keep a decent
dialogue with the students physically present at the department, one day
receiving the joyful message of a sudden availability, through electronic mail, to
10 000 new students nationwide and 5 million more worldwide! Telecast
lectures and remotely accessible text bases is one thing, that will work;
unlimited return channels are something entirely different, and will not work.
Telephones and Videophones

The telephone is definitely a pre-computer invention. Computers do not change
them very much, except for sharply reduced communication costs and the
feeling of freedom associated with the cellular phone—the illusion that we may
now roam the sea, the forest, the mountains, or the prairie while remaining in
touch with the office and the world that brings in our money. The cellular phone
does not entail any real revolution in our communication abilities, though.
Combined with a modem it will make a real difference for people who need to
conduct their business on the go, but, for the majority, it will only afford a
number of conveniences and a marginal increase in efficiency. It will not usher
in really significant, broad changes in organization.
A part of the new telephone environment is voice mail. It has largely the
same kind of advantages and disadvantages as email, with the added advantage
of the extra information contained in voice inflection, and the added
disadvantage that it is not text and thereby not as useful in our predominantly
chirographic work environments. Some managers like the way their messages to
subordinates get a more “personal” touch when delivered in their own voice.
There are examples of systematic use of this effect. One of the most well-known
ones is Debbi Fields’ use of voice mail to communicate daily to the store
managers in her chain of Mrs. Fields cookie stores (Walton 1989). The voice
mail system also allowed the store managers to send voice messages to her,
which they often did. For Debbi Fields it was mainly an instrument of control,
but also a channel for informal feedback. It consumed a considerable amount of
her time every day, but it gave her a very direct channel for influence, and the
store managers, a feeling of direct contact. It did this while preserving her
power to decide when to listen and when to speak, and she could still keep
command of her schedule.
The slow but steady progress toward affordable videophones is probably of
far greater interest. But even videophones will not increase our
capacity for information absorption—their contribution will be that they may

reduce the need for travel and make it possible to have a close working
relationship without meeting in person too often. We do not know for sure yet,
since the high cost of videoconferencing up to now has precluded widespread
use, and the necessary facilities have not been suitable for desktop installation—
making it inconvenient to use even in those organizations where it has been
made available.
However, the videophone is already on its way to become an add-on card that
can be integrated into the PC or workstation, or even just a feature of the
graphics adapter, and the larger screens gradually becoming the norm (17”–21”)
should allow us to conduct small video-meetings (up to seven or nine
participants) on our desktops as soon as the video equipment and transmission
itself become cheap enough. The even bigger screens that are bound to take
over as soon as technology can provide them at a reasonable cost should be able
to display pictures of an even larger number of participants, as well as a
common working space for sketching, writing, and presentation of pictures or
output from various programs.
The arrival of affordable desktop videoconferencing could make a really
significant difference to cooperation over distance. The added information
provided by the picture could make the videophone an instrument for
comfortable conferencing—good enough to replace “real” meetings in many
instances when travel (short or long distance) is involved. It is therefore likely
that video conferencing, within the scope of a couple of decades, will finally
provide the necessary means for reducing the need for “physical” meetings
significantly. Whether this will actually happen, however, depends not so much
on the technology itself, but on people’s preferences—they must actually prefer
video meetings in instances where they would previously have traveled. It is
still too early to tell, but we might guess that most of the really frequent
travelers will welcome the opportunity to spend less time on the road or in
airports, whereas those who make only a couple of trips a year will want to hold
on to what they experience as a welcome escape from the daily routine. Several
studies support this view (Long 1987).
A significant catch, of course, is that if desktop video conferencing really
reduces the perceived threshold for holding meetings and becomes popular, it
could well increase the total number of meetings. This is reported in a study
quoted by Long (1987). About half of the respondents reported an increase in
the total time used for communication after the video-conferencing system was
introduced, whereas the other half reported no change. Such an increase in
meeting activity could be productive in some instances, but not necessarily
always. A reduced threshold for meetings (through video conferencing) also
means that the threshold
for follow-up meetings with superiors will be reduced—something that could

once again reduce the independence of geographically dispersed organizational
units.
Video conferencing will not eliminate the preference for colocation for
groups working really close together, especially groups doing creative work
where cross-fertilization is important. No electronic channel can yet replace the
richness of informal, personal dialogue, the chat in the door of an office you
were passing while you were really heading for the copier, the chance meeting
by the coffee machine, or the inspired but unplanned discussion during a break
in a late-night dash to meet an approaching deadline. As a manager I once
talked to said: “It’s so much easier when you meet people in the corridors.” The
attempts to recreate such avenues for unplanned, informal communications by
electronic means have had scant success (Johansen 1988). We may also suspect
that problems of a more intricate or delicate nature will lead to travel no matter
how widespread video conferencing becomes. Indeed, another study quoted by
Long (1987, p. 58) found that some corporations experienced that their “travel
costs have increased as better communication with distant operations reveals
problems which require in-person appearances of top-level executives.”
But, for more routine administrative work, for coordination, following up on
work in progress, and for sorting out the daily problems cropping up in every
organization, video conferencing should be adequate in many instances. It could
therefore make geographically dispersed organizations more feasible, especially
if they are otherwise advantageous—for instance, because of market
considerations or the availability of energy, raw materials, or suitable personnel.
It may also be very useful for bringing cooperating, but otherwise independent
organizations closer together.
Better Hoses, Same Nozzles

To sum up: Computer-based systems provide a couple of new “hoses” for
information transport, and significantly improve some of the old ones. The
“nozzles” at both the transmitting and receiving end have not changed very
much, however. Our basic capacity for input and output of verbal information
(be it oral or written) still puts the same iron constraint on our communication
process. We already use so much time for communication (especially if we
include the time used for reading printed material) that we can hardly increase
our total communication volume very much, except for using the time now
spent on travel and unanswered telephone calls. Increased use of new channels
will therefore normally entail a reduction for old ones.
A Summary of the Main Impacts

Computer-based systems usher in a revolution in the performance of our
externalized memory. That revolution is built on three pillars: compact and
cheap storage; universal access; and automatic search, retrieval, and
registration. The most important application is the database.
In contrast to the written file, which requires that you physically walk up to it
to retrieve information, digitally stored information can be accessed by anyone
so authorized, regardless of geographical location or time of day. It is,
moreover, simultaneously available to a large number of users. Just as
important, digitally stored information can be indexed, sorted, and compared
with enormous speed. The computer’s outstanding ability to search vast
amounts of information in incredibly short time, and extract, combine and
concentrate data, makes for a momentous difference between computerized and
paper-based files. This is especially true for information in highly structured
form, but, even for text and other information items stored in free-form
databases, accessibility is dramatically increased.
However, the greatest achievement effected through the digital computer is
the externalization of processing. The fact that we can now have information
processed outside the human head will prove to be at least as important as the
externalization of memory brought about by the invention of writing. But the
computer can only mimic certain aspects of the mind—notably, logical
operations, especially all kinds of calculation. The narrowness of the computer’s
proficiency must not fool us into believing that it is inconsequential, however. It
allows us to manipulate vast amounts of quantitative information very cheaply
and quickly, something that translates into a revolutionary ability to handle a
great number of complex matters—from budgeting to finite element analysis. It
allows us to keep track of a vast number of variables and their
interrelationships—complexity on a scale that we were not even able to
approach before—and through this it will also allow us to develop automation
to a level of sophistication that will completely overshadow all that mechanical
automation has ever achieved.
The great breakthroughs in human communication were all developments of
the pre-computer world. Computer technology has so far not provided
breakthroughs of a similar magnitude, and its ability to do so in the near future
is in my view generally overestimated. The reason is simply that
communication is not only a matter of transporting symbols from the desk of
one person to the desk of another—there is also the problem of the actual
absorption and dissemination of information, which represents a far more
formidable problem. However, the computer can be of great help in
transforming information for faster absorption—notably quantitative
information, which can be represented through graphics and animations. The

possibilities are far less promising for verbal information, which is one of the
reasons why email and computer conferencing will have limited impacts. The
bottleneck is still in our heads; and the nozzles there remain the same size, no
matter what the width of the hoses leading up to them.
Videophones and video conferencing will definitely become common when
quality transmission becomes cheap enough, but the effects are uncertain. So
far, telecommunications (which we have had for more than a hundred years,
even internationally) have not led to any measurable reduction in physical
travel—but the technology has perhaps contributed to more meetings overall
and stronger centralized control in organizations.
235
10 Emotional Barriers and

Defenses
“Men live but by intervals of reason under the sovereignty of

humor [caprice] and passion.”
Sir Thomas Browne, A letter to a friend, 1690
In Chapter 6 it was briefly noted that technology and scientific methods are
indeed used to affect emotions—from drugs and psychotherapy to movies. IT
does not seem to bring much new in this respect, except for making existing
products more sophisticated. We may, for instance, observe that Morris’
remarks about our need for sex by proxy can be extended from literature and
movies to computer-based systems. Just as in video, the sex industry has been
among the pioneers in multimedia, something that can be ascertained just by
browsing the back alleys of the World Wide Web or looking at the last few
pages of the classified ads in PC Magazine. Indeed, according to frequent
reports in the media, pornography downloading is so widespread that it is
periodically straining the capacity of corporate communication networks. If we
are to believe the more easily excitable journalists in the trade press, virtual
reality is the next frontier—although I have an inkling that it will be harder to
provide adequate feedback than the enthusiasts seem to believe. Reading such
speculations, one is reminded of one of the bleak worlds described by Olaf
Stapledon in his 1937 novel Starmaker, in which broadcast brain-stimulation
had advanced to a stage where simulated experiences became more important
than reality. Ultimately, this civilization developed the possibility for their
citizens to retire into a completely vegetative, simulated existence: lying
permanently on a bed, connected to life-supporting machinery, one could
indefinitely immerse oneself in broadcast simulations. Seemingly, the
broadcaster networks have been working hard toward this goal ever since.
Some may feel it inappropriate or at least not very serious to draw such
matters into the discussion. However, the immediate and widespread
exploitation of new technology for sexual purposes can serve to remind
us that our basic drives and emotions are with us still, and in no small way
either—if the fervor of the development efforts reflects the size of the market.
We are not intellectual beings with rational purposes all the time, and raw
emotional cravings can easily override even strong rational criteria and
organizational as well as social norms—on all organizational levels. Emotions
are an issue in organizations whether we like it or not, and the whole range will
be present, from the despicable to the noble.
So far in Part III, technology and the rational use of it has had the whole
focus. Before we go on to analyze the rational use of technology in
organizational contexts it seems therefore highly appropriate to consider some
emotional issues as well.
Organizational Effects on Emotions

That emotions were something that affected workplace behavior was not
generally acknowledged in the modern era until after World War II, and the first
significant impetus in this direction only came after the Hawthorne studies (the
experiments were conducted between 1927 and 1932) established it as a fact of
social science (Hollway 1991). Hollway quotes Roethlisberger and Dickson 1
(Hollway 1991, p. 72), describing the early experimenters at Hawthorne as
. . . carrying around in their heads the notion of “economic man”, a man primarily
motivated by economic interest, whose logical capacities were being used in the
service of this self-interest. Gradually and painfully the experimenters had been
forced to abandon this conception of the worker and his behavior . . . they found
that the behavior of workers could not be understood apart from their feelings or
sentiments.
Since the Hawthorne studies were published, the scope of work psychology has
been both broadened and deepened through human relations, organizational
development, and the concept of organizational culture. Practical efforts have
focused not so much on job content and design as on selection and motivation
building, which is far less intrusive with respect to work processes and
organization design. The modern emphasis on organization culture can even be
interpreted as a return from a focus on organization and job content to the more
pure motivational effort of early human relations (Hollway 1991), and the
proliferation of psychological testing could indicate a preference to fit persons
to jobs rather than the other way around.
1FromRoethlisberger, F. J., and Dickson, W. J. (1939): Management and the Worker, Cambridge,
MA, Harvard University Press.
10 Emotional Barriers and Defenses 237
I am not sure if this can be said to represent an advance in work psychology.

It could well be that the development of the modern organization poses new
challenges to our basic emotional apparatus, which it is poorly equipped to cope
with, and which a recharged motivation only can serve to gloss over. It seems
reasonable to believe that our emotions co-evolved with our physiological
abilities to cope with the challenges of a life that for perhaps 95% of our
existence as a species 2 has been life in a hunter/gatherer band, and for the rest
(save the last 200 years) a life of subsistence farming and simple crafts.
Physiologically (genetically), we are therefore probably best equipped to do
physically varied work of a routine character, interspersed with limited amounts
of problem solving and crisis management. During the last 200 years, however,
an increasing number of people have entered into jobs that consist mainly of
problem solving and crisis management, and the problems have grown
increasingly complex and abstract. Quite a few of these problems may even be
impossible to solve in a satisfactory manner. Moreover, many jobs are also such
that the people doing them cannot easily see their significance either for the
organization or for any particular end product, and they are deprived of the
inherent meaning we find in work that is whole and with an immediate bearing
on our own or our family’s survival—as is the food foraging and tool making in
a hunter/gatherer band.
It would be understandable if such situations generated considerable stress
and emotional problems—which they seem to do. In a survey conducted by an
American insurance company in 1991, 46% of American workers felt that their
jobs were very or somewhat stressful, and nearly 27% reported that their jobs
were the single greatest source of stress in their lives (Quick et al. 1992).
Factors such as high work pace, repetition of work, lack of control over work
and work situation, quantitative overload (too much work), and qualitative
overload (too difficult work) are reported to be among the chief sources of
occupational stress (Ross and Altmaier 1994).
Growing complexity is, in fact, an increasingly prominent characteristic of
modern society as a whole—and an increasing number of people find it difficult
to understand how it works, what their options are, and how they can claim a
meaningful place in it. Even everyday life requires a growing sophistication in
abstract thinking and symbol manipulation, from filling in forms to using
computers and other electronic devices, and many people probably feel that
society is closing them out.
2Homo sapiens, which has existed for about 300,000 years. If we count in the whole genus Homo
(about 3 million years), we talk about more than 99% of our existence.
Many jobs are also much more monotonous than any we would encounter in
our “natural” state, and probably produce a strain on our emotions. Finally,
many modern service jobs demand a significant degree of “emotional labor”
(Hochschild 1983, Putnam and Mumby 1993)—the effort exerted when
“individuals change or manage their emotions to make them appropriate or
consistent with a situation, a role, or an expected job function” (Putnam and
Mumby 1993, p. 37). Typical examples are the consistently smiling attitude of
the McDonald’s salesperson or the airline hostess, and the professional
consolation of the cancer ward nurse.
We should be careful, however, not to think that our emotional situation has
necessarily deteriorated as a whole. Small peasant villages pose their own
emotional strains, as does life in the extended family. The work of the serf or
the tenant farmer could be harsh and monotonous enough, and serfs as well as
servants have always had to do emotional labor to please their masters—who up
through history have had considerably more gruesome sanctions to apply than
the sack.
Emotions and Organizational Constraints

Our emotional apparatus was shaped in small groups and tuned to the needs of a
life in the roving band. It might easily have dysfunctional effects in larger
organizations, especially organizations built on the rational model. As we made
the transition to modern society, emotions became a double-edged sword.
On the one hand, positive emotions bring social bonding and cohesiveness
and oil the inner workings of organizations. Emotions are vital for producing the
esprit de corps and individual motivation that makes organizations flourish and
that can bring about success even in the face of what seems like insurmountable
problems. They also provide us with many of the spices of organizational life.
On the other hand, as we noted, emotion-based social bonding functions (in
line with our primate nature and hunter/gatherer origins) primarily within the
local group—the people we meet in the flesh and interact with on a daily or
almost daily basis. Although it is extremely useful for building cohesiveness in
small organizations, this trait is just as likely to bring divisiveness in large
ones—pitting departments against department, and work group against
management.
This mechanism works on many different levels, not least geographical. It is
quite common, for instance, in large multiplant companies, that the different
plants exhibit strong, even fierce, independence and view company or division
headquarters as a remote and largely irrelevant entity, only noted for its
“interference” in local affairs. Clearly, such
strong plant loyalty can be an asset in the efforts to improve locally, but a
liability when it comes to achieving close cooperation and coordination in a
larger organization. Building the necessary minimum of loyalty among people
and organizational units who are geographically dispersed and who seldom
meet requires strong, persistent efforts, and considerable energy is necessary on
a continuous basis to maintain it.
In the world of humans, however, almost any difference can serve as a basis
for divergence in culture and loyalty—not only geography, but also product
affiliation, profession, status, and age, to mention some of the important ones.
The natural preeminence of such local cultures and loyalties over an
identification with the organization as a whole probably constitutes one of the
most serious constraints on the size an organization can attain and still operate
effectively and efficiently, and successful large companies devote considerable
resources to foster common, company-wide sentiments. Especially in
multinational companies, the problem of fragmentation is one of management’s
main concerns, and a driving force behind their demands that prospective
managers circulate through several countries and companies during their career
buildup (Bartlett and Ghoshal 1989).
All organizations experience emotion-related problems, and much energy is
daily spent coping with them. From time to time they may become serious and
require considerable attention and effort, and sometimes an organization will
encounter conflicts so severe that it simply goes under—is dissolved, taken over
by others, or simply goes bankrupt.
IT and Emotions
If it is correct that the increasing complexity of society as a whole and of many
work situations does not match our emotional predispositions, the situation is
hardly improving. Based on the rapidly increasing use of computers and ever
more complex tasks, the tendency toward greater complexity and abstraction in
work is continuing. A steadily growing number of jobs are performed with
computer-based systems as the only tool or as an important support tool,
whereas simpler jobs are eliminated in large numbers.
This is not unproblematic—after all, why should our emotions be better
adapted to computer-based work than our bodies are to screens, keyboards and
mice? If we contrast the tools provided by information technology and their
possible use with the life situation that has formed us, we will easily see a
number of areas where the exploitation of the new tools in the name of logic and
reason will clash with older parts of our psyche. These main areas seem to be
the increasing abstraction and
complexity of work, the inherent relentlessness of a tireless technology built on

logic and an unwavering demand (and capacity) for preciseness, and the social
isolation that can grow from screen-based work.
Abstraction and Complexity

The process of abstraction is fundamental to the use of information
technology—you always interact with a representation of the thing or process
you work with, not with the real thing itself. Even in word processing, you leave
behind the paper and work on a virtual document. This means that you must be
able to understand the relation between the representation and the thing or
process behind it, and you must learn to relate to unfamiliar cues and
impressions. Usually, complexity also follows abstraction; the ability to manage
more complex tasks will often be one of the main reasons to use IT.
Zuboff (1988) has described and named some of the most important changes
that have happened to our everyday work in the course of this process. She has
done so by exploring two central concepts: (a) the qualitative changes in the
required set of skills as we make the transition from action-centered to
intellective skills, which we will discuss here, and (b) the informating of work—
the deepening of the understanding and responsibility that increasingly
sophisticated control systems thrust upon us—which will be discussed in
Chapter 14.
From Action-Centered to Intellective Skills

In Zuboff’s terms, then, we are experiencing a transition from a reliance on
action-centered skills to an emphasis on intellective skills. Action-centered
skills are the skills of manual labor and of the direct control of machinery. They
are, so to speak, the skills of the body—acquired through extended hands-on
experience, and relying heavily on tacit knowledge (Polanyi 1967) and theories-
in-use (Argyris 1980), even intuition. In many ways, action-centered skills
correspond to the oral mindset. This correspondence is also noted by Zuboff.
Explaining the transition to fully computerized process control in a paper mill
she studied, she quotes a manager (1988, p. 71): “You need a new learning
capability, because when you operate with the computer, you can’t see what is
happening. There is a difference in the mental and conceptual capabilities you
need—you have to do things in your mind.”
The transition here is even more dramatic than the passage from an oral to a
literate mindset. Not only does work in a control room involve the acquisition of
a set of abstract symbols and categories that describe the machinery and
processes used in production (artifacts and events in
the real world), it also requires the ability to understand what is happening in the
production process through the nature, values, and interrelations of those
symbols. On top of that, the operators must also be able to exert precise control
over the production process (real-world artifacts and events) through a highly
abstract, symbol-based interface, without any physical contact with the actual
process at all (with the exception of some error situations).
Evidently, the abstraction of work—which is one of the outcomes of this
transition—represents quite a dramatic change in the work situation. It therefore
seems reasonable to expect emotional problems to crop up, both because of the
changes in the individual work situation and as a result of the concomitant (and
inevitable) changes in organization roles. There are probably many people who
will find it difficult to adapt to this kind of work, and to a situation of greatly
increased personal responsibility and the loss (at least partially) of the protection
that has traditionally been built into workers’ collective culture and bargaining
(Lysgaard 1961).
Responsibility and Role Conflicts

As we shall see later (Chapters 12 and 14), the new computer systems not only
force an abstraction of work, but also impart a deeper understanding of the
processes they are used to manage and increase the possibilities for improving
the performance of that part of the organization—or even the whole
organization, as in the case of the control room operators in the paper mill
Zuboff studied. With such an increase in understanding and control follows a
corresponding increase in responsibility, which means a transition from a
relatively simple job with a limited and stable set of rather concrete problems to
tackle, to a job where a lot more time is devoted to problem solving, and where
the problems are more complex, more abstract, more varied, and more difficult.
This change is noted also by Walton (1989).
To cope successfully with the new job, people have to make deep changes in
traditional attitudes to work, turning away from the old tenet that one is not
responsible for anything outside the narrow frame of one’s own job. A change
here will in turn undermine old peer group identifications. If you start to feel
(and respond to) responsibility for a larger part of the organization (or the whole
organization), your identification is already shifting out of your peer group and
is becoming more like the attitude formerly exhibited only by managers. This
can cause considerable problems and conflicts between long-time fellow
workers, with serious emotional fallout, as reported by Zuboff (1988).
But computers do not only tend to make work more complex and more
oriented toward problem solving. The changes described are also often
accompanied by a greater emphasis on teamwork and flexibility, which, for
many people, can be another source of stress and insecurity, since such demands
can also easily be perceived as resulting in less autonomy: Instead of carrying
out well-defined tasks with a mutually recognized, limited set of
responsibilities, one is suddenly at the mercy of events over which one has little
control. Problems crop up and demand solutions, and one never knows what one
will have to do next. Zuboff quotes an operator reflecting on such a job situation
(1988, p. 405):
They say the new technology will require a flexible system, you have no choice but
to go where they send you, when they send you. You can get to earn high pay, but
you have no choice about what your job is, and you can’t have your own job. You
never know what to wear to work—do you wear your good Levi’s or do you wear
your greasy ones?
Rigidity and Relentlessness

Then there is the issue of the untiring nature of computer-based systems, their
inherent craving for precision, and their narrowness in only responding to and
reporting information types that have been defined beforehand. Taken together,
these characteristics can easily translate into rigidity and relentlessness, if
special care is not taken to avoid just that. Zuboff (1988) describes several
examples. An interesting case is the Work Force Supervisory System (WFSS)
of what Zuboff calls Metro Tel, a part of a large telecommunications company.
Prior to the introduction of that system, each worker was a member of a crew
assigned to a particular electronic switching station (ESS) and headed by a
foreman. The crew was responsible only for the maintenance of its local ESS,
and the foreman decided job priority and assigned individual workers to tasks.
When error detection and analysis were centralized to switching control centers,
each covering an extended geographical area with many ESSs, a new situation
arose. Because error detection and analysis were no longer local, tasks did not
have to be locally assigned either. That meant that maintenance workers did not
have to be “wedded” to one particular ESS; they could be dispatched from the
center as the need arose, with precise instructions for each job.
With this new angle of attack, there was suddenly a need for a considerable
new bureaucracy to manage the queue of tasks and assign them to capable
workers, and the situation quickly approached a mild chaos. There was also the
new problem of supervising workers working in isolation in the (now more or
less deserted) ESSs.
The WFSS was designed to solve these problems. From a work identification
number, it prioritized tasks and automatically determined the time they should
take to complete. Then it assigned jobs to the individual workers that put
together should match the priority listing, match the worker’s skill level, and
give each worker a workday lasting the prescribed 8.5 hours. Instead of
reporting to the foreman at their ESS, the workers now reported to the system
each morning, receiving a description of that day’s work. Of course, they also
had to report back to the system each time they completed a task, and this
information was available to the centrally located foremen—giving them a very
accurate view not only of each worker’s progress through the day (or night, for
this was a round-the-clock operation), but also of their accumulated
productivity. Failures to meet the calculated repair times would show
immediately.
For the workers (and even for the foremen) this represented a dramatic
change. Under the old regime, they lived in a traditional work organization, with
ample human contact, and with all the ambiguities and flexibility of normal
human interaction intact. With WFSS, they could work for days without any
face-to-face contact whatsoever.
Initially, many foremen favored the new system, since it gave them an
unprecedented overview of their subordinates’ work. As two of them said
(Zuboff 1988, p. 331): “It is beautiful now. I can track my people’s work. All I
have to do is type the craft’s initials in and see how he is progressing and see
what his total workload was. What is his productivity? Before, we had to judge
people more on hearsay. Now we have it black and7 white.”
The workers, however, quickly perceived the system as rigid and
unrelenting—it was no longer possible to negotiate tasks and times, and it was
hard to gain acceptance for extra time if unforeseen difficulties cropped up.
When management started to use the system’s efficiency ratings to evaluate
workers, even many foremen cautioned against it. Among other things, they
pointed to the fact that the best people tended to get the most difficult jobs (the
system was biased to do just that), jobs that often required more than the
calculated time to finish. That would not show up in the statistics, and the best
people could therefore end up with the worst ratings.
The foremen also discovered that they lost touch with their old people, and
knew next to nothing about new hires, whom they only met as numbers in the
system. They started to lament the loss of flexibility, of joint problem-solving,
and the circumstantial but important knowledge that only diffuses through
personal contact—such as if anyone had problems in the family or any other
legitimate reason for receiving lenient treatment for a while. After some time,
the foremen also learned that the system was being used to rate them as well,
and they were beginning to feel the same sort of misgivings as the workers
about the omnipresent
monitoring the system represented. As a former worker, recently promoted to

foreman, said about the WFSS system, also reflecting the feelings of the
majority of workers (p. 352): “I hated it. It was too close. I could no longer hide
anything. Management could monitor me hour by hour, and that was kind of
scary.”
After a while, the system was being widely experienced as too precise (in
quantitative terms), too rigid, and too unrelenting to live with—it left no room
for human judgment or for the pockets of privacy, ambiguity, and personal
relations that human emotions crave. The outcome should have been
predictable: to reclaim some of their lost room for maneuvering or to correct
what they perceived as distorted reporting, people started to cheat the system—
feeding it false information, ignoring some job assignments and claiming they
had never received them, and so on.
Cheating the system can of course be interpreted as a manifestation of bad
discipline and general irresponsibility among employees—after all, we know
that discipline in such matters is, to a large extent, an acquired quality. The
development of industrial culture through history as well as the differences
between national cultures today testify to that. But, for most of us, there also
seems to be a final threshold, beyond which we cannot be pushed without
serious consequences. We all need a sphere of privacy, a certain room for
maneuvering, a minimum of slack. That men are not machines is generally
regarded as a truism, but sometimes we seem to forget that truisms are in fact
true. Our basic emotions are not rational, they cannot be eliminated, and
systems that encroach too much on them are bound to cause problems.
The Significance of Design

The objection can of course be made here that the WFSS system mirrors a rigid
and inhumane management philosophy more than features that must necessarily
be a part of computer-based systems, and that other design decisions could have
produced a system with quite different characteristics. That is true to some
extent, and a lot of people working with systems design and development have
been quite concerned about this and sought ways to build “humane” systems (a
good approach is presented in Eason 1988). The WFSS system could, for
instance, have been designed not to schedule work and calculate necessary time
automatically, but only to provide foremen with information as a basis for their
decisions. It could have provided fields for comments and other unstructured
information. It could have provided the workers with some latitude in choosing
the order and priority of tasks.
However, computer-based systems are in their very nature based on logic;
their strengths are first and foremost the storage, retrieval, and
automatic processing of structured, preferably numerical, information.

Therefore, they have an inherent tendency toward rigidity and relentlessness,
especially in environments where increased efficiency is highly valued (which
includes most places in today’s industrialized societies). To escape the worst
outcomes, it is especially important to avoid machine-paced work and detailed
surveillance, and to position computer-based systems primarily as tools, not as
automated managers.
Social Isolation
Humans are social animals and generally dread social isolation—loneliness is
indeed a very negatively loaded word. Isolation is also a traditional punishment,
and extreme isolation has always been a central instrument for those who want
to break someone down psychologically.
Common sense would imply that isolation in the workplace is generally no
more desirable, and research seems to bear this out (Sundstrom 1986), although
Sundstrom remarks that “isolation and its effects have apparently not been
systematically studied in work places” (1986, p. 295). Isolation is reported to be
more tolerable if the work is interesting (Sundstrom 1986), but even then a
certain level of social contact is preferred—there are few people who do not
venture out of their offices for a chat a few times during the day.
Technology-induced social isolation is nothing new. Noise and machine
layout in factories have often created jobs where it is impossible to talk to or
even have eye contact with fellow workers, and, as mentioned earlier, it was not
until after the Hawthorne studies that feelings and social relations really started
to be acknowledged as important factors in the workplace (Hollway 1991).
Social contact has been a consideration in factory layout since then, but it is
impossible to tell how strong the impact has been across industries and national
cultures.
As with older forms of technology, the introduction of information systems
can easily produce social isolation for workers and professionals alike. If
information is stored in databases and channeled through computer systems
instead of human contacts, if email replaces a significant part of all telephone
calls, and video conferencing does the same for travel and face-to-face
meetings, our social interaction at work can be reduced significantly both in
volume and quality 3 if steps are not taken to avoid it.
Social isolation was one of the complaints raised against the WFSS at Metro
Tel, but it can be more pronounced in other situations. The obvious
3To me, it is obvious that a reduction in “bandwidth” means a reduction in quality of social
interaction. Being there is better than video contact; a live voice is better than letters on a screen.
case is the caseworker supplied with a terminal giving him or her a total set of
tools and all information necessary for completing work. In addition to the
isolation produced by the systems themselves, seclusion is often intentionally
maximized to eliminate chat and increase productivity. A benefit analyst is
quoted by Zuboff (1988, p. 139):
We used to be able to see each other and talk. Sure, sometimes we just talked about
what we were going to make for dinner, but we always worked while we talked.
Most of the time, we talked about something related to a claim. Then with the new
system, they put in two filing cabinets between us, because we weren’t supposed to
see each other anymore. But there was a small space between the two cabinets, so
she could still turn around and look at me, and we would talk across to one
another. Well, one day a manager walked by, and I was asked who left this space
there. I said that was how they left it when they put the cabinets in. The manager
had them move the cabinets together because they don’t want us talking.
Most of the workers who were affected by this system reported sharply reduced
satisfaction with their work, both because of the isolation they experienced as
well as the monotony of screen-based work and what they felt was the relentless
tempo of their new, system-paced work situation. It represents another example
of heedless exploitation of the technology’s strong points.
In addition to the negative effects that such isolation has on the individual
worker, it seems fairly obvious that it is detrimental to any efforts in the
direction of improving employee morale and building an enthusiastic corporate
culture. It may therefore run counter not only to employee well-being, but also
to the total interests of the organization. If there is no social interaction at work
and if the work situation is experienced as stressful and socially impoverished,
loyalty is bound to drop and corporate culture will suffer.
Emotional Barriers to Virtual Organizations

A popular theme of the last few years has been the prospect of “virtual” teams
and “virtual” organizations. The meaning is seldom clearly defined, but a
“virtual team” generally implies that the people in the team work in at least two
different locations. Often there are more locations, or one or several team
members may be traveling most of the time. The defining feature is that the
team members use one or more electronic media, such as email, computer
conferencing, video phones/conferencing, common calendars and common
information bases as their main communication channels, and they have little
face-to-face contact.
“Virtual organization” usually has two main meanings: It may (a) either
designate several more or less conventional companies working
very closely together (even fronting the market as one organization) with
electronic channels or even common systems as communication medium, or (b)
an organization where a large number of the organization members use
electronic channels as their main (or even only) medium for contact with each
other and with the rest of the organization, thus forming the virtual teams that
carry out the work and represent the main organizational structure.
Today, many seem to believe that the virtual organization is the main
candidate for the title “organization of the future.” I think that is a superficial
judgment with a very weak foundation. We shall return to this question in
connection with a discussion of groupware in Chapter 11. Here, we shall
consider how emotional aspects may act as barriers or brakes on the
establishment and success of virtual organizations—at least of the latter type,
where the organization members use electronic communication channels only
(or mainly) for their interaction.
True to our heritage as humans, we tend to achieve closest contact and build
the most durable trust and loyalty toward the people we meet most often and
over extended periods of time—whether they are (originally) involved in our
work or not. We like to look people properly in the eyes to assess their worth.
Face-to-face contact is the richest communication channel we have, and any
electronic channel is significantly poorer—even top-quality video equipment
cannot measure up to physical presence, let alone the barren dialogue of email.
Of course, our extraordinary flexibility (Berger and Luckmann 1967) will allow
us to build human bonds by the help of very narrow channels, such as email (or
even through old-fashioned letters, as some people actually continue to do), and
some people may even build very strong relationships that way—just like
people earlier have done by writing letters. Indeed, the first Internet marriage
(where the couple met and courted on the Net) has already taken place—
although, undoubtedly to the chagrin of true cyberspace devotees, the bride and
groom chose to appear before the parson in (physical) person.
My assertion is that, given our basic psychological makeup, the richer
channel will in general produce the stronger bond. And, if this is true, it stands
to reason that organizational loyalties in a virtual organization, where members’
face-to-face contacts mostly involve people outside the organization, will be
significantly lower than in a more conventional organization where people meet
physically almost every day. The chances that organization members will be
tempted to let their ideas and initiatives take off in other directions should also
increase. It should also be much more difficult to build a strong organizational
culture and a corporate identity that people can identify with when the
organization is virtual and offers few opportunities of normal social encounters.
A virtual organization, then, should normally not be able to display the same
cohesiveness, resilience, and endurance as a “physical” organization, and should
therefore experience a handicap that must be outweighed by other factors. I
therefore doubt that the fully virtual organization—without a physical location
and built fully on electronic communication—will become a common form.
Flexibility in structure and personnel is good up to a point, but extreme
flexibility may all too easily translate into instability, disloyalty, and
inefficiency. I also suspect that if a really important problem cropped up, the
responsible person(s) in a hypothetical virtual organization would still pack
their suitcases and go—to bring into action the intangibles that are impossible to
convey by electronic means: the sensing of an atmosphere, of a handshake, or
the intimacy of a lunch or dinner conversation.
A New Gender Gap?

In an age where the equal status of women in society is a very important issue,
and their victories often must be defended for long periods of time before they
become ingrained features of social life, it is not without peril to talk about
differences between the sexes. However, it is probably not too controversial to
point out that there seems to be a difference between men and women when it
comes to social abilities and the need for deep personal relations and rich
dialogue. Men are generally acknowledged to be less interested in personal
matters, to pay less attention to feelings, and to make do with a terser dialogue.
Women get more involved in social relations, seek out more personal
information, and generally try to build more complete relationships.
If this is true, it follows that women will find virtual organizations and
electronic communication channels less satisfying than men, will be less ready
to enter into such more narrowly based interactions, and thus prefer
conventional organizations to a larger degree than men. This could lead to a
new kind of gender gap in working life. It could also offer at least a partial
explanation of why there is such an overwhelming majority of men in IT-related
occupations (normal office use excluded): The human-machine dialogue could
simply be more to the male liking, devoid as it is of emotional content.
Information Technology as an Emotional Booster

After all these reservations, it is necessary to point out that information
technology can also help to strengthen both organizational loyalty and culture in
situations where colocation is impossible for old-fashioned reasons, such as the
need to locate close to markets or sources of raw
material or energy. One of the main problems in national or international

organizations is just to build and, not least, to maintain a sense of organizational
unity. Elkem, for instance, one of the organizations that has sponsored this
study, has a number of smelters both in Norway and in the United States. Many
of them started as independent enterprises and were later acquired, and many of
them are located in small, otherwise rural communities—a long distance from
company headquarters or even the division headquarters. These plants are by
tradition fiercely independent and often resent “meddling” from corporate
management. In addition, the company has the Atlantic divide to overcome.
Forging a sense of organizational unity in such an organization is not easy, and
the lack of a community feeling is perhaps the single biggest obstacle to
organizational streamlining in this type of organizations.
Even in organizations with less geographical and historical distance to
overcome, the task of building company identification can be formidable. Some
of the energies bound in the social identifications at the primary and secondary
level (team/department and site/plant) must be transferred to the higher levels.
Mature international organizations have established a lot of practices to achieve
this, especially at the management level (Bartlett and Ghoshal 1989).
Mandatory temporary relocations to one or more foreign sites as part of a career
track is but one of them.
In such organizations, electronically mediated contact is of course better than
no contact, and email and conferencing (particularly video) can build a sense of
community, especially among like-minded people. It can therefore be an
interesting tool for building coherence and cooperation among experts of the
same trade who are spread out among different sites, as well as cooperating
units at different sites.
A General Caveat
Industrialization and the emergence of large organizations have brought us far
away from the content and conditions of our primeval work conditions. IT will
most likely transform our work further. We should therefore take the
opportunity from time to time to remind ourselves that, in our discussions of
technology, systems, and structures, we must not forget that people are what
both society and organizations are about.
Even if they have a rational side that interfaces well with the technology,
people are also living humans with a profoundly emotional nature that must be
taken into account and that has great value in its own right. When discussing
matters like the subject of this book, it is easy to forget that economic efficiency
is not an end in itself and should not be pursued to the detriment of basic human
values. In a competitive world, many people have a tendency to regard such a
view as a luxury one cannot afford, and
it is indeed often difficult to harmonize with a realistic attitude toward pressing

economic realities in organizations fighting for survival. We can only hope that
the steadily increasing material prosperity in industrialized societies will
gradually lead to a more relaxed attitude and a greater interest in using the
increased productivity to improve our lot in a broader sense. We should
remember the words of Blaise Pascal (from Pensées, 1670): “The hearth has its
reasons which reason does not know.”
251
IV Extending the Space of

Constructible
Organizations
In Part III of this book, we discussed how computers have delivered another
quantum jump in the history of organizational tools. In my view, information
technology already ranks on the level of writing itself, and it leads to profound
changes in some of our preconditions for organizing. The most important
improvements over pre-computer technology, as analyzed in Chapter 9, are:
• Computers allow, for the first time, the externalization of processing, and
certain kinds of work that previously required the attention of human minds
can now be offloaded to machines. A potentially limitless source for these
categories of work has been created.
• Processing of quantitative information is presently most important.
Computing tasks that seemed impossible to carry out 50 years ago are now
routine.
• Computers vastly extends the scope for automation and elimination of tasks
in both manufacturing and administration. The potential of computer-based
automation is so great compared to mechanical automation as to be
unfathomable.
• Computers can concentrate quantitative information enormously by
presenting it in graphical form, especially when combined with animation.
By thus exploiting the large bandwidth of our visual system, information
technology allows us to absorb such information much faster than before.
• Computer processing also greatly enhances our insight and understanding of
complex matters and improves our ability to handle complexity. It
significantly extends the coordinative reach and power of one person or a
single team.
• Artificial intelligence and embedded rules and information can support
work, both in time-critical and in knowledge-intensive activities.
252 IV Extending the Space of Constructible Organizations
• Computers also usher in a new revolution in memory. The database offers an

improvement over paper-based files so large that it is of a quantitative as
well as qualitative nature.
• Structured databases are so far the most important, with vastly improved
implicit coordination achieved through global reach, enormous capacity, and
blistering speed.
• Free-form databases (text, sound, and pictures) are interesting and have
important economic potentials but are considerably less important for
organizational purposes.
• Computers increase available communication bandwidth by several orders of
magnitude. The most important aspects are direct system-to-system
communication and remote access to databases.
• Of less but still significant importance are email, computer conferencing,
video conferencing, and other team support tools.
When we go on to discuss what kind of new possibilities these advances open
up for organizing human work, however, we must remember that not even
information technology can cure all ills, and that a number of important
constraints remain in force:
• The human input/output capacity is basically unchanged, with the exception
of quantitative information that can be represented graphically. All
information in verbal form must still be read or heard, written or spoken, and
this iron constraint shows no sign of yielding.
• This constraint still also puts absolute limits on the number of people with
whom we can maintain a meaningful two-way communication—all new
electronic media notwithstanding.
• The usability of databases, especially free-form information bases, is also
severely constrained by the limits of human input/output and processing
capacities. Simply increasing the amount of available information is not
necessarily beneficial.
• Just as with previous technology, human internal processing and deliberation
are not speeded up, even if our ability to handle complex work and parallel
work processes is greatly enhanced.
• The limits of our own mental capacities now manifest themselves in a new
way. The recent (and imminent) advances in information technology, especi-
ally in hardware, place such storage and processing capacities at our disposal
that the main constraint for building more sophisticated and complex sys-
tems has become our own ability to first adequately analyze and understand
the problem domain, and then design and install the intended systems.
• Neither our physiology nor our emotional makeup is adapted to the kind of
highly abstract, problem-oriented work that fills an increasing
253
part of our workdays. The result is often physical and mental strain that can
lead to reduced morale, reduced performance, and even injuries.
• The preeminence of face-to-face contact in the establishment and
maintenance of primary group identification may reduce the viability of
virtual organizations. This constraint may affect women more than men.
Part IV attempts to establish how this new set of preconditions will allow us
to extend the space of constructible organizations. The first chapter discusses
the individual and team level—because they represent the primordial elements
of organization as well as the fundamental building blocks of larger
organizations, and because there are a number of application types (among them
some of the most hyped-up ones) that apply first and foremost to this level.
Then I will move on to discuss the core of the matter: the larger
organizational context and the tools and potentials that apply to the organization
as a whole. This discussion is centered on the three themes that I think embody
the most important potentials provided by information technology for
organizational change and improvement.
Each theme is treated in a separate chapter. The first is “Routines and
Automation,” which will continue to represent an extremely important
contribution to the development of modern societies, allowing enormous
increases in productivity—something that will have a number of interesting side
effects. Computer-based automation also includes automatic routines at various
levels, which is a very important prerequisite for the two other themes. The
second theme, “Coordination by Default,” is about how the use of databases can
contribute to the age-old problem of coordinating the work of all organization
members, both improving on existing arrangements and providing new ones.
The third theme, “Comprehension and Control”, is about how information
technology is used to procure previously unavailable information and to make
information more accessible, thus improving our understanding and control of
both our work and the organization. This has clear implications for organization
structure and the way organizations can be run.
At the end of each of these three chapters, I will discuss the possible
extensions that the examined aspect of information technology may offer to the
space of constructible organizations.
255
11 The Individual and the Group
“One man may hit the mark, another blunder; but heed not these
distinctions. Only from the alliance of the one, working with and
through the other, are great things born.”
Saint-Exupéry, The Wisdom of the Sands, 1948
In Chapter 9, we discussed how information technology can improve individual

capabilities in those areas that are most important for our ability to organize. In
accordance with the view of organizations as constructed and constituted
through individual actions, these improvements represent the foundation for any
IT-induced change of a systemic nature.
However, there is another side to these improvements: they also improve
individual productivity. In this chapter we shall discuss these possible
improvements and try to assess whether they can in turn induce organizational
changes and improvements. We shall also discuss the group level before
moving on to the subject of larger organizations in the next chapters, since the
small group has, throughout human history, represented a basic level of
organization with its own distinct needs and priorities.
The Individual Level

Support Tools
When talking about gains in personal productivity from computers, people
mostly think in terms of increased efficiency for standard office work—for
example, faster production of documents, budgets and related calculation
chores, presentations, and communications. Even though these “Office Suit”
applications have driven much of the investments in computer systems from the
late 1980s and onward, their potential impact on organization is in my view
limited; it may represent no more than a significant reduction in the number of
typists—a contribution to the
general trend of eliminating routine work. Such tools provide mainly what we
might term bounded improvements in productivity—local improvements within
the confines of the jobholder’s usual set of tasks and responsibilities. A typical
example might be the salesperson who can complete a few more customer
contacts and dispatch a few more letters and offers every day with the help of a
PC-based sales support program and a word processor.
We will find the same effects in the realm of science and engineering, such as
systems for statistical analysis, computerized equipment for chemical analysis,
and CAD (computer-aided design) systems. They generally deliver much higher
gains in personal productivity than the standard office tools mentioned earlier,
however—simply because they tap much deeper and more directly into the
numerical processing power of the computer. The combination of simulation
and animated graphics alone has been extremely advantageous. Just imagine the
difference for an engineer designing the front wheel suspension for a new car:
Before the computer, it meant trial and error supported by time-consuming and
rudimentary manual computations; in the first three to four decades of
computing it meant poring over computer printouts, trying to envision what the
numbers really implied. After the advent of cheap and powerful workstations,
the simulated behavior of the new suspension can be seen in real time on a
computer screen as it travels over various simulated surfaces.
As long as they are isolated systems, however, just supporting the work of
the individual professional, even engineering and scientific workstations do not
have any more impact on organization than office tools do. Their main effect
has been a significant reduction in the number of draftsmen and calculation
assistants (a parallel to the decimation of typists). The really exciting processes
do not start until the systems are linked into design databases or planning and
production systems, but, then the systems become more than personal support
systems.
In the ordinary office environment, there is always the danger that increased
productivity will be eaten up by increased output of material of low significance
or through unnecessary embellishments such as fancy layout and presentations
laden with ornamental graphics. It takes both a conscious approach and good
management to really make savings stick.
Cell Automation
In organization terms, some of the same can be said about isolated automation
of single tasks—what we might term cell automation. In both offices and
factories, automating single tasks can increase local (cell) output per employee
many times over. In an office it can, for instance,
11 The Individual and the Group 257
be used for address selection and printing; in a factory, for computer-controlled

machine tools.
In both instances the computers provide not only greater speed, but also
much greater flexibility than previous automation efforts, because of the almost
infinitely greater complexity computer programs will allow. However, unless
the automated cells are linked into some sort of department- or organization-
wide system, traditional coordination methods and organizational structures will
most likely prevail, and the bounded productivity improvements will not
translate into significant changes on the organizational level.
By this, I do not mean that bounded improvements in productivity are
unimportant. There are significant (in some instances even spectacular) savings
to be had, especially in science and engineering, and as the price/performance
ratio of processors continue to improve, processing power that used to be
reserved for multimillion dollar supercomputers is invading the desks of rank-
and-file engineers and scientists. This opens up for dramatic increases in
productivity and an ability to tackle problems with a complexity many orders of
magnitude greater than before. The demand for processing capacity is rising
fast, and, in many areas of science, extensive use of very powerful computers is
greatly accelerating the pace of progress.
Increasing the Span of Competence

Are there, then, any personal support systems that support significant changes in
organization? I think there are—and that the key notion is the attainable span of
competence. The area of interest here is the degree of specialization, and the
amount of coordination and information transfer it necessitates.
When we discussed the emergence of functional specialization in Chapter 7,
it was attributed both to the resulting increase in productivity as well as the need
to reduce the time used for training. But there is of course also another and more
fundamental reason for specialization, rooted in both the limitations of human
memory and our low rate of information absorption: it is simply not possible for
anyone to become proficient in everything.
This is of course not a barrier for the narrow, repetitive jobs of mass-
producing industry, but it becomes an important constraint and a determinant
for the design of organizations or parts of organizations where more complex
tasks dominate. Typical examples are thoroughly professional organizations like
universities, research laboratories and hospitals, but most organizations (and
every large one) will have jobs where the limits to a person’s effective,
attainable span of competence
become a design parameter. In a travel agency, for instance, no one can give a
customer expert advice on travel in every part of the world. In a large bank, no
clerk can advise you on all aspects of the bank’s services. In a government
department, no single person will have the necessary knowledge to carry out
more than a fraction of all the varied tasks falling within the department’s
responsibility.
This is not to say that all specialization in such organizations is based on the
natural limits in the diversity of knowledge that humans are able to maintain—
on the contrary, in most organizations, there is considerable room for
broadening the area of responsibility for individual employees without the need
for recourse to new tools. My point is only that it is not possible to broaden jobs
indefinitely without coming up against fundamental human barriers, and risk a
rapidly decreasing quality of the work in question. This is undoubtedly the
cause behind a significant part of the functional specialization in modern
organizations. Computer-based systems, however, do have the potential to
expand our effective span of competence through artificial memory, artificial
intelligence and embedded knowledge.
Artificial Memory
Even with present text retrieval systems, it is possible to offer much easier and
more comprehensive access to laws, regulations, precedents, guidelines, policy
handbooks, solutions to previously encountered problems and so on than when
relying on printed or written media alone. Future systems will improve this
further. With the fast and exhaustive information retrieval provided by advanced
computer-based systems, it should be possible to support decisions and problem
solving for broader fields of work than we can safely master without such
assistance.
Artificial Intelligence
Artificial intelligence clearly also has the potential to help stretch our span of
competence. It has already been demonstrated that expert systems can improve
decisions. Rasmus (1991), for instance, reports how an expert system introduced
by Southern California Edison incorporated the company’s policy for computer
purchases and allowed the departments to configure their own PC purchases in
adherence to the central guidelines without assistance from the DP department.
DEC’s XCON system and the successor XCEL is well-known (Walton 1989,
Heller 1991). XCEL helped DEC’s salespeople to arrive at the best systems
configurations for their customers. The pace of development here has been
slower than predicted, however—even for XCON, perhaps the most extensively
used
expert system and certainly one of the most widely reported successes—as it
was deemed necessary to have a human expert check each and every “decision”
(Long 1987). Nevertheless, there is little doubt that such systems will play an
important role in the future within particularly well-defined problem domains.
Embedded Knowledge
Embedded knowledge is not the same as an expert system. It simply means that
a computer-based system may have “knowledge” embedded in it as a part of its
data structure and its functions. A simple system for computing annuities, for
example, has embedded in it the rules for such computations. A bank clerk with
access to a system like that can advise a customer on the size of his annual
payments for a particular loan without knowing anything about how to compute
annuities himself. All computer systems have such embedded knowledge to a
greater or lesser extent, and many computer systems are therefore able to extend
the span of competence of their users.
One interesting problem turns up as more and more knowledge and rules are
embedded in systems—both in this simple sense and in connection with the
rule-based inference engines of AI programs: Work rules, regulations and even
the substance of laws may end up as embedded information in computer
programs designed to support office work. The problem is particularly
important in the government sector, where an increasing number of regulations
and even law clauses are embedded in systems used for administrative purposes.
When this happens, it becomes more difficult not only for the public to fully
understand how the laws and regulations are applied, but also for the lawmakers
to control whether their laws are actually represented correctly in the systems.
Experience has shown that you cannot always trust programmers to render law
into code and carry the lawmakers’ intentions through unscathed. We can
therefore anticipate a growing need for system auditors, people who can
scrutinize systems and see if the embedded rules are in accordance with the
regulations or laws they are meant to express. We may even see laws passed
demanding that all systems with laws or government regulations embedded
(which will include accounting systems) must store all rules pertaining to those
regulations or laws in separate tables (and not have them “hard-coded” into the
body of system code) for easy auditing. In my view, such legislation is long
overdue already.
Embedding laws in systems may also make them harder to change, because
of the limits of the systems they reside in, and simply because nobody may have
a complete knowledge of the systems involved. There
are already stories circulating about how proposed changes in taxation have had
to be abandoned or postponed because the necessary changes to the internal
revenue service’s computer systems demanded rewrites too extensive to meet
deadlines.
I have no doubt, though, that such problems will be overcome and that
systematic use of all the tools available—both artificial memory, artificial
intelligence and embedded knowledge—will make it possible to broaden the
range of tasks people can carry out with an acceptable level of quality, in some
instances considerably, and thus make it possible to eliminate even more
coordination and control activities.
Another aspect of this combination of support tools is that it will make it
possible to improve the quality of professional work overall. Appropriate
systems built on this technology should allow physicians to make better
diagnoses, judges to pass sentences that are more consistent, and caseworkers to
achieve greater consistency and quality in their work—in short, to help most
professionals to adhere more closely to professional standards. Viewed in this
perspective, this collection of tools should provide us with a much improved
version of Mintzberg’s (1979) coordinating mechanism standardization of skills,
which is defined more narrowly to standardization of explicit skills in Figure 3-
1 on p. 51. I suggest the name system-supported skills for this new, computer-
dependent coordinating mechanism (see also the extended taxonomy of
coordinating mechanisms in Figure 13-1, p. 315).
Response Assistance
Finally, systems based on the concepts behind artificial intelligence should be
able to help by suggesting responses in complex operative situations, especially
during cognitive overload or when time is a critical factor for other reasons. The
systems could even be designed to take action without “consultation” with the
human operator if the time allowed for response is so short (as it may be in an
emergency) that the human operator cannot be expected to react fast enough.
Such systems could probably prevent tragedies like the airplane crash
mentioned in Chapter 4 and disasters like the Chernobyl nuclear reactor
meltdown.
But Personal Productivity Is Not the Key

The analysis in Chapter 7 showed that the main stimulus behind
industrialization and the developments of the modern, bureaucratic organization
in the nineteenth century was the tremendous productivity increases that could
be obtained through functional specialization and the use of new tools. If a
technology-induced increase in the workers’ individual
productivity could drive this great change, it may seem natural to ask if
computer-based increases in personal productivity today could have the
potential to play a similar role. In my view, the answer is no.
The reason is that the decisive innovation in an organizational perspective
was specialization itself—not the tools that followed. The potential for
organizational change built on specialization and an increase in individual
productivity was therefore largely exhausted already by this first transition, and
even the tremendous increases in productivity that has taken place since the
modern factory was developed in the last half of the nineteenth century has not
changed the organizational principles of the Machine Bureaucracy in any
significant way. We have therefore no reason to believe that a further increase
in productivity at any isolated step in a process—even if it is substantial—in
itself should be enough to change the picture significantly. Output of the total
organization may well increase greatly, but as long as the improvement is built
on isolated achievements at single steps in the process, the organization itself is
not likely to change very much.
The basic characteristics of the technologies involved are very different,
moreover, and the keys to exploit them therefore quite dissimilar. In contrast to
the specialized machinery of traditional industry, the computer is a general,
information-processing machine that is able to adapt to an extremely wide array
of tasks. The strength of information technology is therefore first and foremost
its ability to support coordination and planning, and to carry automation
(including automated coordination) to new levels of complexity and
sophistication. Information technology should therefore be expected to affect
first of all the design and coordination of work processes and the linkages
between different tasks, and achieve its greatest effects through directing
physical processes of far greater complexity with superior efficiency and
flexibility and with much less overhead than before.
Extensions to the Constructible Space

Isolated elimination of routine jobs in itself, then, offers fairly limited
extensions in the space of constructible organizations. Nevertheless, the
potential increases in personal productivity should allow some changes. The
most important opportunities are probably connected to de-specialization and an
increased use of self-service.
Elimination of Routine Jobs

The bounded improvement in personal productivity effected through the office
tools and cell automation described earlier has fairly little to offer
with respect to new organizational structures. The main opportunity in that

direction lies in the elimination of routine jobs. We have touched upon this
already: Many of the tools in the personal support category do the kind of work
that was previously provided by secretaries and various kinds of assistants.
Routine jobs may also be eliminated by more comprehensive changes, such
as in the accounts payable function at Ford (Hammer 1990, Hammer and
Champy 1993), which will be described in some detail in Chapter 13. However,
in that case we are talking of a thorough reorganization facilitated through the
use of the coordinative power of a database, not of reductions based on
increases in personal productivity.
The groups hardest hit by elimination based on personal productivity tools
have up to now been filing clerks, typists, draftsmen, and assistants performing
various kinds of calculations. In the long run, most routine office jobs are in
danger—as their functions are either automated or eliminated. The routine jobs
with the best chance of survival are the physical or personal services: The
janitor and the cleaner will survive, for instance; we will still need some people
in the canteen, and most organizations will prefer a human receptionist to greet
and direct visitors.
De-specialization and Knowledge Support

Broadening the span of competence through the use of system-supported skills
has somewhat more to offer, since it may allow us to decrease job
specialization. Perhaps we should rather call this re-integration or even de-
specialization—to emphasis that we are now able to alleviate some of the
problems that job specialization created in the first place.
De-specialization is not a universal option. It builds on two pillars: easy
retrieval of information on the one hand, and embedded knowledge and AI on
the other. These tools primarily support de-specialization of jobs that require
people to collect information from many sources for further processing, or for
use in decision making on the basis of law, rules, or regulations—the archetypal
bureaucratic kind of job after Weber’s definition. The important aspect of de-
specialization is that it, by reducing the number of steps in the work process,
also reduces the need for information transfer, one of the most time-consuming
activities in any large office, and a major source of errors and
misunderstandings.
The main reason that functional specialization met with much less success in
the office than in the factory can be found precisely in the much higher volume
of information that has to be transferred from person to person as part of the
work process there. In the factory a piece of hardware coming down the
assembly line embodies most—if not all—of the information needed by the
workers. The information is absorbed quite
literally at a glance, and, consequently, one attracts very little penalty—if any—
in the form of increased time for information transfer when one increases
functional specialization.
A transaction so simple that it only needs to be registered or stamped can be
processed in much the same way in a white-collar “line.” As soon as it becomes
a little more complex, however, requiring some kind of assessment and decision
making (what might more readily be termed a case), it will normally be
accompanied with a lot of written information—usually both the basic
information collected at the outset as well as the information accumulated while
it has been passed along the various steps in the work process. Often, there will
also be a need to transfer informal, oral information.
Any increase in functional specialization in the office will therefore normally
incur a considerable overhead in the form of information transfer. Not
infrequently, absorbing all the relevant information and making sure that one
understands it correctly takes longer time than doing the actual work. As noted
earlier, numerous information transfers also create ample opportunities for
errors, misunderstandings, and loss of information. We have probably all been
victims of such mishaps in our encounters with bureaucratic structures. Indeed,
many of us have been guilty of producing them as well.
Reducing the number of information transfers in an organization will
therefore contribute greatly to its productivity, especially since interpersonal
communication itself is so difficult to make more efficient. As our previous
analyses have shown, this is the most recalcitrant of all our innate constraints
when it comes to tool support. Despite all our gadgetry, it takes about the same
time to transfer information from one mind to another today as it did a hundred
years ago—and, if we talk about people at the same location, it takes the same
time as it did 10 000 years ago.
How far can the concept of de-specialization be developed? Can we, for
instance, imagine computer-supported superprofessionals covering many
disciplines, or supermanagers taking over the responsibilities of entire present-
day management teams? What the distant future will bring is not possible to
foretell, and history has taught us not to try. In the foreseeable future, however,
such a scenario is simply impossible, because the knowledge that can be
embedded in systems, even in AI-systems, is mainly of the “hard” kind: simple
facts, or pretty simple if-this-then-that rules. Even advanced AI systems are
extremely limited compared to a human mind.
All our “softer” knowledge; our experience; our tacit knowledge; our ability
to interpret facts from a context and previous experience; our ability to discern
the important from the unimportant, to judge and weigh information and
decision alternatives is impossible to embed or
mimic in a system. In a professional and managerial position, extensive

experience and background knowledge is always needed to respond sensibly to
problems or execute tasks in a satisfactory way. Even if we could build a system
that would allow persons without such experience and background knowledge
to respond adequately in many or even most instances (a daunting, but perhaps
not impossible task in certain circumstances), they would be at a complete loss
when more complex situations arose. Or, even worse, they might think—
erroneously—that they had a good answer and then happily execute it, since
they did not know enough about the implications to understand their own
shortcomings.
The limiting factor, then, for integrating professional and managerial jobs is
not so much the nature of the tasks themselves, but rather the extent of the
knowledge and experience that is necessary to fully understand their
implications. For some jobs (for instance, in sales) the number of personal,
external contacts that must be maintained is also a limiting factor on the number
of functions one person can shoulder. Everyone with experience in sales
activities knows that personal contact is extremely important and that it cannot
be totally supplanted by more “efficient” computer-mediated, semiautomatic
communication (except for fairly inexpensive items).
Self-Service
However, the opportunities range further than this. Of particular interest is the
possibility of offloading tasks onto the customer, thus removing it from the
organization altogether. Supportive systems with elements of AI and/or
embedded knowledge may allow for much more extensive self-service than we
have been used to. Automatic teller machines have already introduced us to
personal support systems that allow us to complete some kinds of bank
transactions ourselves. The types of transactions that have been made available
for self-service so far have been few, but there are more advanced machines
(and systems) coming up, and the concept should be possible to develop to the
point where the bank itself all but ceases to exist (we shall return to that
particular case in the next chapter).
There are doubtlessly a large number of areas where computer-supported
self-service will surface. Both airline tickets and other tickets are already sold
this way, betting systems should be eminently possible, and insurance (at least
some kinds, and more advanced than the travel insurance you can buy from
vending machines at some airports) is a product that should also lend itself to
similar self-service systems. The filing of applications for various purposes is
another area open for computer-supported self-service solutions.
Although self-service is a phenomenon all by itself, it can also be seen as an

aspect of de-specialization, since the logic behind both is the same: Systems
guide us through; because they “know” how things are to be done, they help us
get at the necessary information, prompt us for our contributions, and then
perform the transactions on our behalf. The organizational effects may be
profound—a lot of specialized jobs will disappear because customers take over,
and large parts of existing organizations may be eliminated.
Conclusions
The various improvements in personal productivity discussed earlier have
already made possible changes that have had significant impact on
organizations, and more is bound to come. As personal productivity continues to
improve and cell automation and self-service proliferate, organizations of all
sizes will be able reduce their payrolls further—at least in the parts of the
organization where the improvements are implemented. This is of course not the
only source of workforce reductions—it is not even the most important one, as
we shall see later. However, it will allow for significant reductions. A reduction
in the number of employees will also allow organizations to reduce the number
of administrative layers somewhat—in particular, de-specialization should
contribute to this.
However, the changes are relatively simple: by and large, they consist in
workforce reductions. Even if de-specialization may involve an integration of
jobs and thereby a marked reduction in the need for information transfer, it
provides no particular platform for really inventive organizational changes.
There are no genuinely new principles involved—the IT-based advances in
personal productivity mainly represent improvements and extensions of the
development process started in the eighteenth century.
Granted, the improvements are dramatic in some respects and may foster
significant local changes in many organizations, as when typing pools are
dissolved, assistant draftsmen made superfluous and jobs broadened. The
improvements in productivity can even be said to be of epochal proportion in
quite a number of scientific and engineering disciplines. Organizationally
speaking, however, they do not significantly expand the space of constructible
organizations, nor do they build significant pressures for evolution in totally
new directions.
Groups and Teams

All organized activities are instances of cooperation, and, in that sense,
cooperation can be thought of as more or less synonymous with
organization. When talking about cooperative work in connection with the use
of information technology, however, it is the team and the work group that is in
focus. The discussion in this section will therefore be limited to that level; a
group small enough to let each member have more or less direct contact with all
the other members.
I have quite intentionally made a distinction here between the concepts
“team” and “work group.” Although these expressions are frequently used as
synonyms, at least in everyday speech (I often do so myself), they have
distinctly different connotations in a more precise theoretical context. A group
or work group is a fairly loose term, designating any relatively limited number
of people who work in conjunction with one another for a common purpose. A
team in its more precise sense is a small and tight-knit group with a common
purpose, a strong sense of commitment, and a genuinely shared responsibility
for the outcome of their work (Katzenbach and Smith 1993). It is this genuine
commitment and shared responsibility that serves to distinguish the true team
from the work group, not a particular way of working. Since many authors are
less stringent about this term, however, and everyday use is far less rigorous, I
will use the word “team” fairly broadly in the following. Regardless of
definitions, moreover, teams and work groups should have the same kind of
needs for coordination and work support and thus reap the same benefits (and
share the same problems) from using information technology.
The use of computers in group support is one of the aspects of computer use
that receives most attention as we make the transition from the twentieth to the
twenty-first century (together with multimedia, the Internet, networking and the
concept of “electronic highways”). In my view, this attention is not warranted
by its actual contribution to organizational transformation and efficiency (this
goes for multimedia and the Internet as well). It is, however, easy to understand
why it arouses so much interest: It talks directly to our primate, emotional side;
it is all about humans being human together, rather than being machine-like
parts in a machine-dominated organization. Using a term from organization
theory, Cooperative Computing and the development of groupware may in
many ways incarnate the human relations movement of the computer scene.
Cooperation among humans is almost synonymous with communication. The
exchange of views and ideas, the transfer of information, and working out
decisions and making them known involve copious amounts of communication,
with meetings as the main instrument. People who continuously grumble about
“all the time thrown away in meetings” only demonstrate that they do not
understand the nature of human cooperation or the burden of coordination
placed upon us by any organized
activity. They may simply believe in commanding instead of cooperating, and

thus feel no need for advice, for discussions, or for building motivation.
There are of course good and bad ways to conduct meetings, and in many,
more work could be accomplished in less time. Some meetings are undoubtedly
even unnecessary—but I also know about a good number of necessary meetings
that were never held, to the detriment of the organization in question. The fact is
that any organized activity will require meetings, and, the more dynamic the
situation is, the more meetings will be required. It is no accident that the
supreme military commanders in critical situations or during major offensives
meet continuously during the most intensive phases to coordinate the efforts of
their respective services—they are not locking themselves into their separate
offices to do “real work.”
It is no wonder, then, that much of the work being done in the area of
cooperative computing involves either support for face-to-face meetings or tools
for electronic meetings. Johansen (1988) even concludes that such efforts can
best be categorized according to their support for meetings or activities related
to meetings. I would like to propose another classification scheme, however:
meeting support, work support, and infrastructural support.
Meeting Support
Meeting support involves both systems to support face-to-face meetings and
systems designed to allow fully “electronic” meetings by way of computers.
Work on such systems started quite early in the computer era—for instance,
quite a lot of the original work of Douglas Engelbarth, the “father” of
groupware and graphical user interfaces, involves systems for the enhancements
of meetings. Work on support systems for face-to-face meetings and
“electronic” meetings started at about the same time.
Meeting Support Systems

The main approaches to meeting support systems have been various forms of
electronic white boards and group decision support systems. The aim has been
to provide tools for better structuring of meetings, easier integration of
contributions from the participants, and better documentation of the results.
The results so far have been fairly meager. It is difficult to make tools that
truly contribute to real-time, human communication processes, and it is even
more difficult to make them so easy and intuitive to operate that they are
adopted for use outside the rather narrow circle of groupware
developers and enthusiasts. I think there are potentials for improvement here,
however, and I believe that the electronic white board will slowly develop into a
useful tool—but that development will take many years. Perhaps the most
important initial contribution will be the ability to retrieve and display
information from corporate and external databases, to provide a common
platform for discussion, and to quickly satisfy needs for ad hoc information that
may arise during meetings.
Most of the development in this area has been devoted to meetings in
administrative environments that work mainly with language (text)—that is the
case with all the systems described in Johansen (1988). This can perhaps be
explained by the background of the researchers, their institutional settings, and
the all-too-common preoccupation with the problems of top management.
However, the analysis of where the computer contributes most to enhance our
own abilities suggests that the potential for useful meeting support tools should
be much greater in the professions already working with highly graphical
applications, such as printing, advertising, architecture, and engineering design.
Electronic white boards in those environments, in the form of large-screen,
common workspaces, could serve as very important productivity instruments for
design groups. And if, in the future, one succeeds in harnessing more advanced
graphics and animation for the display of more administratively oriented
information, the electronic white board may gain in importance even here.
Electronic Meetings
Among the tools for meeting support, those meant for supporting fully
“electronic” meetings have aroused by far the most widespread interest. They
may involve telephone, video, computer conferencing, and screen sharing.
Screen sharing means that everybody participating in the meeting can see and
access the same picture on their displays. The focus of development lies in the
direction of video conferencing, preferably combined with screen sharing. With
a sufficiently large screen and sufficient bandwidth on the transmission lines, it
would then be possible to conduct workgroup meetings onscreen. One part of
the screen could be occupied by the live pictures of the participants; the rest
could be available to material for presentation and manipulation.
Simpler solutions may involve telephone conferencing with screen sharing or
real-time computer conferencing (with all the participants on-line at the same
time). Traditional computer conferencing, in which participants log on at
different times and keep exchanges going for days and weeks, does not seem a
viable tool for meetings, but for easy exchange of written
statements and expositions—it is more like an electronic journal or bulletin

board.
The analysis in the previous chapters indicates that such solutions will be a
good tool for groups where the participants know each other, and will make it
easier to maintain cooperation in spite of geographical separation. It also
suggests that the improvements will be especially important for those who work
with strongly graphical applications—geographical separation is more of a
handicap for them than for groups working mainly with text and numbers.
However, the need for colocation and face-to-face meetings cannot be totally
eliminated—at least not yet.
Work Support
Meetings are certainly indispensable to coordinate and reach decisions. They
also quite often function as problem-solving groups. Thus they encompass most
aspects of group work. However, groups do not only work when they meet, the
members also work by themselves on their part of the group assignment. Most
of that work is probably accomplished with the help of various personal support
systems, but, in addition, groups need tools that provide a common framework,
and help integrate the various contributions. Electronic mail, conferencing
systems, and group authoring programs (programs supporting the production of
joint documents or other forms of joint information presentations) are such
tools, along with common databases.
A typical example of computer-supported group work was quoted in Chapter
9 (from McKersie and Walton, 1991), where telephones, a conferencing system,
remotely accessible databases and word processing were used to produce a joint
document with a number of remotely located managers. As we noted, that
document could doubtlessly have been produced without computers, but
probably with lower quality. More advanced tools such as video conferencing
and screen sharing will further increase the edge that computer-based systems
will give over pre-computer tools. Even here, however, I think the potential is
greatest for work involving strongly numerical and graphical applications, such
as engineering design (CAD). By having a common database representing the
total object to be designed as basis, a true design-group–oriented CAD tool
should offer both full coordination and coherence of the overall design
parameters and of the interfaces between modules, while, at the same time,
allowing the individual designer to work on his/her part of the assignment.
When fully developed, however, such a system becomes much more than
groupware—it becomes a very complex system for coordinating the work of a
total organization or even many organizations. We shall return to this subject in
Chapter 13.
Although workflow tools do not, strictly speaking, belong to the groupware

class, all the vendors insist that they belong there. They probably do so because
groupware as a concept is very much in vogue, and also because they do not
have very many other products to include under the groupware heading, except
for calendars and email systems. Workflow tools are meant to manage work
processes and to speed work along from point to point in the process, and they
also take care of some of the most routine aspects of that process (Thé 1995).
As such, they build on a rather traditional approach to office work, aimed more
at speeding it up than changing it. The underlying model is still that of a chain
of individual caseworkers each doing an incremental part of a total task. The
products provide instant transport of all electronically stored material between
caseworkers and also make it much easier to monitor progress and to find out
where in the process a particular case is at any particular time. Householding
functions save time for caseworkers with chores such as filing and routing.
Workflow tools can undoubtedly increase productivity in most procedural
environments, but in the process they tend to cement existing routines and
inhibit more creative solutions built on automation and elimination of tasks.
Infrastructural Support
To function properly, groups require a certain infrastructure. By tradition, we
would say that they need office space for work and meetings; they need a
common “memory” in the form of files and archives; they may need support
personnel of different kinds. We would also prefer groups to work at the same
location and in adjacent offices, not only because of the efficiency (easy access
and no travel time), but also because colocation is generally deemed necessary
to build the team-spirit that is so important for successful teams. Friendship and
team spirit need a certain volume of interaction to grow, preferably ample and
regular informal contact. The close proximity of a common work area is
required to achieve that.
Computer-based systems can improve such group infrastructures in several
ways. The most obvious one is perhaps the database—when everyone have easy
access and can use the same information as a basis for their work, the general
coordination of a group is automatically improved. Group calendar systems can
also be of help if members spend much time away from a common office and
have difficulty keeping tabs on each other to arrange meetings, etc. Computer-
based project management systems represent an improvement over earlier tools
and can significantly increase the flexibility in larger projects when it comes to
tackling unforeseen events and changes in plans and priorities.
Then there is the issue of personal communication. When analyzing the

merits of electronic communications in Chapter 9, I concluded that the new
channels still did not measure up to face-to-face contact. However, I also
concluded that video conferencing would allow us to reduce the number of
required face-to-face meetings and could function quite satisfactorily in many
instances, especially for people who already knew each other well. When we
now discuss the subject of group infrastructural support, it is interesting to
consider the possibilities for using electronic communication to improve team
building in teams that cannot, for various reasons, work in the same place.
The attempts that have been made to create electronic “spaces” for
spontaneous and informal communication have not been very successful
(Johansen 1988), however. The technology has been too constraining to allow
for the casualness required for successful informal interaction to develop. Small
screen formats, mediocre sound quality, and the high price of video
communication have combined to limit the usefulness of video conference
systems. Computer conferencing systems and bulletin boards have actually
shown some capacity for creating and maintaining electronic “cliques,” but
mainly among young people or others with time to spare. The trouble is that
communication is a time-consuming affair, and, no matter the quality of the
channel, communication with more people subtracts more time from the time
available for “real work.”
With the greater availability of bandwidth that is bound to come, possibilities
for “virtual groups” may improve, however. With sufficient bandwidth we can
have not only videophones, but video-wall rooms (let us nickname them
vidwams) where one or more walls consist of a high-definition video screen
showing a corresponding room in another location. With sufficiently advanced
sound systems, it should be possible for geographically separated groups to
achieve a fair semblance of the experience of actually being together in the
same room.
Such rooms should also contain or give easy access to video cubicles (let us
call them vubicles) where single persons or small groups could sit down to have
a closer chat with someone at “the other side.” Vidwams could serve partly as
relax areas, where people could come in to see if anyone was there, partly as
meeting rooms or rooms for presentations—or may be even as canteens. It is
likely that the effect of seeing and hearing even in such an electronically
mediated way would help build stronger ties and loyalties than geographically
separated groups could otherwise achieve. Let us just remember that the time
constraint, mentioned earlier still apply: vidwams will not make it possible to
work closely with significantly more people than before—even if successful,
they will only allow about the same number of people to work closely together
in spite of geographical separation.

Could group support tools then give rise to new ways of group cooperation, or
even new organizational schemes? Many people seem to think so—especially,
of course, proponents of groupware and what is termed computer-supported
cooperative work (CSCW). Greif, for instance, contends that (1988, p. 6):
“CSCW research is examining ways of designing systems—people and
computer systems—that will have profound implications for the way we work.”
It is not entirely clear what those implications will be, however. The
speculations cluster around a small number of themes. The main argument is
that computer-based tools will increase group productivity and creativity
through improved coordination and communication, by allowing groups to do
more work in real-time group mode than before (direct work on the same
screen, for instance), and by providing better support and structure for the work
done individually. It is also believed that this improvement in group
productivity and increased communication capacity will result in increased
emphasis on teams, more horizontal communication and thereby flatter and
more democratic organizations (see, for instance, Drucker 1988, Johansen 1988,
Greif 1988, Keen 1991, O’Hara-Devereaux and Johansen 1994).
This is by no means clear. Johansen, whose account of groupware products
and research is very factual and realistic, makes his own reservations in the
introduction to his book Groupware (1988). After underlining that his
conclusions “lean toward the upsides of groupware,” he lists his concerns—
among them that there may be too many meetings, overdone teams (too many
participants without reduction in other responsibilities), increased control over
team members, too much structure, and a tendency for people to only join teams
that use the systems they know. Rockart and Short (1991), on their part,
question the long-term implications of a disintegration of usual structures and
reporting relationships.
It should come as no surprise that I also belong to the skeptics in this area. I
do not doubt that computer-based systems can provide valuable tools for
improved group coordination and communication and, in some instances,
increase productivity significantly. But I do not think that the improvements
will be sufficient either qualitatively or quantitatively to transform group work
or make groups and teams so much more effective that organization structures
can be radically changed.
As mentioned earlier, the gist of group work is communication—sharing
information and views, building a common understanding of problems, and
forging common decisions. These are exactly the areas where the analysis in
Chapter 9 suggests that computers can contribute the least. To comprehend the
ideas and thoughts of other people, we still
have to listen and read, and it matters little how fast information can be
transferred from computer to computer when the time spent to write, read,
speak, and listen does not change. We still need to reflect upon that information
at our own pace to respond sensibly to it. Working in groups simply takes a lot
of time and effort, and it does not seem very likely that we can escape these
fundamental constraints in the foreseeable future.
I do not mean to suggest that computer-based systems will be entirely
without impact on the way groups work. Neither do I think there will be no new
opportunities for organizing. But I do think that the effects in this area will be
much less dramatic than the CSCW enthusiasts predict. The kind of
improvements in group productivity delivered by systems such as coauthoring
systems, meeting assistance, electronic mail, and group calendars will not
change the nature of the tasks that groups can tackle in any significant way.
They will only allow groups to become somewhat more efficient, produce work
of somewhat higher quality, and function more independently of physical
proximity.
Moreover, an increase in the productivity of groups and teams in itself should
not have too much of an impact on organization, aside from strengthening the
general trend toward reduced manpower requirements. The production aspect of
groups and teams really only gets exciting when we transcend the boundaries of
the group itself and look toward integration within the total organization—as
when engineering design groups work through systems that link their work
directly not only to other design groups, but (through a common database) also
to the groups working with production, sales, and distribution.
For the organization as a whole, the reduced dependence on colocation may
turn out to be the most important opportunity, especially if effective means
(such as vidwams) can be found to foster real team spirit and the development
of a common organizational culture across geographical divides. That would
make it easier to exploit other aspects of computers as well. For example, many
companies with geographically dispersed manufacturing operations could
benefit considerably from a close coordination of those operations—to the point
of running them as one integrated factory based on integrated computer systems
encompassing both sales, production, and warehousing/distribution. However,
the spirit of independence of such plants will mean that they will often resist
close coordination with sister plants under the auspices of what is perceived as a
remote and faceless division management in the division headquarters. If
vidwams, videophones, conference systems, and electronic mail could help to
establish cross-plant work groups and a primary-group identification among key
personnel from all the plants as well as the headquarters, such an organization
would be much closer to succeeding with tight coordination than before. The
same technique could possibly be used to
build cross-company loyalties at select levels in multinational companies,

supporting more direct coordination efforts built on other aspects of computer
use.
To achieve this is not easy, however. Even systems such as those discussed
here do not make it possible to increase the total communication volume very
much. The people who participated would therefore either have to reduce their
communication with the people they work with locally, thus putting some of
their local relationships at risk, or significantly increase the amount of time they
allocated for communication. To use expressions from network theory (Lincoln
1982), computer-based systems would not make it possible to significantly
increase neither the density nor the connectivity 1 of organizational networks,
although the links could possibly be others, span greater geographical distances,
and thus change the structure of both inter- and intracompany networks. It is
therefore doubtful whether the use of information technology will be the
decisive factor that allows team-oriented organizations to achieve flatter and
more democratic structures. Computers can help flatten the organization, but
not because of groupware—it is rather the systems for personal support
discussed in the preceding section and the automatic coordination coming up in
the next chapters that will provide the major impetus in that direction.
It is also quite doubtful whether it was the prospects of groupware and
computer support that arose the interest and faith in teams in the first place. As
Johansen (1988) quite correctly notes, the increased emphasis on team-oriented
organization probably has reasons other than computer technology. Johansen
himself points to deregulation, the trend toward contract work, increasing
geographic spread for companies, and the declaration of team-oriented
companies as models for the business world.
There may even be more basic reasons for the emphasis on teams and the
interest for groupware, however. Both the debates about teams and the
development of groupware are products of the academic and professional
communities (which include most managers), which have a well-developed
propensity to prefer teams. Their educational backgrounds have accustomed
them to professional discourse as an indispensable tool
1The density of a social network is a measure of the number of links between the nodes (nodes can
be persons or groups, depending on the level of analysis). It is calculated as the ratio of actual ties to
potential ties. The connectivity denotes the degree to which nodes are linked to other nodes either
directly or indirectly (through other nodes). A maximally dense network (for instance, a group
where every member has a direct link with every other member) will also have a maximum
connectivity, but a fully connected network (for instance, a hierarchical organization where
everyone is linked to the top manager through their bosses) can have low density (as when no one in
a hierarchy has links with people other than his or her boss.)
for developing ideas and solving complex problems, and their jobs more often
than not require them to work in close cooperation with colleagues in their own
profession as well as people in other professions. Most of them like discourse-
rich environments; they want to work in groups, and would quite naturally like
to see the team concept gain ground—hence a fascination for groupware.
By this, I do not mean to say that groups and teams are not important, or that
professional discourse is superfluous—on the contrary, discussions are indeed
indispensable for much of the work that professionals do, such as planning,
product development, business development and administration and problem-
solving in general. Let us also not forget that after the family, the team is
probably the oldest and most basic organizational structure we have. But the
feeling that teams are all-important and an answer to most of the problems of
contemporary organizations may reflect just as much the local work
environment of the team champions as the functioning of large organizations in
their entirety. To the extent that the team concept is growing in importance
throughout the organization, as indeed seems to be the case, this growth could
also be an artifact of progressing automation and elimination of routine jobs:
automation largely passes by jobs of the kind that professionals use to have, and
thus increases the proportion of team-oriented work in the organization even if
the absolute number of people working in teams remains the same.
Consider, for instance the control room operators in the fully automated
paper mill mentioned in Chapter 10. To tune the factory and squeeze maximum
production and the desired quality out of it, operators can no longer work in
isolation: they may have to team up with both process engineers, product
specialists, and marketing people. In that sense, the workers have been changed
from isolated operators responsible only for discrete steps in the production
process to team members with joint responsibility for the total result. Thus, it
seems to constitute another example of the transition from hierarchical,
command-chain organization to a team-oriented approach in industry.
Appearances are deceptive, however. A closer analysis also suggests that the
job as control room operator is not a continuation of the earlier manual work
but, rather, an (incidental) appropriation of the work of the production manager
and his immediate subordinates—a group of people that has always had to
function more or less as a team. It is thus not a question of the transformation of
jobs or job roles: the former jobs of the control room operators—when they
controlled the discrete production steps in the factory—have simply been
eliminated through automation, and the operators themselves have been thrust
upon a totally new set of duties and responsibilities, requiring an entirely new
set of job roles—resembling rather closely that of the production management
team.
The sum and substance of this is that teams are important because they are
superior instruments for coordination and problem solving, and, as more and
more routine work is eliminated, because a larger and larger proportion of the
work that is left in the organizations will be the kind of work that requires
cooperation and teams. Paradoxically, therefore, computers may enhance the
importance of the work group and team not through support for them, but
rather by eliminating most of the jobs that do not belong in groups or teams in
the first place. Then, of course, the technology may also facilitate group work
and make groups and teams even more useful and flexible than they used to be.
277
12 Routines and Automation
“The machine yes the machine

never wastes anybody’s time
never watches the foreman
never talks back.”
Carl Sandburg, The People, Yes, 1936
Automation—The Cornerstone of Computing

Routine Automation
Information technology has the potential to let us attack all kinds of tasks that
involves handling and processing of information. In general, we may say that
most work will be touched by information technology, at least as a supportive
tool, and many types of tasks will be wholly or partly automated or eliminated,
since the programmability of computers has given automation a strong, new
boost. The more routine the task is, the easier it will be to dispose of, but
computer-based automation will continue to evolve over the coming decades—
much as mechanical automation and the use of energy sources such as water,
coal, and oil were developed during the nineteenth and twentieth century—and
is likely to reach levels we cannot even imagine yet.
The first and most basic application of computers has been just to automate
simple routine tasks. This is still the dominating way to use computer-based
systems—from word processing (automating important parts of the tasks earlier
associated with the production of typed or printed text) to accounting
(automating the arithmetic and most of the reporting) and claims processing in
the insurance business (automating a great deal of filing, writing, and control).
Such automated routines are in fact the most important part of any computer-
based system—even those that seem to concern quite different matters. A
program for finite element analysis, for instance, helping engineers to decide on
the optimal form and
thickness of mechanical parts, works simply by repeating programmed routines

based on mathematical formulas; a CAD system draws, redraws, fills in
surfaces, and adds shadows by doing just the same. Computer-based systems
will also often direct and even pace the work of their users. Typical examples of
this are systems for caseworkers in large white-collar bureaucracies, such as the
system for dental claims processing in a large insurance company described by
Zuboff (1988), which is briefly mentioned in Chapter 10.
Computer-based systems thus generally incorporate explicit routines on two
levels: the closed routines “hidden” in the applications program’s internal
functions and the open routines that incorporate the dialogues with the users and
structure their work. Quite often, in the heated discourse about the wondrous
feats that information technology can pull off, and its great potential for society,
organizations, and individuals, we seem to forget that in the end it all boils
down to this: routines consciously designed and programmed by real humans,
and dumb machines that ultimately derive their great powers from their
immensely fast execution of these routines.
The creation of such programmed routines is obviously a development that
falls within the bounds of the basic coordinating mechanism that Mintzberg
(1979) calls standardization of work. Its immediate forerunner is the explicit
routine, which was developed on the basis of writing and became the main
coordinating tool of the modern organization. I have positioned programmed
routines accordingly in Figure 13-1 on p. 315. However, in my view represent
an advancement in relation to explicit routines that is greater than the original
development of explicit routines and the blueprint of the modern organization.
The traditional use of explicit routines requires that the workers learn all the
routines belonging to their task, or at least learn those that are used most often,
and remember when and how to retrieve the others. Experience shows us that
only routines that are thoroughly learned (internalized) are systematically used
in the daily work situation; others may be overlooked, forgotten, or fall into
disuse for various other reasons. The process of renewing or changing routines
is also difficult, because it requires workers to “actively forget” the old routines
and thoroughly learn the new ones. Since the number of routines that can be
retained as active in a work situation is fairly low, the repertory of any one
organization member will be naturally limited, and the capacity for branching
(alternative routines) will be low.
When routines are programmed into computer-based systems, the situation is
quite different. First of all, a significant number of routines can be automated
completely. Second, the routines that enter into the user dialogue can be much
more numerous and diverse, since the user does not
12 Routines and Automation 279
have to remember them all actively, but just how to operate the system and
relate to the dialogue. This can be compared to the difference between our
active and passive vocabularies—which is quite significant, as anyone who has
learned a second language will know. In addition, the system can incorporate
assisting features giving users a broader span of competence.
A good example is the system developed at IBM Credit (Hammer and
Champy 1993). IBM Credit finances the computers, software, and services sold
by IBM—it is a profitable business to IBM, and quite large: if independent,
IBM Credit Corporation would rank among the Fortune 100 service companies.
Prior to redesign, each application for credit went through a five-step procedure,
taking on average six days to complete before a quotation letter could be
delivered to the IBM field salesperson who had requested it in the first place.
During these six days, the deal was still vulnerable for several reasons: the
customer might obtain financing elsewhere, fall prey to another computer
vendor, or even cancel the acquisition altogether. The pressure to reduce the
turnaround time was therefore considerable, and it was also highly desirable to
reduce the number of calls from impatient sales representatives wondering
where their customer’s application was sitting.
A closer look revealed that the actual work on an application averaged only
90 minutes—the rest of the time it was either sitting on a desk or was on its way
between the five desks it had to visit before completion. A total redesign was
then undertaken, where most of the applications were completed by a single
caseworker, supported by a new computer system. How? Hammer and Champy
explain (Hammer and Champy 1993 pp. 38-39):
How could one generalist replace four specialists? The old process design was, in
fact, founded on a deeply held (but deeply hidden) assumption: that every bid
request was unique and difficult to process, thereby requiring the intervention of
four highly trained specialists. In fact, this assumption was false; most requests
were simple and straightforward. The old process had been overdesigned to handle
the most difficult applications that management could imagine. When IBM
Credit’s senior managers closely examined the work the specialists did, they found
that most of it was little more than clerical: finding a credit rating in a database,
plugging numbers into a standard model, pulling boilerplate clauses from a file.
These tasks fall well within the capability of a single individual when he or she is
supported by an easy-to-use computer system that provides access to all the data
and tools the specialist would use.
IBM Credit also developed a new, sophisticated computer system to support the
deal structurers. In most situations, the system provides the deal structurer with the
guidance needed to proceed. In really tough situations, he or she can get help from
a small pool of real specialists—experts in credit checking, pricing, and so forth.
Even the handoffs have disappeared because the deal structurer and the specialists
he or she calls in work together as a team.
Hammer and Champy claim that IBM Credit in this way cut the average process
time to four hours and increased the number of deals handled 100 times without
any increase in workforce.
The IBM Credit case represents a quite innovative use of computers to
redesign the work process, it is not exactly an implementation of a run-of-the-
mill administrative system. However, even in more commonplace systems,
which often represent little more than an “electrification” of older, manual
routines, we can usually find traces of all the strong points of computer-bases
systems exploited by IBM Credit: automation of simple routines, implicit
structuring of work, embedding of rules, and support for decisions. As the
sophistication of the user organizations as well as the system vendors and
software developers grows, we can expect them to increasingly take advantage
of the more advanced possibilities.
What and How Far Can We Automate?

Automation has proved to be a very powerful approach for increasing output
and improving an organization’s competitiveness. Especially in material
production, it has been the most important determinant of organization for the
last 150 years at least. Therefore, we can expect organizations in general to
continue to explore the possibilities offered by automation, and to seek to
increase their output per employee. In my view, the potential is still great, and
just as great—if not greater—in the white-collar as in the blue-collar sector.
Computers are new as human tools, and it stands to reason that we are only in
the beginning of a long and exiting development. If the technology’s history so
far has any predictive value at all, the coming decades (and even centuries) will
see continuous, rapid improvements both in the basic technologies, in available
hardware products, and in application systems that will consistently dwarf
earlier achievements. Both in the factory and in the office, our efforts to
automate work have just started.
It is not easy to define the kind of work that will be automated into oblivion
and which tasks that will survive—our present knowledge and experience
provides a meager model for extrapolation and our imagination is a guide of
dubious merit when we speculate about the possibilities in the longer run.
However, with due caution, the possibilities seem to be greatest in three broad
areas:
1. Material production (especially factory production).

2. Immaterial production and services—any product or service
that mainly consists of information, information processing or
information procurement.
3. Internal administration in all trades.
In spite of 150 years of improvement, there are still massive opportunities for
increased automation in factory production—indeed, considering the immature
nature of information technology (compared to the long history of mechanical
technology), we have barely scratched the surface. Still, we tend to automate on
the conceptual basis of mechanical engineering, and the most astounding
innovations, in my view, await the development of production methods that are
natively dependent on a copious use of processing power.
So far, we have probably come farthest in this direction in the process
industries, where we have also seen some of the most spectacular improvements
in productivity over the last couple of decades—whole production units such as
refineries or paper mills have been totally automated. Great strides have also
been made in mechanical industries, however. As early as in the late 1970s, for
instance, Fujitsu built a metalworking factory not far from Mt. Fuji that covered
20 000 square meters, employing 82 workers on the day shift and only one
control room operator during the night. His only task was to surveil the working
industrial robots and automatic machine tools from a central control room. A
traditional factory of the same size would have employed almost ten times as
many people, and there were still plans for reductions at that time (Hatvany et
al. 1985).
If we turn away from material fabrication, the possibilities for automation are
generally excellent in almost any business that deals mainly in information,
especially when we include computer-supported self-service as an aspect of
automation. The most extensive automation can be achieved when the
information is structured, and especially when it is quantitative. Banks are prime
examples of such businesses. They have been the subject of major changes over
the last decades, and there is more to come. I shall elaborate somewhat on this
in a moment.
For businesses such as newspapers and publishing houses, often hailed as the
archetypal information-mongers, we must distinguish sharply between the
editorial side and the distribution activities—of which printing has been (and
still is) the central part. Writing and editorial work is highly labor-intensive and
it will have to remain so (even if it can be computer supported through the use
of word processing and the like). On the distribution side, printing is already
highly automated, but electronic channels and media do offer the possibility of
further automation. However, drastic changes here will require the customers to
change their habits and to leave paper as the preferred medium. I think this will
happen more slowly than many enthusiasts believe, and the reason is simply that
screen technology still falls far short of the portability, comfort, and ease of
reading offered by printed media—and it is likely to do so for a good number of
years yet. The exception is the kind of concise factual information that up till
now has been found in dictionaries, encyclopedias,
directories, news clippings, and the like, and where reading comfort is not a
very important issue. For this kind of information, digital media have already
gained an important position, and may rapidly achieve dominance.
Leaving the subject of single lines of business, there are also significant
automation possibilities in internal administration in all kinds of trades. The
function where computers are most widely applied is probably accounting,
where it has lead to considerable staff reductions. However, even areas such as
the administration and use of customer data (as in insurance companies), sales
(on-line order registration and semiautomatic fulfillment, or even self-service
ordering over Internet or through customer terminals), and logistics (automatic
restocking through point-of-sale registration, etc.) have been the focus of much
change.
The prospect of automation in the office has been the subject of much
discussion. Arguments have centered on whether general office work can be
automated at all, and many research projects have concluded that such work in
its very nature is too unstructured and dependent on human judgment to allow
significant automation. I shall return to this discussion later in this chapter, and
will only say at this point that I think the possibilities are far greater than many
people would like to think. However, the development of automation will
happen in close interplay with the development both of information technology,
other technologies, and methods for analysis and design, and it is not possible to
predict the developments very far into the future.
It is perhaps easier to say something about where the opportunities seem to
be most slender—to point out the work that depends too much on the human
faculties computers cannot mimic (at least not yet), or that require physical
skills and dexterity machines cannot match. There seems to be three broad
classes of such work:
1. Work where judgment and creativity are central—for instance,

research and development, design, policy making, journalism,
artistic work, and management other than routine supervision.
2. Work where human physical dexterity and skill are paramount,
as in handicrafts, the performing arts, domestic work, and
chauffeuring. Some jobs are safe because we want humans to
perform them, as in handicrafts and the performing arts; other
remain safe because they are (at least for the time being) very
difficult to automate, such as much domestic work, transport,
and repair work.
3. Work where the emotional component of dealing with a fellow
human is important, such as psychiatry, much sales and service
work (especially personal services such as hairdressing or

waiting at tables), and teaching.
Many jobs have components from more than one of these classes—waiters
also depend on their dexterity to do their job, and craftsmen often combine skill
and creativity. Work such as nursing combines all three. When a job scores high
on one or more of these properties, it means that it depends on human qualities,
and the incumbents will be difficult to supplant by nonhuman agents or
automatic procedures.
A word of caution is warranted, however. Even if we may think so, the
human aspect of a job is not always the most important to us. We readily forgo
the social pleasures of exchanging everyday niceties with a bank clerk in order
to retrieve money faster and more conveniently from an automatic teller
machine, and we have been swift to prefer the low prices and fast throughput of
the self-service store to the old over-the-counter shop. Our culture has put
increasing value on efficiency and in many ways fostered an acceptance and
even glorification of neutral impersonality in business matters—conditioning us
to tolerate or even prefer the self-service concept in more and more situations.
Indeed, the reluctance many senior citizens show in front of self-service devices
is not only grounded in their unfamiliarity with the appropriate techniques; it is
just as much grounded in the fact that their cultural values have not adjusted to
accept the absence of human contact in those situations. Jobs that look safe now
because of their emotional component may therefore be in danger if this trend
continues—such as the more routine aspects of teaching, which may become
seriously threatened by “self-service” learning based on multimedia computers,
with their combination of programming, video, sound, and databases.
The Potential of Evolving Automation—An Example

To illustrate some of the potentials of automation and the iterative nature of its
development, I would like to elaborate on these ideas in an example. And since
automation so far has progressed farthest in the factory, I think it is more
interesting to use an example from the white-collar world—where the changes
have scarcely begun.
Up until the computer entered the scene, automation in the realm of
administrative work was sparse. Punch card equipment was probably the most
advanced, and it may have been the only example of true automation.
Bookkeeping machines and mechanical desk calculators more approached the
nature of tools. Even punch card equipment was a modest achievement
compared with the extensive automation in the production of material goods.
The computer, however, is profoundly changing this state of affairs, and the
changes have proceeded further than most people realize. A large part of the
work that is strictly procedural and routine in nature has alreadybeen automated
to a greater or lesser extent, especially work associated with large files of
administrative information—and the pace is accelerating. But the automation is
often gradual and fairly unobtrusive (for everyone except those made
redundant). It is not always easy to spot for an untrained eye. For a familiar
example, let us take a short look at the development most banks have gone
through since the early 1960s.
Traditionally, banks were mainly filing and accounting organizations. Before
the computer era, they used mechanical bookkeeping machines of various kinds.
For each transaction, the customer’s account card was manually located in the
filing cabinet, placed in the appropriate machine, and the amount deposited or
withdrawn was entered manually on the keyboard. After the record was
completed, the file card had to be put back into its folder in the cabinet. This
work was carried out in central filing or bookkeeping departments, and the
inputs for their work were the receipt forms and vouchers they received from
the various branch offices and departments that had the direct customer
contacts. Typically, there were separate departments for different types of
accounts—one for savings accounts and another for checking accounts, for
instance. To get an overview of the bank’s total relationship with a particular
customer would therefore involve several persons and quite a lot of work.
This was a very labor intensive setup, and was only feasible when there were
few transactions. The number of transactions was low because society still
mainly operated in the cash mode—wages were paid in cash; goods and
services were paid in cash. Besides, it was cumbersome both to deposit and to
withdraw money—you had to go to a branch office of your bank, 1 bring with
you your bank book, and wait your turn at at least two different counters. I can
still remember the stuffy atmosphere of the savings bank of my childhood
savings account—where you first walked up to the appropriate counter,
presented your errand and your bank book, and then waited until the teller
called your name through a loudspeaker. The counters, moreover, were different
for withdrawals and deposits, although the teller window was the same.
There were also a number of instruments to conclude transactions outside the
premises of a bank, such as the check, the giro, and the credit card.
1In some countries, such as Britain, one even had to go to the particular branch that administered
one’s account. In Britain this system partly survived even up until quite recently—as late as 1990, a
Norwegian journalist working as a correspondent in London lamented the fact that he had to have
one account in a branch office in the suburbs where he lived, and another at a branch in the city
center, where he worked. The downtown branch office would not allow him to draw money on his
suburban account and vice versa.
As long as bookkeeping was manual, even they depended on a fairly low

volume to be viable, and it was not until the introduction of computers in the
1960s that the banks were ready to promote a more active use of bank accounts,
with personal checks as an important feature. It was also computers that made it
possible for the credit card companies to start their rapid expansion.
At first, the computer-based systems only replaced the manual files and the
bookkeeping machines. Still, the customer’s interface with the bank was as it
had always been, and the receipt forms, the vouchers, and the checks were still
collected and registered at a central location. Some of the manual operations
were eliminated even at this stage, however, such as the retrieval and
replacement of account cards, work with the bookkeeping machines (being
replaced by punching), and much accounting work. The punching and the
automatic processing by the computers were so much faster than the old
methods that the transaction volume could increase many times over without an
increase in the workforce. Moreover, printouts of the account balances could be
distributed to the branch offices, giving them much better information on their
customers.
Then came the next step—terminals at the counter, allowing the clerks to
register the transaction directly in the database, eliminating the need for a
central punching department. At first, the systems generally did not operate in
realtime—transactions concluded at the counter terminals did not update the
production database directly but were collected for batch processing (usually
during the night). Since then, the trend has been a development toward true on-
line systems with real-time updating of the production database.
This change did eliminate many routine jobs in central bookkeeping
departments, but the instant availability of customer information also facilitated
a significant reorganization at the customer interface—the counter.
Specialization was reduced, to the effect that most of the usual transactions
(deposit, withdrawal, currency exchange, cashing of checks, etc.) could be
completed by any one of the clerks working at the counter. A lot of paper-
pushing was eliminated in the branches, and branch offices were furthermore
authorized to give loans and credit to a larger degree than before.
The next step was to introduce automatic teller machines, allowing customers
to wrap up some of the transactions themselves. Later development has
provided EFTPOS 2 terminals in shops, allowing you to use your smart card to
pay for what you buy, thereby concluding a direct and immediate transfer of
money from your account to the shop’s. Quite a
2ElectronicFund Transfer at Point Of Sale—the card reader and auxiliary equipment that lets you
pay with your bank card in shops and elsewhere.
few banks now also allow customers (at least professional customers such as
companies) to link up to the bank’s systems, and complete certain types of
transactions from their own computers, and an increasing number are making it
possible for customers to access such services through the Internet. The net
result is that a lower and lower percentage of the transactions are concluded on
the bank’s own premises or involve any of the bank’s employees.
The development just described has been gradual, with each new step
building on the preceding one. The degree of automation has increased for each
step, both by direct automation of tasks and by eliminating the need for certain
operations altogether. If we look at the rise in transaction volume over the last
40 years, productivity has increased enormously, and the service level for most
customers (those who can handle cards, teller machines, and computerized
answering devices!) has improved dramatically. But, to a surprising degree, all
this has happened without the banks making any real changes to the basic
definition either of what a bank account is or of what a central file is. Most
banks still regard the account as their basic entity, not (as one should think) the
customer—some even to the extent of sending separate account statements in
separate envelopes for each account a person might have. What we consider a
revolution is so far not a result of a radically new concept of banking; it is just a
consequence of having made the mechanics of record keeping infinitely more
efficient through automation. It is also another example of how change that is
basically quantitative can have results that are perceived as qualitative by the
user.
If you take a closer look, moreover, the banks are still plagued by a solid
heritage of their original paper-based systems. Giros and checks, for instance,
still require a great deal of manual handling (including punching) and represent
a drain on the banks’ profits, since banks are generally not able to (or do not
dare to) charge their customers what it really costs to process paper-based
payments.
But, the story does not end here. Banks seem ripe for much larger changes
over the next thirty years than over the last thirty, and some banks may be able
to operate with only a fraction of the workforce that is common today—even
with the most sophisticated of current systems. We shall return to that a little
later, but, at this point, we must first confront the debate on the limits to
automation, especially for automation in the office.
Limits to Automation—Real or Imaginary?

The Debate on Office Automation
There is significant disagreement on the future possibilities of automation in the
office. Although no one would deny that a lot of jobs in accounting
and filing have disappeared, and that even more such jobs will go in the future,
doubts have been raised about whether less narrow jobs can be automated. Most
office jobs are simply seen as being too diverse, involving too many exceptions,
and requiring too much judgment to be defined in the exact algorithms needed
for a computer. In one sense this is true; in another it is not.
It is certainly correct that most classical “office work” does not readily lend
itself to straightforward automation. Numerous studies on this subject in the
early 1980s, where the researchers monitored the activities in various kinds of
offices, showed that office work was very complex, and even seemingly trivial
tasks required quite a lot of knowledge, judgment, and nonroutine activity (see,
for instance, Maus and Espeli 1981, Lie and Rasmussen 1983, Strassman 1985,
Long 1987, Schmidt and Bannon 1992).
The general conclusion from these studies and many later discussions is that
early hopes of automating the office in the same way as factory production were
naive and built on an superficial and overly simplistic understanding of the
nature of office work: There were simply too few repetitive activities that could
be automated, tasks were too unstructured to lend themselves to automation,
there were too many exceptions to the rules (insofar as there were any
formalized rules at all), too much of the activity was concerned with uncovering
and correcting errors, and the activities generally required the collection of
information from many different sources and the execution of considerable
judgment. Long, summing up his review, says (1987, p. 51):
Overall, conclusions based on a realistic picture of the office and its occupants
suggest that the scope for the outright “automation” (elimination) of jobs in the
near future is quite small, except with respect to semi-professionals and some
routine information-handling and coordinating roles.
The conclusion drawn by Long, Strassman, and Schmidt and Bannon, as well as
many others, is that the main scope for computers in the office is to support the
work of professionals and managers: “Stimulating an improved quality of
performance or the provision of new and/or better services,” as Long puts it
(1987 p. 46). The growing number of people occupied with R&D activities
concerning computer supported cooperative work, or CSCW (to whom Schmidt
and Bannon belong), are especially vocal in this respect.
It is undeniable that the possibilities for directly automating more complex
office jobs are limited, in the sense of having a computer system replace
humans by more or less mimicking their behavior. But that does not mean that
automation in a wider sense of the word is blocked.
Even in material production, directly mimicking human behavior is not the

way we normally automate—we do not design machines that wield or directly
mimic the use of traditional hand tools. True, there are some examples where
precisely that happens, as when a paint line is robotized by letting human spray
painters guide robot arms equipped with spray guns by hand until they “learn”
the painting movements. But in most industrial automation, automation is in
large part achieved by exploiting the intrinsic properties of machines, not by
building human-like automatons. In my view, that will also be the strategy that
will continue to revolutionize material production: The real potential in the
future lies in matching the intrinsic properties of computer-controlled
production systems with new materials and production processes that cannot be
utilized without them. New advances such as chemometry—the use of
sophisticated sensors and mathematical models for the control of production
processes—is perhaps a harbinger of things to come (Lundberg 1991).
I believe that the situation is the same even when it comes to office work and
that the studies mentioned earlier overlook the fact that even if most of the
activities of office work themselves cannot be automated, the peculiar properties
of computer systems—especially their processing power and the coordinating
effects of their databases—can nevertheless be harnessed to eliminate the need
to carry out large sets of activities altogether. The job cuts achieved through
this can be as just as dramatic as those effected through classical automation.
Consider the following example.
Task Elimination: An Example

One of the most cited examples of the elimination of a whole set of tasks is the
reorganization of the accounts payable function in Ford in North America
(Hammer 1990, Hammer and Champy 1993). Indeed, this project is probably
one of the cornerstones behind the term “business process reengineering.”
The accounts payable function consisted of typical, old-fashioned office
work—the clerks in the AP departments checked invoices against purchase
orders and receiving documents, and (if the three matched) then authorized
payments. The work sounds simple, but of course it was more complicated than
this. Quite often, the three documents did not match: The delivery might be
different from the order, and the invoice could easily differ from both. Several
kinds of information had to be collected and compared, missing papers had to
be located, inconsistencies cleared up, and so on. There was need for copious
amounts of communication, with people within the company as well as
suppliers sending the invoices.
Looking at an accounts payable office in isolation, one might well conclude

that the scope for automation is scant, and that the best solution would be to
offer the people working there various support tools to make their work more
effective (for instance, electronic mail).
Initially, that was probably also the project team’s conclusion, since they
were working to reduce the head count in its account payable departments,
which totalled more than 500 employees just in North America. The project was
part of a company-wide offensive to regain ground lost to the Japanese in the
late 1970s and early 1980s. The initial analysis proposed a project that would
use computer support to reduce the number of people to 400, which must have
seemed pretty good. However, Ford had recently bought a 25% interest in
Mazda, and Ford executives noted that the (admittedly smaller) Japanese
company handled the comparable functions with only five people.
A deeper analysis—sparked by this, and taking the total problem domain into
account—revealed that most of the work in the accounts payable departments
was a consequence of the intrinsic shortcomings of paper-based administration,
and that computer-based systems could simply eliminate the need to carry out
of most of the work in the first place, by offering a superior integration of the
information with a more far-reaching implicit coordination as a result.
Ford’s subsequent project ended up eliminating the accounts payable
departments in its traditional form altogether. Instead of using 500+ labor-years
to check and compare invoices against purchase orders and receiving
documents, and then authorizing payments, all purchase orders were registered
in a database. When a shipment arrived at the receiving dock, it was
immediately checked against that database. If matched by a purchase order, it
was accepted and registered as received (if a match is not found, the delivery
was returned). The system then automatically generated a payment transaction
and prepared the check. As the system went into operation, Ford notified its
suppliers that invoices were no longer accepted (they would go directly to the
trash bin); they should just send the goods. Ford estimated that the change
reduced the work needed to handle the control and payment functions (which is
really the reason for an accounts payable department) by 75%. In addition, there
were no longer any discrepancies between the financial and physical records to
worry about, material control became simpler, and financial information more
accurate.
This is a prime example of the possibility for eliminating work through deep
analysis of the problem domain, of the strong effects of the inherent
coordination in a unified database, and of the value of the integrity of the
information it delivers.
Mazda’s achievement was even greater and apparently also effected through
the coordinative effect of the database: 3 Mazda in effect delegated the full
responsibility for stocking the production line to their suppliers. The suppliers
were therefore allowed access to Mazda's production planning and control
system and could deliver their parts directly on the line, coordinated with the
succession of cars coming down it. All Mazda needed to do, then, was to count
the number produced of each model—knowing that, they also knew exactly
how many parts they had received from their various suppliers. The beauty of
that system was not only the total elimination of parts administration and
accounting on Mazda's hand, but also the automatic exclusion of faulty parts
from the payments (they would be eliminated by quality control during
production).
Banking: A Possible Next Step

To elaborate somewhat, let us return to banking to discuss what extended
automation and new concepts for using the strong properties of information
technology can allow. Until quite recently, what has been exploited is relatively
straightforward automation through the exploitation of the range and speed of
the database, as the accounts (the central administrative files of the bank) have
become available for access not only in the main office, but also in branch
offices and even in shops and the customer’s own office or home.
This concept can be extended further, however. It is already technically
feasible to conclude nearly all kinds of payments—be it the purchase of a new
car or of a bus ticket—as direct transactions against bank accounts. It is just not
economical for small amounts yet, due (mainly) to the cost of
telecommunications and the banks’ transaction systems. But, in not too many
years, it will become economical for almost all purposes. Further, apart from
simple payments, a much broader spectrum of transactions will be possible to
complete via office terminals, automatic teller machines, or home computers.
What will then happen to the banks?
Consider the following. As a private bank customer, I have fairly limited
requirements—I need to keep my money in a safe place, I need to pay bills and
receive money from employers and others, I need to deposit money to earn
interest, and from time to time I need to borrow. I also need to receive
information on my transactions and the current balance of my account(s). How
can I get these services most conveniently?
3Strangely enough, I have not found any account of what Mazda actually did to achieve this result,
neither in the book by Hammer and Champy nor anywhere else. The information about Mazda was
related to me by the manager of a productivity program sponsored by the Royal Norwegian
Research Council.
Not by venturing out on the streets to seek out a branch office (mind the
opening hours) or by mailing checks!
To me, the perfect solution would be a “banking system” residing on my own
PC. For safety, it could incorporate a card-and–code based identification system
(there are already PC readers for traditional bank and credit cards available, and
the PCMCIA standard for external peripheral devices could provide even more
advanced possibilities). Off-line, I could set up my transactions, and then ask
the PC to execute them. A short (and thereby cheap) automatic call to the bank’s
central computer would download my instructions, upload confirmations,
upload notifications of other transactions concluded toward my account since
my last connection, and update the balance and transaction history kept by my
local database. That way, I would always have a complete history of
transactions available without bothering the bank (after all, it is my money and
my information), and I could play around with statistics and budget information
as much as I wanted.
The upload could also contain the bank’s current offers on interest rates, it
could include electronic invoices from my creditors (which means that I could
send my own invoices through the system as well). Perhaps I could also deposit
a mortgage bond on my house in the bank, giving me a credit limit within which
I could grant myself loans at the then-current terms (contained in the latest
upload). Another advantage would be that I could have the same access to my
“local” banking services no matter where I was in the world, as long as I had
access to a public telecommunications network.
There are already services available from a number of banks incorporating
parts of this concept. However, they are not yet complete and do not yet adhere
to common standards. If the interface between such a local system and the
bank’s system was standardized, and not proprietary to the bank (or if one of the
proprietary interfaces was emulated by others and thereby established as a de
facto standard), I could not only use the system for my business with one
particular bank, but use it to obtain competing offers and conclude business with
other banks as well, not to mention insurance companies, stockbrokers, mutual
funds, and others. In my view, the growth of financial services on the Internet
will greatly contribute toward such standardization, and create a very different
type of market for financial services than the one we are used to.
A logical conclusion to such a development could be that there would no
longer need be any need for banks in the traditional sense—what I would need
would first of all be a clearing central that could carry out the money transfers
(also the many small transfers coming as a result of electronic payments in
shops, etc.) and keep an officially authorized version of my transaction
account(s). Then I would need various service
providers to offer me alternatives for depositing money, for loans, for buying
stocks or parts in mutual funds, and so on. The clearing central could even be
organized as a public institution (a new role for the central bank, when paper
money becomes almost extinct?), operating on a regional, national, or even on
an international basis.
Technologically, such a development is already perfectly feasible, and there
is no need for exotic new inventions. Commercially, development along these
lines is highly probable, as the development in basic technologies makes the
necessary equipment and communication capacity cheap enough—even if it is
too early to predict the specific directions and speed of change. As hinted
earlier, the new development in Internet banking will probably speed the
development toward the fully electronic bank further. Of course, such banks
will not take over the whole market, at least not in the foreseeable future. There
will still be sizeable customer segments that prefer more old-fashioned services.
But we will have a much greater segmentation of the market, and there will be
more competition—also internationally, since fully electronic banks should be
able to compete equally well on a global basis for many types of services. The
main constraints will be legal provisions (a number of countries would perhaps
not allow such banks to do business with their citizens) and the question of
confidence—to put your money into an account, you have to trust the bank. Not
everyone would feel attracted to even high interest rates offered by— for
instance—an Internet bank located in Grozny, Chechenya.
“Digital money” in the form of reloadable smart cards are another interesting
development in this connection. Such a card could be loaded with money from
one’s bank account, and then used for all kinds of purchases (as cash and credit
cards are used today) as well as for paying bills over the telephone network. The
card and the payment system would incorporate advanced cryptography to
ensure maximum security and should allow payment without leaving “electronic
traces” in the form of name or account information. Interestingly enough, a
number of the European central banks have supported research on such systems
through the European ESPRIT program. If we view this in the context laid out
earlier, there could be a possibility for an international network of clearing
centrals run by the central banks 4 (or by the European Central Bank), serving
cash cards as well as the “personal bank” described earlier, or even a (very
potent!) combination of the two.
4A Norwegian commission presently working to propose new banking laws has already suggested
that the clearing function in Norway should become a responsibility of the Norwegian central bank.
The basis for the proposal is the delays in money transfers consciously implemented by banks in
order to hold on to their float revenues.
Whatever the direction and speed of development, one thing is for sure: The
banks will have to change more during the next 25 years than in the previous
50, and the facilitator will be the mounting automation provided by information
technology—automation that increasingly exploits the special properties of
computer-based systems. Banks, or the corporations that replace them, may end
up as largely automated organizations with very few employees, and there may
also be considerably fewer of these companies left. If earlier communication
revolutions in historic times teach us anything, it is that improved
communications brings death to local business. When it becomes possible to
reach out beyond geographically delimited markets, it also means that all will
face a proportionally larger set of competitors. As some take advantage of the
situation and expand aggressively, local businesses who thrived mainly because
of lack of competition will find themselves in great trouble, and they will be
bought up or driven out of business in large numbers.
The larger and more perfect markets emerging from this process will foster
greater focus on price and performance, and the result will be—just as in
conventional brand-name business—that a small number of players will grow
large and destroy the others. If the Internet becomes as important for a number
of trades as many people think, it spells not only opportunity for all in those
trades, but also ruin for most of them. We can already see this development
starting for booksellers and record shops. The only alternative to agressive
growth will be to concentrate on niches where the big players cannot or do not
bother to compete—but these niches are not big enough to sustain more than a
fraction of the original players. The Internet, then, is no more of a boon to the
small, local business than the railroad, the car, and the telephone were. Like
them it is, on the contrary, an exterminator.
Circumventing the Maginot Line

If we relate these examples—one real and the other (so far) imaginary—to the
debate on the scope for automation in the office, we can see that the barrier
created by the inherently indeterminate nature of office work—by many
regarded as a Maginot line against automation—can simply be circumvented.
Quite spectacular achievements can be made without having to force the
presumed barrier at all.
Consider the Ford example: With a traditionalist approach, it would indeed
be very difficult, if not impossible, to develop a computer system that could
automate the accounts payable function—that is, to make a system that could
automatically compare purchase orders, receiving documents and invoices,
check for consistency, investigate and resolve mismatches, take corrective
action if necessary, and finally authorize
payments. Taking those tasks for granted would therefore create problems for
automation. In fact, that was in all probability what the project team at Ford first
did (Hammer 1990, Hammer and Champy 1993). If their initial efforts (in line
with the arguments of Long and the CSCW proponents) only aimed at providing
the people in the accounts payable departments with better support tools for
their jobs, they were actually quite clever to achieve a projected improvement of
20%.
When they rethought the problem in light of their discoveries at Mazda,
however, the people in the project realized that most of the tasks and routines in
accounts payable were nothing more than consequences of the way work was
traditionally defined and organized, it was not intrinsically necessary for the
reception and payment of shipments from Ford’s suppliers. When they managed
to analyze the deeper functional necessities behind the existing procedures, they
could therefore specify a system where the computer’s strong properties were
used to completely rearrange work and eliminate a whole slate of operations.
This example becomes even more interesting when we consider the subject
of groupware and computer supported cooperative work: One can easily
envision a solution to Ford’s problem along the lines of CSCW—with workflow
tools to speed electronic or scanned documents around the caseworker circuit,
with email to enhance their cooperation and their contact with the suppliers’
accounts receivable people, and with conference systems and videophones to
solve the most difficult cases. It is, however, difficult to see how even the most
exquisite system along these lines could have approached the efficiency attained
by the project finally carried out by Ford, which relied on task elimination
instead.
Of course, the Ford case is not an example of pure automation; it includes
both automatic routines and the coordinative effects of a common database for
purchasing, inventory, and accounts payable. However, that is the nature of
successful computer-based systems—they usually exploit several of the strong
aspects of information technology simultaneously. This creates problems for
orderly analyses and expositions such as the present one (I must wait until a
later chapter to discuss the coordination part), but not for the application of the
actual systems.

Shrinking the Organization
Automation, then, allows us to abolish work both through straightforward task
automation, as in a pulping plant, and by task elimination, as described in the
example from Ford. The development in banking really incorporates both. Quite
often, we will see that organizations are able
to reduce their head count even as they manage to increase their total
production—banks are good examples of this, even if they have increased the
size of their organizations over the years. Many people take this as a proof that
computers do not deliver the productivity they should, but looking at size alone
is grossly misleading. If we look at the volume of bank transactions—any kind
of transactions—the number of transactions per employee per year has
increased dramatically over the last 35 years, as computers have taken over for
bookkeeping machines, counter terminals for forms and vouchers, auto-giros
and customer terminals for manual giros, and finally card-operated teller
machines and EFTPOS-equipment for checks and cash transactions. Viewed in
this light, the banks have achieved a very impressive increase in productivity.
Moreover, they have done so while drastically reducing the size of that part
of the organization that performs the bookkeeping operations, the original main
function of the bank. Because of the increasing use of computers, bookkeeping
has actually been collapsed to a small fraction of what it was.
The reason the bank organizations have not shrunk dramatically in the same
period has been an increase in other aspects of the banks’ activities. Tasks such
as arbitrage, sales, and advisory activities have grown considerably in volume,
along with customer contact at the counter and the various functions necessary
to assess risks and give loans, manage funds, and so on. The structure of the
bank’s organizations has thus changed quite markedly, away from an
overwhelming emphasis on mass transaction processing toward a relatively high
proportion of more varied work of a professional nature.
If the development in banking conforms to the scenario outlined earlier,
however, the future reduction in workforce requirements will not be offset by
new demands for services, and we will witness a further contraction of the
banking organizations, as more and more functions are automated or eliminated.
We may also see more specialization, not only in niches, as today, but as a
general trend that almost no bank can escape, and where some banks will
specialize in highly automated routine services (high-volume, low-value
transactions), whereas others will develop into highly competent financial
service organizations concentrating on nonroutine (low-volume, high-value)
transactions.
Organizational Truncation
There are more dramatic examples of task elimination, however. Especially
within the process industries, such as paper production and oil refining,
advanced automation has led to an outright truncation of the organization:
Almost all of the manual tasks in production—which
means practically the whole factory organization—have been eliminated. This

development has been analyzed in some detail by Zuboff (1988), who bases her
book partly on the automation-based transformation of two pulp mills and one
pulp-and-paper mill.
Before automation started in these factories, each step in the process was run
by skilled workers who controlled locally their particular vat, boiler, or blender.
They had some contact with the production steps directly ahead or after their
own but were otherwise isolated from the rest of the process—except when
something really went wrong, and the whole factory had to stop. This
fragmented control of the production process naturally meant that a
considerable number of coordinating positions were required—the total
production process had to be coordinated by foremen, supervisors, and, finally,
the plant manager. Mechanical automation had allowed a fair degree of
centralization of control, but it was only when computer-based systems entered
the scene that it was possible to thoroughly automate the production processes
and eliminate next to all manual positions.
What happens, then, when a production process is fully automated and the
control of the entire factory is centralized not only to a single control room, but
(in principle, at least) to a single terminal? 5 As Zuboff (1988) shows, the
persons in the control room are suddenly, with the support of the system’s
processing power and information concentration abilities (modern process
control displays are highly graphical), in a position to directly control and run
the entire plant, without any intervening organizational apparatus. Of course
they do not run the plant in the sense that they manually control the process
(which execute under the control of computer programs), but they supervise it
and are able to improve it by tuning the program parameters as they gain
experience with the equipment and the way it operates. The depth of their
control has been dramatically increased—they almost literally run a joystick-
controlled factory.
What has really happened here is that the systems have eliminated the entire
operating organization at the factory floor, the entire operating core in
Mintzberg’s terminology, and left only the roles of the production manager and
his support team relatively intact. The organization has not only been reduced in
size; it has been truncated—one part, which was earlier the largest one, has
simply disappeared, and only machines have come instead. Of course, there is
still need for small teams of workers
5Normally, you will see a number of terminals in a control room. That is more an expression of the
present state of the technology (requiring several screens to display all vital information) and safety
precautions (the need to have more than one person available in case of emergency), than of the
technology's basic characteristics.
to maintain the plant and to tackle emergencies, but the daily control of the
production process can be left to one person (in principle), or (more likely in
practice) a small team of persons. According to Mintzberg (1979), such an
elimination of an operating core configured as a Machine Bureaucracy will
mean that the total organization is going through a metamorphosis: Its character
changes in a profound way, since staff and management, populated by many
more team-oriented professionals, will now come to dominate its structure.
It is very interesting to note how this has generally not been understood in
plants that have been automated. Almost without exception, the jobs in the
control room have been defined as transformed versions of the local control jobs
earlier performed by skilled workers, and it has been the workers who have
been trained to fill them. Our analysis here, however, indicates that the control
room jobs are not a continuation of the work on the factory floor at all; rather,
they represent the key plant management responsibility: to direct the operation
of the plant such that it achieves optimal performance, given the existing
business objectives (product and quality mix).
Before automation, the managers and their process engineers had to pursue
this goal indirectly, working through supervisors and foremen, trusting both
their judgment and the judgment of a large number of skilled workers.
Improving the quality or yield was very difficult, since so much of the process
depended on tacit knowledge, and the control over process parameters was quite
crude.
When such a factory is fully automated and the whole organization at the
factory floor is eliminated, control can be exercised directly, without human
intermediaries. In addition, control over the production process is dramatically
improved, since it can be based on accurate measurements, parameters can be
adjusted in minute increments, and systematic experiments can be made in order
to improve both yield and quality. The direct control of the total process and the
tuning activities are definitely not a customary part of rank-and-file factory
work; they belong squarely to the traditional domain of management and
engineering.
Quite naturally, this mismatch has led to conflicts between control room
operators and production managers, their subordinates, and staff. Zuboff
describes this in some detail (1988) and explains the reasons quite accurately:
Clever control room operators will, after some time, develop a deep
understanding of the total process and will increasingly be able to tune it for
greater economy, increased production and/or improved quality. Managers, who
do not have direct access to the control systems and the information they
provide about the processes, will fall behind and will not be able to either direct
or control the operators’ work in a meaningful way. Frustrated over losing their
positions as those who can
best comprehend the total process, they will often try to reassert their authority
by giving the operators directions anyway—directions that are likely to be
inferior in most cases, since they are based on inadequate information and a lack
of experience with the system. The operators, in their turn, will feel this both as
an encroachment on their newly found responsibility and an affront to their
professional competence. Both parties will suffer, together with the plant’s
economic performance.
Ideally, then, control room responsibilities should have been left with plant
management and the engineering staff, and the systems designed accordingly. It
is they who should have been trained to use the systems for controlling and
optimizing production. If it is desirable, for moral or political considerations, to
appoint former workers for such jobs, one should be very conscious of the
implications and provide educational programs, discretionary powers, and
benefits that match the real responsibilities of the job—because, in the highly
automated factory, it is the people who master the systems and understand the
information they provide who decide the profitability of day-to-day operations.
Hyperautomation
Already, information technology has helped us develop the extent and
sophistication of automation far beyond what was possible by mechanical
means alone. Looking into the future, the scope for progress is still vast, and the
limits are difficult to define. Advances have already been dramatic enough to
warrant a new term to distinguish this new breed of automation from
mechanical automation as we have seen it develop over that last century: it
could be called hyperautomation. Hyperautomation is the computer-dependent
variant of automation, and it can be mapped under that entry in the taxonomy of
coordinating mechanisms shown in Figure 7-1 on p. 165 (as I have done in
Figure 13-1 on p. Error! Reference source not found.315). In principle,
hyperautomation is not different from automation, but, just as for computer-
based information storage, the sheer power of the new tools is so great that they
are nevertheless qualitatively different.
We may see great organizational changes in connection with
hyperautomation, as in the examples described earlier. By shedding almost all
the workforce in its operating core, a company can be transformed from a
Machine Bureaucracy to something much more like an Administrative
Adhocracy (Mintzberg 1979). There is also no doubt that the use of information
technology will make such transitions possible for a much larger number of
organizations than mechanical automation ever could.
Hyperautomation also makes it possible to integrate a much greater span of
organizational activities into one coordinated process, not least because it allows
the automation or elimination of significant
administrative processes. We have thus already seen process-oriented

automation expand along value chains (Porter 1985), even outside the
boundaries of the principal organization.
Prominent examples of this can be found in the automobile industry, which
has for a long time been at the forefront of automation. When building their new
factory in Sunderland in northern England, for instance, Nissan invited
important suppliers to establish their own factories at the perimeter of their main
plant site and tie directly into Nissan’s production control system. The objective
was to have the suppliers deliver their parts directly on the assembly line, to
save storage space and handling costs.
As soon as the basic body of a new car is put on the painting line in the
Nissan main plant, a transponder is attached to it, containing the complete
specifications for that particular car. 6 This is particularly important, since the
broad range of colors and options offered today’s customer virtually ensures
that no two adjacent cars coming down the line will be identical (the company
claims to offer customers 20 000 varieties of their vehicles). When the body
leaves the paint line, the transponder is read by the the central production
control system, which broadcasts the information to the subassembly stations
and component supply points as well as to the suppliers that are tied into the
direct delivery system. The manufacturer of seats, for example, receives the
necessary specifications three hours before the seats are to be fitted (Christopher
1992). Only then do they start their own production, assembling the front and
rear seats to match the car model, colors, and other details determinded by the
model and the customer’s choices. Every 15 to 20 minutes, a transport shuttle
leaves their factory, taking the finished seats directly to the appropriate supply
point at the assembly line, where they arrive just before the car they belong to.
What we see here is an extremely tight coupling of a number of independent
organizations, a coupling that is even tighter than you will normally find
between departments within a single organization. Nissan’s own plant, by the
way, operates according to the same principles—its press line for body panels,
for instance, is carefully synchronized with assembly, to the point where the
total amount of doors, trunks, and trunk lids in process amounts to less than
what is required for one hour’s production.
Even though these supplier organizations all have their own independent
owners, administrations, and economies, for the purpose of producing Nissan
automobiles, they function as one amalgamated organization with
6This information is drawn from the company information package distributed by Nissan.
a common coordination infrastructure. We shall return to this kind of

organizational setup in Chapter 16, since it constitutes a new organizational
configuration.
Hyperautomation is a tool that offers dramatic new opportunities for the
design of organizations, and one that may also greatly affect the development of
society. The possibility of organizational truncation and the establishment of
strongly coupled organizations are genuinely new extensions of the
constructible space. However, the new tools and even the new organizational
configurations work fairly well with established organizational practices and
structures. For all the work that is not eliminated by the new systems, it is
eminently possible to use common structures and coordinating mechanisms.
Though I am not certain, I would guess that the remaining operating cores of
both Nissan/Sunderland and the suppliers are predominantly Machine
Bureaucracies, and that staff and management still operate much as they used
to. It would certainly be possible. This fact is perhaps also one of the main
reasons why hyperautomation has developed so fast, and organizational
truncation and the development of strongly coupled organizations have kept
pace with this development. It is only natural that more unconventional
approaches (if they are possible) will take a longer time to develop and deploy.
Consequences for Society

As we have just seen, the consequences of extensive automation are dramatic
for the organizations involved. The more organizations exploit the potential
computers offer in this direction, the more the consequences will also be felt on
the societal level. Increasing automation will irrevocably change the labor
market, and the great advances in productivity will provide a steadily increasing
material prosperity, if the accompanying environmental problems can be solved.
The developments in the labor market have actually been underway for some
time, apparently as a continuation of a long trend starting with industrialization
and the mechanization of agriculture. First, industry overtook agriculture as the
major employer, but, as industrial productivity increased through automation
and increasing use of energy, it was overtaken by the service sector (including
public services and administration). The further contraction of the industrial
sector can easily be interpreted as a continuation of this trend, trailing the
agricultural sector by a number of decades.
However, even without venturing into a discussion of the development of a
service economy, information economy, or the postindustrial society, it can be
stated that we are experiencing a break with this development; we face a new
situation with unclear consequences. The significant difference
between former developments and the present is that, up until now, the routine
work eliminated in one sector has always been supplanted by routine work in
another. As the available positions for farmhands dwindled, up went the number
of positions for factory work; and as their number in turn declined, the great
white-collar bureaucracies expanded to offer a new set of jobs.
What is happening now is that the remaining routine jobs in both industry
and in the service sector continue to be decimated, but no new ones seem to be
appearing: Almost all new jobs are less routine than those that disappear. The
required level of education rises, and it becomes more and more important to be
able to think abstractly and to understand and manipulate symbols instead of
physical objects. This tends to be true within most occupations, even traditional
ones.
A simple example of this is a subtle change in the situation of secretaries who
do a lot of typing. In the days of the typewriter, their core professional skill was
of the action-centered type—it was the physical skill of hitting the correct keys
very fast. The typewriter was a simple and very concrete tool, and its operation
and few controls were well understood by the secretaries. They were the
undisputed office masters of typing and editing. Today, where most
professionals and managers have their own PCs with the same word-processing
software as the secretaries, the situation is significantly changed. True, the
secretaries are usually still the fastest key-hitters, but they are generally no
longer the masters of their tool. In most organizations and departments, there
will be a number of professionals who are more proficient than the secretaries in
using the advanced functions of word-processing software, and the secretaries
will often have to turn to them or to support personnel for help. The case is the
same for errors and system breakdowns: The secretaries do not have the general
knowledge about their computers to escape from even relatively simple error
situations, and they again need help from someone else. Many secretaries
experience this as humiliating, and as something that undermines their former
position as specialists.
We often see that even extensive training does not change this situation
significantly. The task of typing and editing has become so much more abstract,
and the writing tool itself so exceedingly complex and symbol oriented, that it is
more easily mastered by the professionals, who generally have extensive
training in handling symbols and abstract problems. On the average,
professionals may also have greater natural abilities in that direction to begin
with.
Are we all able to live up to these new requirements? Or will there be a
sizable number of people in our societies who will never find a job they can
master? Will we have to stimulate the creation of more simple service jobs,
which can offer a decent and respectable living to those who do not want
or do not master intellectual work? At the moment, these questions do not have
clear answers; they only echo growing political concerns (at least in some
countries) about the “two-thirds society,” where the fortunate two-thirds of the
population is employed and grows more and more prosperous, whereas the
unfortunate third is unemployed and only becomes poorer. Although it is not the
theme for this book, this question represents a problem that will affect every
aspect of society.
Another aspect of the developments outlined here is that it is not only jobs
that are becoming more complex and abstract—many everyday doings follow
the same trajectory. Paying bills, for instance. Not too long ago, it was possible
to live by cash alone. Today, that is becoming increasingly difficult, and you
suddenly have to write giros or checks and understand bureaucratic procedures.
No big deal for most of us—but to the 5-10% in most western societies who are
functionally illiterate, it is. As technology, abstraction, symbol manipulation and
written directions seep into more and more aspects of our lives, those who have
trouble reading streets signs are having greater and greater problems, and no one
seems to take their predicament seriously yet—neither in business nor in
government.
303
13 Coordination by Default
“Harmony would lose its attractiveness if it did not have a

background of discord.”
Tehy Hsieh, Chinese Epigrams Inside Out and Proverbs, 1948
Implicit coordination achieved through the use of common archives or files was
the first new coordinating mechanism made possible through the use of
technology, and it has played a crucial role in the development of the modern
office organization. Its elegance and efficiency stem from the fact that it allows
coordination to be achieved not by actively directing people, but simply by
recording information and making it available. However, as long as it was tied
to paper, its potential was severely restricted—the information being accessible
in only one location, and normally having only one index. Therefore, it did not
come fully into its own until the advent of information technology, or
specifically, the structured database, which lifts these restrictions through
automatic indexing, automatic search and retrieval, and electronic
communication.
The Structured Database

That Significant Record
As we have noted earlier, writing was most probably created to keep records for
business and public administration (Goody 1986). The first material memory
technology was thus used for storing administrative information—itemized,
often quantitative information such as sums of money, numbers of cattle,
amounts and kinds of goods, names of people, sizes and locations of landed
property. Only later did it become a medium for discourse, for art, and for
accumulation of knowledge and reference material. However, because the
written discourse and the accumulation of knowledge and reference material
were decisive for the development of philosophy, religion, science, and politics,
and thus were more visible (and
exiting), those aspects of writing easily attracted most of the attention in

historical analyses.
Nevertheless, the Industrial Revolution and the evolvement of modern
society depended just as much on the meticulous record keeping of merchants,
master craftsmen, industrialists, engineers, civil servants and, not least, their
clerks. Their tidy accounts, production plans, inventory lists, file cards,
protocols with customers and suppliers, details of business transactions, land
registers, and tax records became the lifeblood of an increasingly complex
society.
The great importance of record keeping is evidenced by the fact that all new
technologies for information storage seem to make their debuts in the realm of
business and public record keeping— punch card equipment and computers
included. When they first ventured beyond research and was adapted for
administrative purposes, it was indeed for record keeping and tabulation: The
only two private customers on Univac's order books in 1948 were Prudential
Insurance Company and the market research company A. C. Nielsen.
In my view, the vast improvement and extension computers bring to implicit
coordination represents one of their most revolutionary aspects—and one we
find behind most of the familiar success stories that circulate in the business.
Paradoxically, it is also among the least talked about. The reason is probably
that it does not reside in highly visible equipment such as personal computers
and scientific workstations, or in their increasingly advanced software and add-
ons. It does not jump at you like a fancy multimedia presentation—you cannot
walk into a computer show or an office and see implicit coordination, unless
you take the (often considerable) time needed to study and understand the
applications and databases accomplishing it.
Tools for implicit coordination are nothing new. Nor is it new, either, that
they are undervalued. Today, however, computers have brought new
dimensions to it—even if the database, logically speaking, is in many ways just
an extension of the paper-based file. The increases it offers in speed, availability
and ability to handle complex information are nevertheless so great that it
becomes qualitatively different. The difference is further increased by the fact
that the information in a database is available for automatic operations.
The central aspects of this new functionality are the reach, capacity, and
speed offered by the implicit coordination achieved through the use of
databases.
Reach
The coordinative reach of a database is a function of the available
communication arrangements. If communication lines with sufficient capacity
13 Coordination by Default 305
are available, the geographical reach can cover the whole earth (and more, if
that should be required!). Thus, with a true, on-line banking system, for
example, a transaction registered against your account in any one branch office
is immediately reflected in an updated total for that account in the central
database—and so instantly available for all the other branch offices as well. You
can therefore expect a coordinated response from your bank—no matter which
branch office you walk into, the amount of money they would be ready to give
you should be the same.
Capacity
Another important factor for of the coordinative power of a database is the
number of people that can be simultaneously coordinated. As with geographical
reach, there are no definite theoretical limits here—the achievable capacity is
determined by the level of the available technology and is rising rapidly all the
time. To my knowledge, the largest capacity displayed by single databases in
1999 were the airline reservation systems. The three largest are SABRE,
Galileo, and Amadeus. At the time of writing, Amadeus is probably the largest,
with 180 000 1 terminals generating (probably) more than 6000 transactions per
second at peak load. When the first such system, SABRE, was introduced in
1964, it taxed the capacity of the fastest machines then available with its 1200
teletype terminals (Hopper 1990).
If the development continues at the same pace—and there are no reasons why
it should not—a single physical database should be able to accommodate at
least 3 million on-line transaction-processing terminals in the year 2020. My
guess is that we will reach that level even earlier—maybe as early as the first
decade of the twenty-first century—due to advances in parallel processing and
new storage media. For less transaction-intensive applications, the number of
terminals could be considerably larger. It is not necessary to test the limits of the
possible performance ranges to extract great value, however.
Speed
In theory, the reach and capacity of coordination described earlier are not
dependent upon computers. Information can travel the world on paper as well as
on wires and airwaves, and a paper file can thus be accessible for anyone,
almost without regard for distance. There are many library services in the world
demonstrating this principle daily, and the Japanese kanban system, the tool
behind the original development
1 Personal communication from a representative for Amadeus in Norway.

of just-in-time production, was originally based on cardboard cards. Indeed,

many smaller Japanese companies still rely on such cards in their daily
production.
But reach and capacity is not everything—it must be coupled with speed. If
we look again at the examples described earlier, we will see that none of them
would be feasible without the instant transfer of electronic communication. And
even that is not enough—it must also be combined with the instant registration,
retrieval, and transmittal offered by computer systems. It is precisely this
combination of reach, capacity, and speed that makes the database qualitatively
different from the paper-based file (and from kanban cards, for that matter).
As an example, we can go back to SABRE, the first airline reservation
system, developed by American Airlines and IBM from 1954 to 1964. Before
SABRE began operation, all flight bookings and changes were received through
telephones (note that telephones provide instant or almost instant transfer) and
recorded manually on blackboards and index cards (Hopper 1990). When the
development of SABRE started, however, the booking department of AA had
begun to look really strained; by the time SABRE was finished in 1963/64, it
was probably coming apart at the seams.
The reason is not difficult to see—the number of persons needed to answer
all the telephones was increasing dramatically and changes to the cards and
blackboards were cumbersome to effect—and, as the number of callers and
clerks steadily rose, the update problems increased even faster. In addition,
there was a significant time lag between the actual confirmation of a seat and
the moment this was known by the other clerks, and that time lag could easily
lead to trouble.
Today’s traffic volume would probably not even be theoretically possible to
handle the old way—already in 1990, SABRE’s database contained 45 million
fares from 650 airlines, there were up to 40 million changes every month, and
more than 500 000 passenger name records were created every day (Hopper
1990). At the time of writing, SABRE handled booking for more than 400
airlines, 35 000 hotels and 50 car rental companies. It had over 30 000 agency
locations, more than 130 000 terminals attached, and in 1996 it processed over
5200 transactions per second and peak load (according to the company’s annual
report for 1996). Without the automatic and extremely fast reads and writes of
the central computers, this would simply be impossible.
Multiple Databases and System-to-System Communication

Even if it is not always feasible or even desirable to coordinate a set of activities
through a single database, I believe that the single-database
solution will become increasingly important for intra-organizational purposes as

both software and hardware improve and communication lines become cheaper.
It simply provides a superior solution with regard to speed, integrity, and
administrative overhead. However, there will be a considerable period where
solutions with multiple databases will dominate, and, in interfirm linkages, they
will probably dominate in the foreseeable future.
Linking can take several forms. A database may be split physically among
several distributed machines but still be logically organized as a single database.
A database may also exist in several copies, with mutual updating at preset
intervals. In principle, this is also simply a single database split physically. The
reason for such a setup is almost invariably that telecommunication costs makes
it a cheaper solution than a centralized database.
Much more common, however, is the situation where the linked databases are
quite different from each other—when they use different software, belong to
different applications, and run on different kinds of machines. We can then use
electronic messaging to synchronize key information between them—let the
databases exchange information about their states, allowing automatic
coordination to take place—let orders update production schedules, invoices be
matched against purchases, and payments against records in accounts
receivable. The messages can either go automatically as a result of processes
internal to the application programs or be triggered directly by the users.
Messaging is going to be particularly important as long as systems remain
fragmented, and even intraorganizational databases are diverse and
incompatible. It will allow automation of many existing routines, increase
speed, and save labor hours without requiring any fundamental logical changes
in the administrative structure. In fact, standardized messaging (such as EDI)
tends to conserve existing practices, because of the nature of the standardization
process itself. The creation of international standards for messaging involves a
large number of countries, standards organizations, and trade organizations. To
be accepted, any standard will have to build on widely used documents and
forms that are part of the traditional way companies of the world organize and
do business. The whole EDIFACT standard is a witness to this—the catalog
over standardized document formats reads like an old-fashioned textbook on
accounting and business administration. Examples (taken from the catalog in
Thorud 1991) are as follows:
IFTMAN Arrival Notice Message

IFTMBC Booking Confirmation Message
CREADV Credit Advice Message
DEBADV Debit Advice Message
DESADV Dispatch Advice Message

DOCAPP Documentary Credit Application Message
PAYEXT Extended Payment Order message
IFTMBF Firm Booking Message
INVOIC Invoice Message
PAYORD Payment Order Message
From the outside, an organization with extensive use of EDI may thus look
thoroughly reformed in the way it operates, due to improved coordination,
increased speed, and reduction of errors. Below the surface, however, most of
the old functions and procedures might still be intact, even to the extent that the
old, isolated applications still run—now only augmented by EDI-compatible
front-ends. In other words, existing procedures will just have been
“mechanized.” But the quantitative change (increase in speed) is so great that
the consequences are often perceived as qualitative.
Bearing in mind the accounts payable function of Ford described in Chapter
12, you will see from the preceding list (where you find both a DESADV, an
INVOIC, and a PAYORD message) that an uninventive solution based on the
use of EDI is eminently feasible, and it could easily have been compounded by
the use of email and other groupware applications. The result would have been
an impressive system, yielding considerably less improvement than the much
more radical solution eventually adopted by Ford.
Networks of linked systems may be very large. The largest existing one is
probably the SWIFT network used by banks for international money transfers.
Most large banks throughout the world are connected to this network, and
billions of dollars are moved over it every hour, constantly updating account
databases in banks worldwide, with an unknown total of terminals attached. A
transaction entered on a terminal in Oslo may withdraw a certain amount of
kroner from an account in a Norwegian bank database, and update an account,
for instance, in Japan with the corresponding amount of yen. 2 In turn, this would
cause a Japanese bank clerk to allow the owner of the Japanese account to
withdraw money without any other notice than the implicit coordination
provided by the linked databases. As we have just seen, even tighter coupling
can be found in the automobile industry.
Speed need not be much lower for systems based on linked databases than for
those based on a single database, but it often is—there may be batch processes
involved, or delays may be deliberately introduced for
2Itmay take some time—a day or two—because there may be old equipment and batch-oriented
systems involved, or the banks may want to sit on the money for a certain amount of time to earn
some interest, but, technically, the transfer could happen immediately.
other reasons, as is often the case with money transfers—the banks want their
float just as in the old days, no matter how fast the technology may allow them
to operate. In my view, however, only unified databases can deliver the
maximum advantage from database technology. Information should be
registered when and where it is created or captured, and stored in only one place
to ensure full information integrity. Single databases are also much better and
more efficient than messaging for coordinating a total value chain. Distributed
databases, where the same information may be stored in several physical
locations, will incur a great deal of processing and communication overhead to
maintain integrity. As communication costs continue to fall, the central database
will therefore grow in popularity.
This discussion has been focused on administrative applications. There are
also, of course, important uses of messaging and common database access in
areas such as sensor information (in process control and other manufacturing
systems, in air traffic control and other types of monitoring systems) and in the
direct control of physical devices (as in manufacturing and military systems).
But the principles are the same, and the benefits derive from the same basic
mechanisms: integrity of information, implicit coordination, and fast responses.

Computers do extend the scope of implicit coordination. They offer real-time
coordination almost regardless of volume and geographical distance, even
across processes. But how does this extend the space of constructible
organizations? In my view, there are opportunities on three main levels: inside
the single organization, on the interorganizational level, and on a level where it
can be debated if we are really talking about organizations at all.
The Single Organization

Banks, Automobiles, and Airplanes
As a coordinative tool inside the single organization, implicit coordination has
already proved some of its mettle. We have talked a lot about banking, and there
is little doubt that the coordinative power of the database is the main force
behind the flexibility of modern banking. The instant or near-instant availability
of account information has made it possible to eliminate a lot of back office
work and control procedures, to broaden jobs, and to extend services reliably
both to self-service devices (e.g.,
automatic teller machines) and to external agents (e.g., shops and filling
stations).
However, even if the database is vital for banking, and its effects powerful in
the sense that it allows any clerk in any branch office to meet a customer’s
request in the same way, the coordinative aspect is nevertheless fairly narrow—
the clerks are coordinated in their assessment of the customer’s financial status,
but, aside from that, the actions of one clerk will have little or no implications
for another. That even this fairly simple and narrow application of implicit
coordination has led to revolutionary changes in banking is a proof of its power,
and with the addition of automation—especially in combination with
“intelligent” self-service solutions such as those postulated in the Chapter 11—
it will eventually effect a total transformation both of single organizations and a
whole industry.
The potential is even greater than this, however. To find an example, we need
go no further than to an example from the previous chapter: the revolution in
Ford’s accounts payable departments. It was the use of a fairly simple database
that made the whole thing possible, and there is even an element of cross-
process coordination to be found—the database not only effected the
procurement of parts and payment of suppliers; it also saved a lot of work in
financial reporting. The reports became both more up to date and more reliable,
since information on parts and payments was available at all times and were
always consistent.
Even more illustrative is the situation at Mazda, where the same database
served as a coordination point for orders, production and supplies from
subcontractors. The data integrity and implicit coordination offered by the
single database provided full synchronization of the entire manufacturing
process (“procurement to shipment,” as Hammer and Champy [1993] prefer to
call it), the sales process (“prospect to order”) and the order fulfillment process
(“order to payment”). The same is, in all probability, the case for the Nissan
factory in Sunderland, also referred to in the previous chapter, although details
are lacking in the available description. Indeed, it is this potential for cross-
process coordination that is the main motive for the general movement toward
registering data only once—at the point of origin—and storing it in one place
only to ensure integrity.
A different example of the strong coordination that unified databases can
provide can be found in the aerospace industry. The Boeing 777 was the first
airplane whose full structural design was done in an integrated CAD/CAE 3
system (Stix 1991, Moeller 1994). A modern jetliner is a very
3CAD: computer aided design. CAE: computer-aided engineering.

complex piece of machinery, and with traditional, paper-based design, a major

part of the job is to manage thousands of drawings; correct them when there are
changes; to ascertain that adjoining parts actually fit together, and that no two
parts (including piping and cables) occupy the same space. This job is so
complex that it is simply impossible to complete on paper—to really find out if
all the parts fit, and whether cables and piping collide, physical models and full
scale mock-ups have traditionally been necessary to sort things out. It was even
usual that last-minute changes had to be made during the actual manufacturing
of the first airplanes, due to problems that had not been discovered during the
design phase.
With the 777, all the design work was done on workstations equipped with a
three-dimensional (3D) design program, which made it possible to display each
part as a 3D picture, rotate it to view it from different angles, test the effect of
movements, and so on. Because of the integrated database connecting all the
workstations, neighboring parts could then be joined together on-screen—any
engineer could call up the parts adjoining the one he or she was working on to
check if they fitted together. The screen even provided the telephone number of
the person working on that particular part, in case there should be need for
consultation.
The parts could also be assembled on-screen to modules and to a complete
model of the entire aircraft, including such vitals as cabling and piping. The fit
between parts and modules could thus be tested without the need to build
models and mock-ups, and the software could detect if any two parts—for
example, two cables—occupied the same point in space. The design program
and the database, then, took care of the coordination and ensured that the work
done by any of the thousands of engineers matched with what the others did—
without the need for human liaisons.
Besides the digital design itself, Boeing also took advantage of the
coordinative powers of the system to integrate about 15 different design and
engineering steps into a single overlapping process (Stix 1991). Manufacturing
engineers were able to write tooling specifications as soon as design on a part
had started, and could provide feedback on manufacturability early in the
process. Some of the CAD data could also be fed into CAM 4 systems and used
directly to manufacture parts. Likewise, the same data could drive automatic
testing equipment, examining parts for mechanical accuracy (mechanical
tolerances are very narrow in this business). According to a November 1991
article in the New York Times, Philip M. Condit, then Boeing vice president in
charge
4CAM: computer aided manufacturing.

of the design project, said that the system allowed the engineers in this large
project to work together just like the team of less than 100 engineers who
designed the B-29 bomber during World War II.
The actual size of the 777 project team, however, is not clear—the number of
workstations and people actually working on the design, has, surprisingly, been
impossible to ascertain. The various articles all report different numbers, and
personal communication with sources at Boeing has not helped very much to
clarify things, since Boeing regards the CAD system and the way it is used as a
competitive advantage important enough to be shielded from detailed reporting.
It seems reasonably certain, however, that the number of people involved in the
design was higher than 5000, and that at least 2-3000 of them—at more than 20
sites in USA and Japan—were equipped with workstations.
The 777 project did not save Boeing time in the period up to the first roll-out
of the new plane, due to the time it took to train the large number of engineers—
not only to use the new system, but to “think” in 3D and to work in cross-
departmental teams. However, with that job done, and the entire structure of the
plane in digital form, Boeing spokesmen were confident of spectacular savings
in the development of new versions. Moeller (1994) quotes Larry Olson,
director of computing systems at Boeing, as saying that custom versions could
be built in eight months, compared to the previous 52—an 85% reduction in
lead time! It seems reasonable to expect that the system will save Boeing
considerable time when they embark on the design of their next new aircraft,
and that it will be extended to cover larger parts of the total aircraft design.
Bigger, Better, and Brisker

These examples illustrate the main strength of the database as an organizational
tool: the ability to provide coordination as a spin-off, as an implicit effect, of the
data storage itself. Coordination that earlier required significant, even massive,
efforts can now be effected without any human mediation at all, with much
greater speed, and with much better precision. Computer-based implicit
coordination should make it possible to build and maintain much larger
organizations than before, to make large organizations much more responsive,
and to improve the quality of their output. The condition is of course that
common information lies at the base of their main activities, as they indeed do
in banks and in design projects. For manufacturing operations such as the
Mazda and Nissan factories, a common information base must be augmented by
advanced automation to achieve the maximum advantage. Indeed, the same can
be said about banking, where the combination of the database, automation, and
self-service will soon make it possible to run vast transaction-processing
operations, covering great geographical areas, with surprisingly slim

organizations.
The increased responsiveness does not only come from the speed with which
one organization member can retrieve information—for example, to answer a
simple customer request. It is equally important that changes in the common
information base are instantly incorporated into the basis for everyone else’s
work. Thus, it takes considerably less time and effort for the organization to
come up with a consistent response to a request that involves more than one
person or group. This will be true for relatively simple cases as well as for really
complex ones, such as an airline’s request for a custom version of the 777. The
consistency of the information and the instant updates will allow a large
organization to respond in ways that were earlier only possible for organizations
small enough to have almost every relevant person working in the same
building—taking advantage of the richness of face-to-face communication. The
integrity of the common information will also contribute significantly to product
and service quality by increasing the internal consistency and accuracy of the
output.
Decentralization
The implicit coordination achieved through the use of databases eliminates a lot
of administrative tasks that used to be necessary to coordinate work. Those tasks
used to be the main responsibility of middle management. However, they were
also combined with decision making—and when the coordinative tasks are
eliminated and middle management is correspondingly decimated, the decision
making is not necessarily eliminated with them. Someone, then, still has to take
care of it, and it tends to go where the information is used—which most often
means a migration toward those parts of the organization where customers’
requests are met. In most instances, this will mean a decentralization of
authority (vertical decentralization in Mintzberg’s terminology [Mintzberg
1979]). To illustrate, let us detail the banking example a little bit.
An important effect of the introduction of modern banking systems has been
the transformation of work at the counter level. Before computers, or, rather,
before they got terminals, the clerks working at the counter were little more than
paper pushers. Their most important decisions were whether to accept a check
or an identity card. The introduction of terminals (especially on-line terminals)
changed that. First of all, it eliminated a lot of registration (that is, most of the
paper work), even if the transaction still had to be registered. However, with an
on-line system, the clerk could now immediately check the customer's balance,
transaction
histories could be retrieved, and accounts could be opened and closed on the
spot, to mention a few of the new possibilities. Since this also meant that
important parts of a customer’s total relationship with the bank were available at
the counter, and additional information on creditworthiness became available in
commercial databases, many banks authorized some of their clerks to grant
small loans and credits to customers.
This was indeed a revolution, and it was the instant availability of
information at the customer interface that invited the delegation. Earlier,
information had to be collected from many sources within the bank itself, where
it resided in paper-based files, and the retrieval process had to follow
established archival rules and mail routines. All this took time and effort—
information travels slowly when it sits on paper that has to pass through several
hands. To collect this information was therefore back-office work, and
managers coordinated it and reviewed the results before making the final
decisions.
Under the new setup, the counters were usually divided into two zones—one
where customers concluded their normal transactions (such as withdrawal,
deposits, payments, and so on) in terminal-equipped teller windows, and another
where they could conclude more “elevated” business (such as opening an
account or applying for a small credit on a salary account). This development
was usually also followed by an increase in the sizes of loans and credits that
branch managers could authorize.
Used for decentralization in the manner just described, computers will
undoubtedly lead to a reduction in the number of organizational levels and an
increase in the authority and latitude for judgment in the bottom layers of the
organization. This leads many people to argue that IT is first and foremost a
technology for decentralization, and that empowerment of employees and the
transfer of responsibilities down the ladder of authority are prerequisites to
success when implementing computer-based systems. I disagree with this view,
which I consider overly optimistic, just like the claims that computers will foster
a proliferation of networked teams. I shall return to this subject in the next
chapter, where I will argue that computer-based systems can also be used also
as a tool for centralization. In my view, computers will not force the
abolishment of hierarchy, but, on the contrary, provide a platform that extends
our options—the space of constructible organizations—in both directions.
Implicit Coordination as an Expression of Mutual Adjustment

In Chapter 7, I classified paper-based implicit coordination as an expression of
mutual adjustment. It should then follow that computer-based

Tacit Explicit
Skills Skills

Computer Dependence
Implicit System- Programmed Hyper- System-

Coordination Supported Routines automation Supported
(by database) Supervision Skills
Figure 13-1: Taxonomy of coordinating mechanisms extended by the

use of information technology (preliminary)
implicit coordination is such an expression as well (Figure 13-1). Just like its
paper-based counterpart, it works by effecting an indirect, mutual adjustment
between all the people who use the database for their work. In most cases, it
cannot provide total coordination—designers of neighboring parts on the 777
had to talk to each other from time to time, as did people working with different
aspects of the design process. But it is nevertheless sufficiently powerful to
allow a radical reorganization of the entire process, and it extends some of the
functionality of mutual adjustment to a potentially vast number of people.
The simple elegance of the principle of computer-based implicit
coordination—where extremely detailed, complex, and time-critical
coordination can be achieved without any direct coordination effort, unfettered
by geographical distance—represents the second great power of information
technology, on par with hyperautomation.
Coupled Organizations
In Chapter 12, we discussed the case of separate organizations tightly bound
together and coordinated through common computer systems.
The focus then was on automation, but there is little doubt that the foundation
for hyperautomation along value chains will always be one or more common
databases, sometimes synchronized by messaging. The same will be the case
with other arrangements for tight organizational cooperation. The implicit
coordination provided by databases is therefore a pivotal factor when coupling
independent organizations. The reason is of course the same as for single
organizations: it is a very efficient coordinating mechanism, it requires little or
no manual work, it works independently of geographical distance, and the upper
limit on the number of persons or organizational entities that can be coordinated
is already very large and rapidly rising.
As long as the coordination achieved through computer-based systems is
sufficient, then, there will be few technical constraints on size. The main
limiting factor is no longer the technology, but simply the will to cooperate and
to undertake the painstaking analysis, standardization and design that is required
to make such structures work. The task is already formidable for systems in
single organizations, and the work required to standardize the format of
information items across organizations can frighten even the most hardened
project manager. Even reaching agreement on simple things, such as the number
of digits in an order number, can be difficult enough to delay projects for
considerable periods of time. This also serves to underscore the fact that such
tight coupling is not easily established, and the easy way in which terms such as
“network organizations” and “virtual organizations” are used today belies the
effort it takes to establish them and the losses involved in opting out once the
cooperative structure is working.
However, I definitely think we will see more such constructions in the future;
I have already hinted that I believe they constitute a new kind of organizational
configuration—but that part of the discussion must wait until Chapter 16.
On the Fringes of Organization

One of the examples I used earlier in this chapter to throw light on the way
databases provide implicit coordination was airline reservation systems. The
operation of all the big reservation systems is quite astonishing, if you really
think about it. While you are on the telephone in the late afternoon with your
travel agent in Cincinnati, mulling over if you should book that last available
seat on the first flight from London to Milan on Christmas Day, I might place an
early morning call to my travel agency in Oslo and snatch the ticket from under
your nose—and you will know it and suffer the consequences in the same
instant (give or take a few seconds, depending the current response time of the
system and the alertness of
your agent). You would then end up as the victim of an implicit coordination
spanning half the world, taking place inside a database physically located on
another continent.
What is really happening here, when tens of thousands of travel agents daily
book flights for their customers on Amadeus? The coordination implicit in the
common database keeps them constantly informed about everyone else’s
bookings; new flights; changes in pricing, departure, arrival times, and so on.
Clearly, there is a strong element of organization present—if they did not have
their computers and database, they would need an enormous hierarchy of
coordinating managers and professionals to carry out the same work—except
for the fact that the whole feat would be totally and utterly impossible to bring
about without the database in the first place.
Of course, this mass of travel agents cannot be said to make up an
organization in our normal understanding of the word. But no one can deny that
their common database connection ensures that their behavior appears organized
in certain key aspects—their actions are coordinated in the sense that any
booking by any agent has the potential to modify the behavior of any other
agent connected to the database, and that such modifications routinely occur. It
is not an organization, but it is certainly organized, representing what can be
considered a new kind of structure. We shall return to this discussion in Chapter
16.
319
14 Comprehension and Control
“Knowledge is the true organ of sight, not the eyes.”

Panchatantra, c. 5th century AD
Comprehending the Complex

Getting to Know
In Chapter 9, we briefly discussed how the digital computer’s processing power
has made it possible to handle much greater complexity than before. It is of
course not only processing that is involved—as with most application areas,
several of the strong points of computers are involved at the same time. In this
instance both registration and storage of structured data (especially in
quantitative form) and communication are essential.
This has at least two interesting aspects. First, computer-based systems make
it much easier to aggregate, communicate, and display key information.
Important information about sales, for instance, that used to be available only
periodically (say, once each month) and lag weeks or even months behind actual
sales can now be updated daily or even in real time. Second, in a growing
number of instances, the computer systems will register and report information
that was simply unavailable before, and thereby create new feedback chains
throwing light on formerly unknown or unfathomable causal relationships. Let
us explore this in some more detail.
Availability of Information
Information technology improves the availability of information in two ways.
First, as Zuboff (1988) notes, the increasing use of computer-based systems
means that a larger and larger part of the information used and processed in an
organization is captured and registered in the
organization’s computers. The systems will often capture and retain information
that has not been collected earlier at all, because it was too difficult or too
expensive (e.g., registering every single item sold in a supermarket at the point
of sale). Second, the access to this information is greatly improved both by
storage in integrated databases allowing remote access and by machine-to-
machine communication.
Information transfers that previously had to involve many people can now
happen automatically, and with great speed. The speed itself is very important—
information registered in a database is immediately available, and information
communicated between computers moves very quickly. This means that you can
have information continuously updated in real time, without any perceptible
delay between the registration of an information item and its use hundreds or
even thousands of kilometers away.
Information Concentration
However, it is of no avail to collect heaps of information if it causes our innate
input channels to clog up. With increased information availability, we also need
information concentration—the refinement of “raw” information into a form
that is easier to comprehend. This may happen in several ways. One of the most
obvious possibilities is through transformations, as when numbers are turned
into graphs. It may also take place through aggregation, as when we compile
statistics. Statistics can then be further concentrated by being converted to
graphs. We may design compound measures—numbers that represent a
weighed synthesis of several other numbers. We may let the computers select
information items for us, and, for instance, only show us values that deviate
from the expected.
The more of these techniques we use, the more we draw on the computer’s
ability to continually trace and display the relationships between a large number
of variables, the more information are we able to monitor. In theory, the number
is almost infinitely large, in practice there are of course limits—but the limits
are only imposed by constraints on data capture and on our modeling and
programming capabilities. And, even a humble spreadsheet represents a real
extension of our working memory, showing us instantly the ramifications of
changes in single or multiple variables, thus greatly enhancing our
understanding of the total system—whether it is a budget or the layout of a
logistics operation. When represented graphically, the information is even more
accessible, as in modern brokerage systems, where brokers can follow (in real
time) the continuous changes in exchange rates, interest rates, stock or
commodity prices both as numbers (in one window) and “living” columns (in
another). The columns make it very easy to note the trends; the numbers provide
14 Comprehension and Control 321
the precision needed for actual trading decisions. Through the constant updates
they get from the screen, brokers are thus able to reflect on and manipulate more
complex relationships than before.
Other examples of systems that allow us to deal with otherwise intractable
complexity are modeling systems for complex physical processes, such as
weather forecasting systems, and modern military fire control systems. The
computerized fire control centers of modern naval units such as an aircraft
carrier group can simultaneously track and engage a large number of targets—
ships, aircraft, and missiles—using a diverse array of weapons (including its
own aircraft). Fighter planes now have computers that allow pilots to engage
several enemy aircraft simultaneously. Indeed, the military forces of the modern
industrialized countries are rapidly becoming extremely computer-intensive.
The Allied offensive against Iraq in January 1991 relied not only on modern
weapons and a lot of firepower, but just as much on a very sophisticated
communication and control infrastructure. With a high skill level in organizing
and operating computerized weapons as well as communications and control
systems, it is possible to achieve a planning capacity and a tactical coordination
on the battlefield that is simply out of reach for less skilled and more poorly
equipped forces. 1
The ultimate goal here is of course to simplify information and crop it down
to a volume small enough to absorb. This represents a new twist to an old story,
but it is also strikingly different: it means simplification by inclusion and
concentration, not by selection and omission. Whereas, as naked humans, we
had to rely on our experience and intuition to choose the few select parameters
we could manage to monitor and process, we can now build systems that allow
us to monitor all or a large number of the parameters we suspect are of interest,
and then have the systems select and concentrate information on the basis of
programmed rules. The systems may even have heuristic properties and be able
to modify themselves on the basis of accumulated measurements (artificial
intelligence). I think we are just in the beginning of a very interesting
development in this field.
1In addition, such coordination and control also build on the impersonal discipline, reliability and
efficiency that have become part of the industrialized cultures. Precise coordination of large
organizations is very difficult when one operates within a more oral culture, with its emphasis on
emotions and personal relations, and where appearance may be judged more important than fact.
That is probably why highly industrialized countries are so overwhelmingly efficient in large-scale
battlefield warfare, much more so than their firepower alone should warrant, and why forces from
less developed societies only stand a chance if they can drag the war down to the guerrilla level,
where more or less isolated man-to-man or platoon-to-platoon battles dominate.
Causal Relationships
The exposure of causal relationships and the establishment of feedback loops
are among the most important contributions of computer-based systems. The
more the activities and information in an enterprise are committed to computer-
based systems, the more the relationships between different parts of the
enterprise’s activities will be revealed and laid open for intervention. Because
information stored in machine-readable form is so much more accessible, and
the computer-based tools for analysis so much better than the old manual ones,
increased use of computers will make it possible for us to uncover deeper and
more complex causal relationships than before, and establish much more
sophisticated feedback loops. Combined with the computers’ outstanding ability
to aggregate, concentrate, and present quantitative information, this will
significantly expand the limits of what single persons or small groups can
comprehend and direct.
If all the activities that lend themselves to digital representation are indeed
represented in an integrated database, and that database is structured after a
suitable model of the enterprise’s business domain, it should be possible to
surveil and tune the total organization’s activities in a very sophisticated way—
especially in manufacturing enterprises with extensive automation. The just-in-
time production control systems of the automobile industry represent precisely
an effort in this direction, and similar effects should be possible in other chains
of enterprises making up an extended value chain (from raw materials to
retailing). On a societal level, intelligent use of computer-based systems should
make it possible to reveal interdependencies and establish feedback loops in
public administration that could allow more efficient use of public funds.
Informating Work
To Shoshana Zuboff, it is this general contribution toward a deeper
understanding and more sophisticated control of complex processes that stands
out as the most important aspect of computer-based systems. We have touched
upon this in Chapters 10 and 12, but I would like to introduce Zuboff’s concept
more directly.
In her book In the Age of the Smart Machine (1988), Zuboff opens with an
explanation of what she calls “a fundamental duality” of information
technology. While the activities of classical machines only result in concrete
products, information technology in addition to this (1988, pp. 9–10) “. . .
simultaneously generates information about the underlying productive and
administrative processes through which an organization accomplishes its work.
It provides a deeper level of transparency to
activities that had been either partially or completely opaque.” This is how
information technology goes beyond traditional automation, says Zuboff, and
coins the word informate to describe this capacity.
To Zuboff, automation and informating form a hierarchy, where informating
“derives from and builds upon automation” (1988, p. 11). Automation is nearly
always the goal when IT-based systems are introduced, says Zuboff, and up to
now informating has come largely as an unanticipated effect, which almost no
organizations have understood and very few have exploited. The informating
aspect of the technology is for Zuboff the real revolutionary one, the one that
will cause most of the organizational changes in the future. Although she
acknowledges that IT has the potential to replace large numbers of humans
through automation, in her opinion it only “perpetuates the logic of the
industrial machine, that over the course of this century, has made it possible to
rationalize work while decreasing the dependence on human skills” (1988, p.
10). Only informating can bring real change, as it “. . . alters the intrinsic
character of work—the way millions of people experience daily life on the job”
(1988, p. 11).
In Zuboff’s view, it is only by exploiting the informating aspects, the insight
it gives in core processes, that it is possible to design systems and work
organization in such a way that one can reap the full benefits of information
technology. In her eyes, the capacity for informating also represents an
appealing aspect of the technology, because it seems to favor increased use of
human intelligence, learning, and teamwork, and a concomitant decrease in
hierarchy and the application of Tayloristic principles: This is simply necessary
to reap the full benefits of computer-based systems.
I fully support Zuboff’s view that the informating capacity of computer-based
systems represents a very important and genuinely new addition to our arsenal
of tools. It is absolutely central to our growing capacity for managing complex
tasks and projects, and, in my view, it is one of the technology’s three most
important contributions—on par with hyperautomation and the coordinative
powers of the database.
However, I disagree at some points. First, I do not believe that information
technology necessarily favors empowerment and a decrease in hierarchy in
general. Like earlier communication technologies, it can be used both for
centralization and decentralization, and it is not a given that decentralization and
empowerment will be more attractive or productive in all circumstances.
Second, I do not agree that computer-based automation operates “according
to a logic that hardly differs from that of the nineteenth-century machine
system” (Zuboff 1988, p. 10). Even if many of the basic principles are the same
as those that apply to mechanical automation, I
nevertheless believe that the degree of automation we can achieve by using

computers is so dramatic in comparison with mechanical automation that it
represents something qualitatively new. The effects of hyperautomation and the
general elimination of work that can be achieved through the use of information
technology will contribute just as much as informating (or more) to the changes
we will experience in our organizations and in society.
Finally, I do not agree that such automation, or the use of information
technology for other purposes than informating, necessarily implies a
decreasing dependence on human skills: on the contrary, it entails an increasing
dependence on knowledge. However, the requirements for knowledge may well
be unevenly distributed in the organization, and I think the narrow statement
that automation decreases the dependence on human skills is based on a “local”
interpretation of skill—that is, on looking only at the concrete (and presumably
lost) skill of a worker who is replaced by machinery of some kind. In my view,
one must look at the total set of skills required for a certain production process.
To achieve a sophisticated level of automation, it is necessary to develop
equally sophisticated skills in analysis, engineering, and planning to design and
build the necessary machinery and computer systems and to operate the
resulting production units.
This is, of course, the reason why advanced automation can only be
developed and maintained by advanced industrial and scientific cultures. The
skill required to automate is actually much higher than the level sufficient to
carry out the work without automation—but the skill is of another kind; it is
more intellective, to use Zuboff’s terminology. It will also normally reside in
another part of the organization, and partly even outside the organization
itself—in consulting firms and the firms that make and install the necessary
systems and machines. Moreover, as an increasing number of routine jobs are
eliminated, the jobs left will in most instances require a higher skill level than
those eliminated, which means that the average skill level in the organization
will rise. However, the skills required in both the automated and the informated
organization will increasingly be of the intellective kind, and the ability to work
through symbols and abstract thought will become much more important.
The fact is that currently available technology already permits us to control
more complex matters than we can tackle at our present level of methodological
sophistication. The scale of manageable complexity is already limited not by the
technology itself, but by our ability to plan and design systems, and to interact
with and through them. The reason is simply that to build a system that can help
us manage complex matters, we must first understand these matters
thoroughly—as well as analyze and describe them very closely. Only then is it
possible to design the control
systems in all their painstaking detail and devise the interfaces that will allow
people to use them effectively. As we proceed along the learning curve, then,
and set out to tackle more and more complex tasks, the ability to analyze and
understand the problem domain, and then design the total system/organization
combination, quickly becomes the crucial factor—not the technology itself.

By making information extremely accessible and increasing organization
members’ understanding of both an organization’s problem domain and its
internal workings, information technology adds a new contribution to its
extensions of the space of constructible organizations. As recognized in the old
proverb “knowledge is power,” the increased knowledge should, first and
foremost, make it possible for those with access to make quicker and better
decisions and to supervise and direct more complex tasks and operations than
before. This was, by the way, the main point made by Leavitt and Whisler in
their pioneering article on the effects of computers on management (Leavitt and
Whisler 1958). They predicted that top managers would be prone to use this
opportunity to recentralize authority that had been delegated only because
overwhelming complexity had made central decisions untenable.
What kind of opportunities will this open up? Does it primarily favor the
development of more centralized, more tightly reined organizations, or of
decentralized organizations, where management layers peel off and
empowerment and self-organized team becomes the order of the day? The
answer is not evident—just as for the telephone (Pool 1983), arguments and
examples can be produced that point in both directions. In fact, the question
seems to function almost like a Rorschach test: those who think central control
is a good thing eagerly eye what they see as the opportunity to use automation,
improved communications, faster reporting, and better information retrieval and
analysis to strengthen management’s grip on the organization, whereas those
who would like to wrestle power away from bosses finally see their chance to
decentralize operations, devolve responsibility, and empower employees.
George and King (1991) has made a thorough review of the debate on
computing and centralization, drawing on 65 studies and discussions. Their
material clearly shows that there are no simple relationships to be found.
Numerous empirical studies can be marshaled in support of all the main
hypotheses—that computer use leads to centralization, that it leads to
decentralization, that they are unrelated, and that their use will only reflect the
already established propensities in the organization. George
and King conclude that there can indeed be a relationship, but that it is not a
simple causal one (1991, p. 70): “Rather, we believe this relationship is filtered
through an organization’s history and context and power structure and takes
form through management action in a manner best accounted for by
reinforcement politics perspective.” They assert that the
centralization/decentralization debate in its traditional form can be declared to
be over, but that research into the matter should continue in order to learn more
about the intricate relationships between information technology and
organizations.
For our purpose, it is still worthwhile to analyze this matter in a little more
detail, taking the nature of the technology and the way it alleviates our innate
constraints as the starting point.
The debate is in fact even older than Leavitt and Whisler’s article, as the
telephone provided some of the same advantages as computer-based systems.
Pool’s conclusion for the telephone was that it both facilitated some
centralization of control, while at the same time allowed decentralization
through a dispersal of activities (Pool 1983).
However, according to Mintzberg’s definition of decentralization (Mintzberg
1979), the physical (geographical) dispersal of facilities alone does not qualify
to be called decentralization in an organizational sense. True decentralization
must involve a decentralization of decision making and power. In this
perspective, the telephone appears to be mainly a tool for centralization, perhaps
with the qualification that it helps to democratize the organization by making it
easier to strike contacts across organizational levels and divides. (Pool cites
several authors to that effect.) Pool’s conclusion seems more valid for
computer-based systems, however. They can indeed facilitate both
centralization and decentralization. The question is what kind of centralization
and decentralization we can achieve, and, additionally, if the potential is greater
in one direction than in the other.
Possibilities for Centralization

Pool (1983) says that the telephone makes it easier to centralize control, and
that, precisely because of this, it allows greater physical dispersal of operations:
the controlled can be given a physically longer rein, since the controller is
confident that the new means of communication will enable him to maintain the
desired level of control anyway. This relationship can be seen as an aspect of a
more general relationship between control, distance, and complexity. Control is
inversely related both to the distance between the controller and the controlled,
and to the complexity of the problem domain.
From this, it follows that any technology improving communication and/or

the handling of complex information and feedback chains will improve control
if distance and complexity are kept constant, and allow greater distance and/or
complexity with an unchanged level of control.
There is much historical evidence to support this. Before the advent of radio
and international telephone and telegraph links, for instance, masters of
merchant ships had great discretion in accepting freight assignments, deciding
which ports to call at, whether to do repair work, and so on. Being away from
the home port for months, often years, at a time, they constantly had to (and
were expected to) make important business decisions on behalf of the owner.
The advent of radio, telegraph, and telephone, however, effectively reduced
them to mere navigators and crew managers, since the improved means of
communication allowed the owner to gradually bring the business decisions
home to his own office.
Diplomacy has seen the same development—when it took months to consult
one’s government, the post as ambassador was really an important one in
political terms. Today, it primarily covers certain administrative and ceremonial
functions, in addition to public relations and local information gathering (and
even intelligence activities). Pool (1983) himself vividly describe the moment
when this development was brought home to American (and other) diplomats,
when President Hoover took to the telephone to placate the French government
after he had unilaterally declared a moratorium on all war debts on June 21,
1931 to deter the German government from defaulting on its loans from US
banks. He used the telephone intensively to maintain hourly contact with the
American representatives in the major European capitals, and they also
conferred with each other. The calls were effective, and resulted in an
agreement with the French government two weeks later. These new methods
caused considerable agitation in Europe, as the instant information transfer of
the telephone forced diplomats and politicians to work at much greater speed
than they were used to. Pool quotes from an article in The New York Times
(1983, p. 88):
This breach of diplomatic precedent has startled Europe, a Belgian politician decla-
red, relating how Europe was being hustled by new American methods. It is a new
world without distances, he said, which makes diplomats feel they have outlived
their usefulness when the heads of States can discuss matters almost face to face. 2
In this example, as well as in the example of General Oyama quoted earlier, the
telephone improved control because it allowed rapid collection of information at
a distance. It also made order giving much more efficient
2HerbertHoover: The Memoirs of Herbert Hoover. New York, Macmillan, 1952, p. 72. The New
York Times, 28 June 1931, p.1; 29 June 1931, p.10.
and swift. It was used as a tool for centralizing both intelligence and command.
However, computer-based systems do much more than facilitate human
communication—as we noted earlier in this chapter, they also make information
available to anyone, and they can concentrate information and expose causal
relationships not previously known or fathomed. In addition, systems with
embedded knowledge, artificial intelligence, hyperautomation or clever use of
implicit coordination can allow wholesale elimination of tasks. By virtue of this,
information technology can facilitate centralization in at least three ways: by
furnishing managers with greatly improved information about real-time
performance, by large-scale elimination of tasks, and by automated supervision.
Task elimination was discussed in detail in the previous chapter, but I will
nevertheless include it here, since the perspective now is a little different.
Centralizing by Informating
As I said earlier, I do not agree with Zuboff’s (1988) postulation that the full
potential of computer-based systems can only be achieved through the
empowerment of organizational members and a concomitant devolvement of
power. I do agree that this is true in many instances, but I maintain that the
informating aspect of information technology also offers a potential for very
efficient centralization of power.
As we have already concluded, computers cannot move verbal information
from one person to another very much faster or over a longer distance than the
telephone can. But, by virtue of their processing capacity, coupled with the
range and speed of database access, they can automatically collect quantitative
information from a multitude of sources, aggregate it, and present it to a human
in an easily accessible form. This process can happen quickly enough to present
the information in real time or very near real time, and provide a central
management with very accurate and adequate information about the main
activities of an organization. Such automatic collection of information from a
multitude of sources was what made SAGE, the first computer-based air
defense system (deployed in 1958) so revolutionary: It made it possible to
organize a real-time, central combat control center for the air defenses of the
northern United States. This facilitation of a central command is still one of the
main functions of combat control systems.
However, let us consider instead a more civil case in point, where computer-
based systems allow increased centralization of control through automatic
collection, aggregation, and presentation of vital business information. The
much-quoted example from Benetton (Clegg 1990) provides an interesting
illustration. This Italian maker of clothing (mostly sweaters
and other knitwear) developed a business strategy that was based on real-time
monitoring of color preferences in the marketplace by the help of a computer-
based system. Their subcontractors (about 200 small family outfits in their
home region in Italy) produce only undyed clothes. Small batches of clothes in
assorted colors (assumed to be the most popular that particular season) are sent
to the Benetton shops the world over in the beginning of each season. Every sale
is registered at the cash register and transferred electronically to Benetton’s
central database, where it is aggregated with data from the other shops. It is thus
immediately available for analysis, and Benetton’s central management know
straight away which colors sell and which do not in their different markets.
They can then go on to dye the clothes that are produced accordingly. Changes
in demand throughout the season are instantly registered and reflected in
production.
The weak point in this system, by the way, is the time lag between the
registration of the sales information and the delivery of the new batches of
garments to the stores—the sales profile may well change in the time it takes to
go through the whole cycle! Ideally, therefore, delivery to the stores should be
daily; based on the sales the day before; and modified by any accumulated
experience about typical variations relating to time of the year, holidays, and
day of the week. The Swedish clothes chain H&M (Hennes & Mauritz) has a
comparable system that lets management pinpoint slow sellers early in the
season and begin selective clearance sales both to preempt their competitor’s
clearance sales and to draw extra crowds into their stores while the rest of the
collection is still “hot.”
The organizational implications of systems such as these are perhaps not
visibly dramatic, but what happens is that management in the central
headquarters has just as good a knowledge of the developments in the local
markets as the shop managers themselves, and they get it just as fast. Indeed,
because of the computer system’s ability for information concentration and
calculation of trends, central management probably knows more about the total
action in the local market than the people in the individual shops. Their
managerial reach then naturally extends much further down the organizational
hierarchy; in fact it will extend right into the shelves in the individual stores:
Because of the informating aspect of the systems, and their greatly improved
overview over customer choices from day to day, management’s effective depth
of control in the organization is greatly increased—and the shop managers’
freedom of action correspondingly reduced. What is left of it can often be taken
care of by less experienced personnel, since it is no longer necessary to have
store managers with a thorough knowledge of the local market or the trends
within the industry.
One can of course argue that the information here presented to a central
management could instead be fed back to the store managers, giving them
a tool for ordering and organizing their sales activities. Although the feedback
itself is technically unproblematic, such a procedure would not necessarily
represent an improvement for the organization as a whole. Marketing and sales
activities for chain stores have to be centrally initiated and coordinated to a
large extent, and orders based on local modifications and expectations may just
as well be less accurate than more so compared to those based on a broader
material.
Anyway, the point here is not that the technology will force a development in
one direction or another (we repudiated such determinism already in the
introductory chapters), or that one direction will, by necessity, yield better
results than another, just that the technology extends the constructible space in
both directions. Information technology thus makes it possible to centralize
command in large organizations with great geographical spread to a much larger
degree than before, and it allows central management to extend its direct reach
of supervision to a much greater depth in the organization. Information
technology here clearly enables a significant extension of direct supervision as a
coordinating mechanism. The extended mechanism is qualitatively different
from the previous version—much more so than the enhancements brought about
by the telephone, and I believe the changes merit a separate term: system-
supported supervision (Figure 14-1).

Tacit Explicit
Skills Skills

Computer Dependence


use of information technology (preliminary).
System-supported supervision usually means conscious direction of work to

great depth and/or great breadth in an organization, based on information
gathered and presented through computer-based systems. Directions to
subordinates can be given directly by personal communication (including email
and voice mail), indirectly as new parameters in application programs or new
routines to be followed, or they can even just follow as a consequence of
deliveries of goods or other concrete actions. The core is that the results of the
subordinates’ work can be monitored and directed in sufficient detail through
the system, in real time or with a negligible time lag.
Centralization by Hyperautomation and Elimination

Hyperautomation in Zuboff’s (1988) mills led to a total centralization of control
in the factory. Prior to the introduction of computer-based control systems, the
physical control of the process was spread throughout the factory; afterward,
control was centralized to one room. In principle, a fully automated factory such
as this could be controlled from a single workstation. The qualified jobs (for
skilled workers, foremen, and supervisors) outside maintenance were more or
less eliminated, and the factory organization was thus effectively truncated
below the level of the production manager and his support team.
One of the interesting properties of this brand of centralization is that it is
often not viewed as centralization at all, and may even be mistaken for
decentralization. Because the positions in such control rooms are almost
invariably given to skilled workers that formerly worked on the factory floor, it
is not infrequently interpreted in terms of devolvement of responsibility and
decentralization of power. To me, it is obviously the opposite: a centralization
of power built on the elimination of lower organizational levels. As I argued in
Chapter 12, the new powers of the control room operators are not a result of
devolvement, but of a de facto functional promotion. The fact that control room
operators often find themselves in conflict with their superiors over how the
system should be run only corroborates this—as their new responsibilities force
them to assume large parts of the role of production manager, it is just natural
that such a conflict will develop. Maybe it will be easier to see this if we
perform a thought experiment with Ceramico, the enterprise of Mrs. Raku—
Henry Mintzberg’s archetype of a growing organization (Mintzberg 1979),
briefly described in the beginning of Chapter 3.
Mrs. Raku started out by doing everything herself, and obviously had full
control of every aspect of her business. In other words, centralization was total.
As soon as she started to employ others, control began to slip, but it remained
strong as long as everyone worked in the same room. As
the enterprise expanded ever further, finally becoming a divisionalized

corporation, Mrs. Raku had to rely on a growing hierarchy of managers, and her
direct control over day-to-day activities diminished sharply. She probably felt
the frustration that so many entrepreneurs show when they suddenly have to
work through others, and she became a “normal” top manager—far removed
from the everyday details of business, and obliged to work through echelons of
people with wills and views of their own.
Now, imagine if information technology made such strides that Mrs. Raku
could totally automate production (except for maintenance and transport), as
well as most of the administrative work—having the salespeople in the field
update the production system directly, the designers’ CAD systems seamlessly
link up with the computers controlling the production machinery, purchases and
payments handled more or less automatically through EDI-type transactions,
and so on. Maybe she could then get almost all the information needed to run
the company directly from the computer systems, and could gather the few
people who really needed to make decisions in one room—or at least along one
corridor. Then much of her lost control would return, and the enterprise would
become re-centralized—in the sense that routine work and the work of middle
management would be eliminated, concentrating control in far fewer hands.
Here, both hyperautomation and system-supported supervision would be used to
its fullest potential, and to use an alluring metaphor, we could say that Ceramico
would become a joystick organization—a company where all or almost all
activities were directly controlled by one person or a very small group of
persons, with the help of sophisticated, computer-based systems.
This possibility is not as far-fetched as it may sound—in fact, the
hyperautomated pulp mill and the modern automobile factory (e.g., the Nissan
Sunderland facility) represent long steps in this direction, and we can expect the
development to continue. The great communication capacity of computer-based
systems also opens the possibility to stretch the depth of control in joystick
organizations over large geographical distances. The most likely early
candidates for such large-scale centralization will probably be process industries
with a global market as well as a global spread of their production facilities and
sources of raw materials.
Centralization by Remote Control

The success of the modern organization was built (among other things) on
extensive use of standardized, explicitly described work procedures. In a way,
the use of such standardized procedures represents a “remote control” of
organization members—their work is, to a large extent, directed by rules laid
down by a combination of managers and staff
personnel, saving huge management resources compared to organizations based

on direct supervision. And, as noted in Chapter 12, administrative computer
programs, with their embedded routines, coordination mechanisms, and
directions for work, make it possible to extend and refine this control.
However, the increasing sophistication of computer programming can be
exploited to increase such “remote control” even further by embedding AI-like
functions and making systems active in supervising and directing workers—to
let systems assist or even replace human managers.
An example of a fairly simple system of this kind is given by Zuboff (1988),
in the case of the WFSS system of Metro Tel discussed earlier. All the
maintenance work was registered in the WFSS, which also held a database with
the necessary information about all the workers. The system automatically
scheduled the workday for each worker, taking into account the location of the
tasks (including which floor in the building) and their expected duration to
minimize travel time and achieve a workload that had the best possible fit with
the length of the working day. The workers started their day by logging into the
system to receive a printout of that day’s work and reported back to the system
for each task that was completed. Even if there were still foremen in the
organization, they now worked primarily through the system, entering tasks and
monitoring workers through the statistics produced as a result of the task data—
the information about completion, etc. logged by the workers. The algorithms in
the system then set priorities, calculated the time for completion, and scheduled
work for each individual worker.
Siemens in Norway employs a similar system for its nationwide crew of
computer maintenance engineers. 3 Their system also includes an inventory
database, because, to offer speedy fixes, the service personnel keep a select
inventory of spare parts in their cars at all times. When the system “sees” that
they need new parts, replenishments are automatically dispatched by post.
An example of a more complex system is a suite of programs used in the
Mrs. Fields cookie shop chain (Walton 1989). The company tried to incorporate
into the programs Debbi Fields’ own expertise in running cookie shops,
developed in the early days of her enterprise when she personally managed one
of them. As Walton notes, cookies are perishable products, and success depends
on good management. As the number of cookie shops grew from a few to
dozens and then to hundreds, it became more and more difficult for Debbie
Fields and her growing numbers of managers to ensure that the store managers
ran the shops the way they
3Personal communication from sources in Siemens Norge.

wanted them to. Training and supervision were extremely time-consuming,

since the shops were geographically dispersed, and the company relied on a
young and inexperienced work force with high turnover rates.
The company attempted to solve these problems by applying a number of
computer-based systems. For instance, a voice mail system allowed Debbie
Fields to address all her employees (or select ones) directly whenever she
wanted, and an email system allowed the employees to address her. The core
system, however, was a program called the Daily Planner. In this program the
company tried to embed as much as possible of Debbie Fields’ own experience,
selling techniques, and management principles.
Every day, the program required the store manager to enter a number of
information items, such as a daily sales projection (based on sales the same day
last year, adjusted for a growth factor), the day of the week, the weather, and
whether it was a school day. Walton quotes one of Mrs. Fields’ regional
managers, Tom Richman, describing how the program worked from a store
managers point of view (Walton 1989, pp. 36–37):
Say, for example, it’s Tuesday, a school day. The computer goes back to the
store’s hour-by-hour, product-by-product performance on school-day Tuesdays.
Based on what you did then, the Daily Planner tells him, here’s what you’ll have to
do today, hour by hour, product by product, to meet your sales projection. It tells
him how many customers he’ll need each hour and how much he’ll have to sell
them. It tells him how many batches of cookie dough he’ll have to mix and when
to mix them to meet the demand and to minimize leftovers.
The computer revises the hourly projections and makes suggestions. The
customer count is OK, it might observe, but your average check is down. Are your
crew members doing enough suggestive selling? If, on the other hand, the
computer indicates that the customer count is down, that may suggest the manager
will want to do some sampling—chum for customers up and down the pier with a
tray of free cookie pieces or try something else, whatever he likes, to lure people
into the store.
On the other hand, the program isn’t blind to reality. It recognizes a bad day and
diminishes its hourly sales projections and baking estimates accordingly.
The Daily Planner was not issuing orders that the store managers had to follow.
It was meant as a guiding and suggestive tool. Inexperienced store managers
followed its advice most closely; older hands allowed their own experience to
override the system when they thought it best. They needed to keep sales up,
however, because their daily results were picked up by a store performance
monitoring system, revealing their results to the headquarters based “store
controllers.” Results lower than expected were sure to attract immediate
attention. The computer systems in the shops also helped in a number of
administrative tasks, such as workforce scheduling, interviewing of applicants,
payroll, and maintenance of bakery equipment.
It is interesting to compare this use of computer systems with the thought

experiment we just performed on the case of Henry Mintzberg’s Mrs. Raku.
Debbie Fields readily admitted how difficult it was for her to delegate authority
when the company began expanding. Like almost every entrepreneur, she was
loath to relinquish her direct control of the baseline activities in her company:
“Eventually,” she says (Walton, p. 39), “I was forced, kicking and screaming, to
delegate authority, because that was the only way the business could grow.” So,
she tried to hold on to that control through the use of computer systems. She
tried to remote-control her shop managers, automate performance control, and
establish a communication system that would let her completely bypass middle
management. In my view, this represents a serious effort to build the kind of
systems suite described in our discussion of Mrs. Raku, to create a real-world
example of a joystick organization.
If we compare this with a traditional approach to standardization of work—
such as the extremely detailed handbook reportedly governing all aspects of
work in McDonald’s hamburger restaurants (Morgan 1986)—the distinguishing
feature of systems like those used by Mrs. Fields are their dynamic features:
They do not only contain static rules; their algorithms adapt their output
according to circumstances (such as time of the year, day of the week, weather,
if it is a school day), to the actual sales volume, as well as to a constantly
updated repertory of past experience (sales were related to performance on the
same day in previous years). Moreover, they do not only answer queries; they
“act” proactively, giving directions for corrective action when input (in the form
of registered sales) deviates from what the systems’ designers (and thereby top
management) deem appropriate or acceptable. In addition, the systems (unlike a
handbook) report sales directly to centrally placed managers, providing them
with the ability to monitor performance on a daily basis, or even in real time, if
they so wish. This combination of system-supported supervision, programmed
routines, and a dash of artificial intelligence creates a very powerful extension
of the explicit routines of the traditional Machine Bureaucracy.
When systems replace managers, as in the examples described here, we
should expect problems to crop up if the systems severely reduce the amount of
human interaction. After all, humans are social animals and normally crave a
certain level of social contact. As we have already seen, this is exactly what
happened in the case of Metro Tel’s WFSS—most workers missed the old days
when they worked in one building as a member of a fairly stable team, and had
frequent contact with their foreman and fellow workers. The new situation
turned their job into a lonely one, traveling between largely unmanned
switching centers and only reporting to computer terminals.
In Siemens there are fewer complaints, probably because many of the service
engineers work in small communities throughout the country and would not
have had any interaction with colleagues anyway, and because they meet users
all the time on their calls (often people they know from earlier visits). They
have been deprived of their daily telephone contact with the central service
organization, though. Finally, Walton does not report any problem of this kind
from Mrs. Fields, which may indicate that there were at least no major
difficulties. In that case, it seems reasonable to assume that store managers
fulfilled their social needs through their constant interaction with crew and
customers.
Possibilities for Decentralization

It seems quite evident, then, that information technology can allow for increased
control and centralization of power, just as the telephone has. But what about
the promise of decentralization of authority and empowerment of employees?
According to Mintzberg’s definition, Pool’s example of the telephone’s
decentralizing effects was not an example of decentralization at all, only of
dispersed activities. We might add that the reason dispersal was allowed was
precisely the improved control over distance that the telephone provided. Thus,
the telephone conserved or increased centralization of power and stands out first
and foremost as a tool for extending the control of the centers of management—
if one is not ready to accept the arguments of McLuhan, Boettinger, Cherry, and
others (Pool 1983) that the telephone is functioning as a democratizing tool,
allowing easy contacts across levels and departments. “On the telephone, only
the authority of knowledge will work,” Pool quotes McLuhan (1983, p. 61) 4.
Although I readily acknowledge that any technology that makes communication
easier and less formal will have some democratizing effects, I think that the
matter is considerably more complex than the quote from McLuhan indicates,
and that the telephone in itself plays a fairly minor part in the democratization
of the workplace we have seen especially in the latter part of the twentieth
century.
When we turn to information technology in its full breadth, however, we face
a different situation—computer-based systems represent more than a new
communication channel and can extend the possibilities for genuine
decentralization in several ways.
4The quote is from Marshall McLuhan’s Understanding Media, New American Library, 1964, p.
238.
Decentralization by Information Availability

Perhaps the most basic contribution is the integrity and availability of
information stored in databases, as many middle-management tasks can be
eliminated by making updated information directly available anywhere in the
organization. Not all of the decision making disappears, though, and it tends
migrate toward that part of the organization where the information is utilized—
often positions at the customer interface. Frequently, this will also lead to
decentralization of a number of other decisions that are connected to the same
complex of information and customer services. In banks, for instance, people
working at the counter were authorized not only to open and close accounts, but
to grant limited credits and small loans as well. The implicit coordination
obtained by immediate registration of the new credit or loan in the system
ensures that an enterprising customer cannot obtain an undue amount of loans
by approaching different clerks in a number of subsidiaries. Before the on-line
age, this problem was one of the reasons why banks had to centralize such
functions.
Decentralization by De-specialization
Another opening that we have discussed already is de-specialization, where the
use of embedded knowledge, artificial intelligence, and artificial memory makes
it possible for single persons to cover a broader set of tasks than before.
Thereby, it facilitates decentralization, especially in connection with increased
availability of information. If a broader set of tasks can be gathered in one hand,
it means that there will be less need to collect information at more central points
in order to make informed decisions. It is not a universal option; it will mainly
support jobs that require people to collect information from many sources for
further processing, or for use in decision making on the basis of laws, rules, and
regulations. De-specialization has its limits, since only “hard” knowledge is
possible to embed in ordinary systems or AI systems—tacit knowledge, the kind
of “feel” developed through experience, is extremely difficult to extract and
implement.
Decentralization by Increasing the Depth of Control

This title may sound self-contradictory—especially after the discussions earlier
in this chapter. However, the informating aspects of the technology should also
be able to serve as a basis for decentralization in the sense of pushing decision-
making powers down and outward in an organization.
The focus here is on the scale of manageable complexity in an organizational

context. We have already touched upon the enormous improvements computers
have provided in tackling complexity in the scientific and engineering sector.
Models of the atmosphere, of waves, of crude oil reservoirs, of pollution—to
mention a few of the applications that can be found on the many
supercomputers and scientific workstations of the world—enable us to study
and predict the behavior of physical systems that are vastly more complex than
anything we could ever hoped to tackle in the past. But complexity is also a
challenge in terms of organization. Summing up his book Images of
Organization, Gareth Morgan says the following (1986, p. 339):
I believe that some of the most fundamental problems that we face stem from the
fact that the complexity and sophistication of our thinking do not match the
complexity and sophistication of the realities with which we have to deal. This
seems to be true in the world of organization as well as in social life more
generally. The result is that our actions are often simplistic, and at times downright
harmful.
Morgan here points to a central aspect of our continuous struggle to cope with
the world, and, more narrowly, to keep complex organizations going and ensure
that our actions have the intended outcomes. But it is not only the simplicity of
our thinking that represents a problem—part of the trouble can also be found in
the limits of our coordinative tools and abilities. Complex tasks require great
efforts in coordination, which tends to build hierarchy: information must be
gathered and related, and informed decisions can only be made by those with a
position central enough to provide them with the necessary information.
Information technology has the potential to revolutionize our abilities even in
this respect—the new tools for personal support, the coordinative powers of the
database, the greatly increased scope for automation, and the informating
capacity of computer-based systems converge to improve our capabilities. This
is perhaps the very essence of the computer as a tool—that it gives us an
unprecedented handle on complexity in almost any area of application: It
informates work, and greatly increases our depth of control by providing a more
complete understanding of the problem domain as well as exposing more of the
consequences of our own actions.
The introduction of CAD in the structural design of the Boeing 777 is an
interesting case in point. The “living” model of the evolving airplane provided
the designers with an immediate and much better understanding of how the
results of their own work meshed with that of others in the project. The
individual designer as well as the cross-departmental teams obtained a depth of
control earlier achievable only through laborious procedures rooted in a fairly
extensive project hierarchy. The new control, moreover, was much more
powerful than the old one, since it was based
on a tool that could adequately reflect the dynamics of the running design
process and thus provide almost real-time control.
The CAD system thus meant that repercussions from an addition or change in
the 777 structural design could be ascertained very quickly, both horizontally
(for adjoining parts) and vertically (for the subsequent steps in the engineering
process). Cross-departmental teams replaced significant parts of the traditional
project hierarchies, and considerable decision making power was moved closer
toward the origins of design. In this case, then, the informating aspect of a
computer-based system provided an increased depth of control, which was, in
contrast to the examples reviewed in the section on centralization, exploited by
decentralizing power and giving new responsibilities to the lower levels in the
organization—which is more in tune with Zuboff’s (1988) conclusions.
There are not many examples of this brand of decentralization, however,
where depth of control—not sharing of information or de-specialization—is the
central feature. There could be several reasons for this. First, management
traditionally seeks control. Where possibilities for increased depth of control are
found, one can assume that managers on various levels will be attracted to it and
motivated to exploit it for centralization. Second, increased depth of control is
often not planned for—the possibility only surfaces as an unintended effect and
is difficult to discover and handle. The pulp mills described by Zuboff (1988)
provide examples of this. Third, complexity is difficult—as noted in Chapter 8,
the bottleneck for building systems to handle really complex tasks is no longer
found in hardware, and not even in basic software, but mainly in our ability to
understand the problem domain, analyze the tasks, and design the appropriate
systems. Fourth, the decision-making power that is decentralized may not really
be centralized in the first place—actually, without the benefit of computer-based
systems, it may not even exist as a practical possibility! To illustrate this last
proposition, which may seem puzzling, let us consider two examples—one
actual and one hypothetical.
We have already discussed airline reservation systems in some detail.
Looking at the numbers associated with them, it strikes one at first as fairly
massive aggregations of information, but perhaps not too complex—after all, an
airplane is an airplane; a seat is a seat; and, even if there are lots of them, the
relationships in the database seem simple enough. In its first version, that may
also have been the truth. But, as always, when new opportunities open up,
humans gradually develop new ways of exploiting them, building complexity
along the way.
First of all, a seat is not simply a seat—there are several classes, each with
different pricing. This may even imply physically different seats, as first class
always has and business class often has—especially on international flights. The
number of seats of each type will vary among the
particular airplanes, and this must of course be reflected in the database. Pricing
is quite differentiated, as airlines try to attract passengers to seats that would
otherwise be empty. Airplane seats are perishable commodities; an empty seat
at takeoff is equivalent to income irretrievably lost, and the airlines will
therefore try to vary prices for particular seats on particular flights to squeeze
the maximum profit out of each planeload.
The art of balancing pricing for maximum profit per flight is called yield
management, which is today viewed as one of the airlines’ most important
competitive instruments. Good yield management can mean the difference
between loss and a handsome profit, and reservation systems allow analysts to
manipulate seat pricing and the ratio of differently priced seats in a way that
was previously simply impossible. To quote Hopper (1990, p. 121),
Computers review historical booking patterns to forecast demand for flights up to a

year in advance of their departure, monitor bookings at regular intervals, compare
our fares with competitors’ fares, and otherwise assist dozens of pricing analysts
and operations researchers. During routine periods, the system loads 200 000 new
industry fares a day. In a “fare war” environment, that figure is closer to 1.5
million fares per day.
The real decision-making power in this case falls to the pricing analysts and
operations researchers, especially since many of the changes will be time-
critical, and there will be little time for review by line management. The better
the systems become, the closer they will come to real-time control, and the less
room there will be for direct management involvement, other than as a source of
general policy directions. In Mintzberg’s terminology, this will amount to
horizontal decentralization. Insofar as the decision-making opportunities are
new and only made available through the (informating) quality of the systems, it
is a power that emerges directly at a decentralized location in the organization.
We can also note that the implicit coordination of the database will then
manifest itself—making the new fares immediately available directly at the
customer interface the world over and guide travel agents’ advice to customers.
It is important to note that this coordination is indeed direct and without human
intermediaries—the actions of the travel agents and the people in the ticket
offices, and even customers who buy directly via Internet, are directly modified
by the information entered in the database.
Now for a hypothetical example of an optimizing system (a kind of yield
management system) for the health sector in Norway. Norway has a very
comprehensive, public health care system, designed to provide (practically free
of charge) every citizen with all health services needed. The administration and
funding of its various parts are split among the national administrative level, the
counties, and the municipal authorities, causing
serious suboptimalization because vital feedback loops have been severed

through the system’s very design.
The system works like this: if you get sick, the national health service pays
your benefits until you are well again and can resume work. Now, say you need
hospital treatment—simple surgery, for instance. The hospitals are run by the
counties, and it does not matter to them (economically) that you are on sick
leave, because your benefits are paid out of central government budgets, not
theirs. So they can safely leave you waiting for hospital admission for months—
even if your operation will cost only a fraction of what you receive in benefits
while waiting.
If the counties had been a little more sophisticated, they might at least have
calculated their tax loss while you were sick and figured out that it would pay to
operate quickly anyway, but they do not. Even if you get so sick that you need
help and care at home, or admission to a nursing home, it does no economic
harm to the county 5—these services are run by the municipal authorities and are
paid over their budgets. When you finally get treated, the hospital (county) will
kick you out as soon as possible and leave you to the municipal services again if
you need further help. This is really a classic case of suboptimalization, caused
by a failure to establish the necessary feedback loops. The authority that
controls the treatment is isolated from the economic consequences of a failure to
treat the patient.
The reasons for the resilience of the established system are in large part
political/ideological (treatment shall be given to everyone without
considerations of economic character, employed should not have advantages
over unemployed), but also that very few people have actually realized the
problems caused by the lack of feedback and the costs associated with it, even if
there have been some very convincing small-scale trials. On top of this comes
the problem of solving the problems by administration—the complexity of a
coordination effort involving all levels and elements of the national health care
system seems truly overwhelming.
Technologically, however, it should be quite straightforward (even if it would
require substantial investments and a lot of work) to create a suite of computer
systems that would allow a health administration officer to calculate the
projected cost of your treatment; match it against the cost of your benefits;
check the costs and waiting times of the nearest hospitals
5Lest anyone who is not familiar with the health care system in Norway is led by this to believe that
you are not admitted to a hospital in an emergency, I hasten to add that you are indeed; the delays
apply to conditions that can await treatment (even if there are bound to be borderline cases). I
should also point out that hospitalization is free—and, as we know from economic theory, when a
service or a commodity is free there tends to be an escalation in demand and a need to regulate it by
queues. Of course, the severity of the cases in the queues is a function of the total resources used on
health services, and that is where the political discussion focuses.
and at hospitals farther off; check the prices and available capacity of certified
private hospitals in Norway and other countries, and then make a decision
representing a sensible tradeoff between your comfort and well-being and the
cost to the health care service. Such a system would also allow the health
administration to determine the total performance of the entire national service
and tune the capacities of its various parts to achieve the best possible result in
view of the given priorities. Because it would lay bare economic causal relations
that have hitherto been obscured by the very complexity of the system, it would
also give politicians a much better instrument for their decisions.
In short, a quite conventional (albeit large and complex) suite of systems
could provide a degree of control over a very complex part of the public
services that would be almost unthinkable with traditional tools, and much of
the control would end up in the hands of officials at the “customer” interface.
The national health service is perhaps an extreme example, but there are
bound to be innumerable large and small spheres of activity where
suboptimalization exists because of obscured causal relations and severed
feedback loops. Computer-based systems could be used to close those loops and
reveal the causal relations, and thus give people in responsible positions much
better instruments to manage the complexities of their domains of
responsibility—and even to extend their responsibilities considerably. The most
serious obstacle for such developments is not technical, but, rather, a question of
that old primal part of ourselves: closing feedback loops, revealing causal
relations, and extending responsibilities for someone inevitably means that
others will lose their large and small empires, have their budgets suddenly
linked with someone else’s, and so forth. Such changes are bound to be painful
and to feed organizational and political infighting, and they will be difficult to
effect—no matter how rational they seem when viewed from the outside.
The Migration of Power

All the examples in this chapter—of both centralization and decentralization—
have one thing in common. The systems build directly on the most central
properties of computer systems: their ability to store very large amounts of
information cheaply and indefinitely; to retrieve that information rapidly,
reliably, and independently of physical location; and to present it in an
accessible form, possibly also with a few analyses performed automatically
before presentation. By eliminating paper-bound information flows and by
automating information processing and presentation, the systems make it
possible to bypass traditional paper-processing administrative hierarchies and
deliver the necessary
information directly where it is needed. And, as noted earlier, when the middle
layers of the organization are bypassed (and partly eliminated), decision-making
power tends to follow the information upon which it is based: It migrates toward
the “hot spots” in the organization, the places where the needs for decisions
arise.
Here, we see the interesting split that is illustrated in the examples presented
in this chapter: Customer-related decisions are pushed toward the customer
interface, supported by systems used to retrieve information relevant for single
transactions, for specific customer-related tasks, or for critical operations in a
production process—whereas coordinating and controlling power relating to the
whole organization is drawn toward the top, exploiting automation and/or
automatic gathering and presentation of information on an aggregate level. In
those cases, as we saw in Boeing, where the total result is dependent on a very
complex process where no single point in the organization can decide, power
migrates toward the process seniors (senior professionals—those who head the
project teams).
The location of such “hot spots” may vary from organization to organization,
but will most commonly be found at the top (the strategic apex in Mintzberg’s
terminology), the customer interface, and the critical stages in the production
process. In service industries, such as banking and insurance, the customer
interface will often also represent one of the most critical stages in production.
Power and authority should therefore tend to migrate toward those critical
decision points, moving aggregate information upward and task-specific
information downward. This can happen to a large extent without disrupting the
general structure of the organization—a Machine Bureaucracy can survive this
process with flying colors and come out strengthened and rejuvenated. As
Thompson says, commenting on what he considers the postmodern theorists’
premature burial of bureaucracy (Thompson 1993, p. 190, italics in original),
Organizations are frequently becoming leaner and more decentralized, but these
trends can be interpreted very differently than the fundamental break with
centralized bureaucracy present in postmodern imagery. Essentially what we are
seeing is a duality in which the decentralization of the labour process and
production decisions (through mechanisms as diverse as profit centres,
subcontracting and quality circles); is combined with increased centralization of
power and control over the spatially dispersed, but interdependent units.
However, not everything is preserved. The middle layers of any organization

easily become big losers in this process, because much of their raison d’être is
just aggregation and processing of information for superiors, and channeling of
information among subordinates. As power migrates toward the strategic and
operational levels, therefore, the middle
layers in the organization should shrink significantly. Moreover, to exploit the

potential benefits of computer-based systems, it is necessary not only to let this
migration of authority and responsibility happen, but to promote it actively. It is
also necessary to empower the people working in the nonmanagerial “hot
spots.” Even if the framework becomes more tightly controlled (such as
strengthened financial controls), the discretionary powers of the people in the
hot spots should nevertheless be increased, if the benefits the new tools can
provide are to be fully exploited.
It is also extremely important to understand that these jobs will change
radically in the process—they will incorporate parts of the roles formerly filled
by managers and professionals on higher levels, and demand more advanced
skills, a more pronounced talent for abstract thinking, and a greater feeling of
responsibility for the organization as a whole. Not everyone will be ready or
able to make this transition.
It turns out, then, that Pool’s proposition for the telephone holds up well for
computer-based systems—the same technology that allows centralization can
also support decentralization. But, as George and King argue, the movements in
those two directions are neither similar nor mutually excluding—the
relationships between computer-based systems and organizations are intricate
and dependent on many factors, both social and technological. However,
computer-based systems will facilitate the movement of aggregate information
for business guidance toward the strategic apex, and information relevant for
actual transactions or production processes toward the customer interfaces and
factory floors—information that formerly had to be painstakingly collected,
processed, and moved in paper-based information flows, involving many people
and organizational levels, and even information so complex in origins or
processing that it had not been available with traditional tools at all. Leavitt and
Whisler’s (1958) prediction that middle management would fall upon hard
times is valid indeed.
Control: The More Sinister Aspects

So far, we have only discussed new opportunities for centralization and
decentralization in a normal, democratic setting, where due respect for human
rights and the right to privacy is taken for granted. Unfortunately, that is not
always a valid assumption. It makes one shudder to think what Hitler or Stalin
could have achieved with information technology—and even more at the
thought that they were probably not the last of their kind to appear on the world
scene.
Information technology gives people bent on control and surveillance a
dangerous new set of tools. For example, in many countries today,
you can have a small chip implanted in your dog (usually in the neck), carrying
a unique identification number (Hesselholt et. al. 1992). If the dog becomes lost,
an appropriately equipped police or veterinarian can read out the number by
placing a reader at the appropriate spot over the chip. A national register will
then inform them of the owner’s name and address. The system is also useful
for breeding purposes, and it is now routinely used to identify dogs in important
sled races. Such systems are even applied to identify fish in fish farms for
research purposes. The possibilities of this technology are daunting indeed.
A similar technology has been implemented for the collection of toll fees for
automobiles—around several Norwegian cities, for example, one can find
electronic toll stations. As a subscriber, you will have a chip glued to the
windshield of your car, and each time you pass through the toll station, a
computer system picks up the car’s identity, checking it against the central
database for the city to see if you have paid. If your subscription is valid for a
number of passages rather than for a period of time, the system will deduct one
passage from the number you have left. It performs this operation in the few
tenths of a second it takes you to pass the antenna and reach the pole with the
warning light and the camera. If you have five passages or less left, a white
signal will flash at you, and if you have none left or do not have a chip at all, a
red light will flash and your car and license plate will be photographed. A few
days later, you will receive the appropriate bill in the mail. You are not
supposed to pass at speeds higher than 70 kilometers per hour, but taxi drivers
have assured me that the system seems to work well past 100. I have not had the
nerve to test that for myself.
One can well imagine that Hitler would have been very pleased with the
ability to implant such a chip into every Jew and put up detectors in their homes
and in relevant public places—and that Stalin would have taken the opportunity
to do the same with suspected “enemies of the state.” Judging by the enormous
surveillance machinery revealed when the former East Germany collapsed, one
might even ask if some leaders would perhaps be prepared to equip all their
citizens with such a convenient identification tag, and have the muscle to
actually do it. Still more chilling, it is probably only a matter of a decade or two
before computers can recognize faces quite reliably, making surveillance even
easier and harder to detect.
Used in this way, information technology would become the first realistic
tool for achieving a measure of control at the level of or even beyond that
described by Orwell in 1984, since much of the surveillance could be
automated. Access to any area could be automatically controlled, every citizen
could be assigned individual restrictions on movement, and the patterns of
movement of persons and even groups could be
continuously monitored and analyzed by computer systems, alerting the

attention of human surveillants only when anything suspicious turned up.
Hopefully, this is somewhat beyond what we would expect in democratic
societies and the organizations in which we work. Indeed, to forestall unwanted
surveillance, governments increasingly impose new laws and regulations
restricting both their own and private corporations’ leeway in collecting and
using information on private citizens. For instance, Norwegian authorities have
instructed the company collecting the toll fees in Oslo that they are not allowed
to store information in their database about when and where a particular car has
passed. Only the balance of the car’s subscription account may be retained.
In spite of this, there is already talk about equipment that can be installed in
every car—recording not only the amount of use, but when, where, and with
what speed. The idea is that it will allow the authorities to price the use of roads
directly, not only through excise duties and toll fees, and to price it differently
according to date and time of day. No doubt, well-meaning people will also
advocate the use of such gadgets for catching speeders and generally controlling
the behavior of drivers in order to increase safety.
Related to this is the ability to monitor performance. Some of you may recall
the controversy raised in the early 1980s when Wang introduced word-
processing systems that allowed supervisors to monitor the performance of
secretaries (the number of characters typed). Similar controls can easily be
devised in many work situations involving computers. Not only can one monitor
the amount of work, but often also speed, quality, the number and length of
breaks, and general work patterns. We have certainly entered an age where
vigilance against unwarranted surveillance and control—private as well as
public—is more important than ever before. Have you, for instance, ever
thought about the record of your travels that some airline reservation systems
keep accumulating? Or the spending patterns revealed by your credit card
accounts? The history of telephone calls registered by your telephone company?
Remember that with the new digital switches now in operation in most modern
countries, the destination and duration of every single telephone call can be
recorded and stored.
To complete this chapter, I will relate a story brought to my attention by
Professor Jon Bing of the Norwegian Research Center for Computers and Law.
This story shows that even the activity pattern reflected in the database of your
local electricity board may have amazing potentials.
In the late 1970s Rudolf Clemens Wagner, one of the central terrorists in the
notorious German Rote Arme Fraktion was surprised by the police and arrested
in his flat in Hamburg. How had the police tracked him down? They suspected
that he lived in Hamburg, but they had no idea
about where. They knew, however, of some occasions when he was

demonstrably elsewhere in Germany, participating in RAF activities. By
matching his suspected travel pattern with the database over electricity
consumption kept by the Hamburg Electricity Service and analyzing the
resulting data closely for other clues, they were left with a very small number of
flats—and in one of them, the suspect was found. 6
However chilling these examples may seem, let us not make the mistake of
labeling information technology an “evil” technology. The general qualities of
computer systems that allow such monitoring are the abilities to record
information automatically; to store very large amounts of information cheaply
and indefinitely; to aggregate, transform, and analyze it automatically; and to
present the results in an easily accessible form. They are the same abilities that
make Benetton’s extended production system possible, and the examples
presented here just show us that IT can, like any other technology, be used for
both good and bad. It is up to us and our own vigilance to ensure that the
technology is not used for oppressive purposes.
6The example is reported in more detail in the third annual report of the German Federal Data
Protection Commissioner, 1981, p. 50.
349
V The New
Organizations
In the four chapters in Part IV, we analyzed the way computer-based

enhancements to our preconditions for organizing extended the space of
constructible organizations. The main extensions can be summed up thus:
• Increases in personal productivity eliminate routine jobs. This tends to
increase the ratio of team-oriented jobs in organizations.
• Computer-based systems allow improved group cooperation over distance
and can improve social cohesion in teams and groups who cannot otherwise
meet. The prospects of dispersed organizations (“virtual organizations”) are
improved, but social and emotional constraints limit their attractiveness.
• Automation and hyperautomation allow large-scale elimination of work,
even of work that cannot be automated directly. Organizations may be
truncated, thereby totally changing character and even structural
configuration—generally in adhocratic direction.
• The implicit coordination achieved through databases also allows for
extensive elimination of work. The potential is great for both simpler
(banks) and more complex work (engineering design). This will allow for
much larger organizations than before; it can make large organizations more
responsive and improve the quality and diversity of their output.
• Computer-based systems also make it easier to couple separate organizations
closer together. The coordination may be very strong, as in extended value
chains that are wholly or partly automated under common control programs.
• Implicit coordination can support large entities that are organized, but still
not constitute organizations in the classic sense.
• Computer-based systems allow extensive centralization of power through
informating. Management can surveil performance in real-time, both
aggregate and in detail, and supervision of subordinates can be automated
(“remote control”). Work elimination also contributes to centralization.
350 V The New Organizations
• Computer-based systems may allow increased decentralization through

improved information availability, de-specialization and by increasing the
depth of control. Decentralization mainly takes the form of a migration of
customer-oriented decisions toward the organization’s periphery (the
customer interface).
In a way, we have now fulfilled the original purpose of our investigation: We

have analyzed information technology’s strong and weak points, we have
established how it allows us to alter the set of organizational preconditions, and
we have analyzed how these improved preconditions in their turn allow for new
extensions to the space of constructible organizations.
However, we have still not fully analyzed the consequences of the combined
effects of these extensions. The discussion has verged on combining two or
more of the extensions at several points in the last four chapters; however, to do
so would have anticipated later discussions and caused a break in the narrative. I
have therefore waited until the last three chapters to bring the whole picture
together. Moreover, since Mintzberg’s structural configurations were so
centrally positioned in the discussion of the organizational platform in Chapter
3, I cannot end this book without discussing if and how the extensions to the
space of constructible organizations combine to modify the configurations and
perhaps create altogether new ones.
This discussion will come in Chapter 16, “The New Configurations.” Before
that, however, I must discuss two other topics that are central for understanding
how intimate the connection between organizations and computer-based
systems are becoming: that is, what will it really mean to build organizations
with information technology? In Chapter 15, “Toward the Model-Driven
Organization,” I will therefore first discuss the status of computer programs as
building blocks of organizations, when organizations are viewed as patterns of
action in line with the discussion in Chapter 2. As such programs become ever
more prominent parts of the organizational fabric, action theory will have to
confront this problem. Next, I will return to the conceptual model, which was
first discussed in Chapter 7. We noted there that this model was at the heart of
the emergence of the modern organization, which was built within the literate
paradigm. At first, organization designers were probably not aware that they
were actually constructing models and using them for organizational
improvements. Later, however, and especially after computers and computer
programming were introduced, the concept of the model and modeling activity
became very explicit. Models will be extremely important for the organizations
of the future—indeed, we seem to be heading toward a situation where active
models will make up the central elements in most organizations.
351
15 Toward the Model-Driven

Organization
“From the moment of birth we are immersed in action, and can

only fitfully guide it by taking thought.”
Alfred North Whitehead, Science and the Modern World, 1925
Organizations: Patterns of Action, Patterns of Logic

Before going on with the analysis, I would like to address the question of the
status of computer programs—the logic governing the computers—as compared
to human actions in organizations. This is especially important, since I have
supported a combination of an action and a systems approach to organization
theory.
I have argued that the modern organization was a product of the literate
society and the literate mind. Its defining feature, and indeed its foundation, was
the explicit, conscious design of the recurring patterns of action that constitute
organizations. Whereas the patterns of action that constitute organizations just
emerged within the traditional oral paradigm, they were to a much larger degree
consciously constructed within the modern literate paradigm. I also held that
automation represented the utmost in such design, but deferred to Silverman
(1970) and others in reserving the term “action” for human actions, talking
instead about “machine movements.” I could also have used the term
“behavior,” which, according to Silverman, only designates observable, outward
conduct, and can thus also be used for the operations of inanimate matter. In
Silverman’s sense, “action” is more than behavior, it implies the meanings the
actor attribute to the actions, and meaning is something that cannot exist outside
a sentient mind.
Even if they embodied a modest amount of logic, in the form of “canned

actions,” automatic mechanical machines could still be seen as belonging to the
old world of tools used to augment human work. As such, they could fairly
easily be contained within the original perspective of the action theorists,
reflected in Silverman’s definitions of “action” and “behavior.” When we enter
the computer age, however, the distinctions begin to blur, because the computer,
as it now appears, is not first and foremost a machine in the old sense of the
word. The ability to store and execute programs has made the computer a new
kind of universal machine, and modern microelectronics has increased the
amount of logic per gram of matter by so many orders of magnitude compared
to mechanical automation as to create an altogether new class of machines.
Even the average PC must chiefly be looked upon as an exceedingly complex
system of logic, a logic that represents (in executable terms) an extremely large
repertory of “canned actions” designed and implemented by systems analysts,
designers, and programmers.
Thus, even if the computers and programs themselves are inanimate and
cannot attach meaning to their behavior any more than a pebble on the beach,
the programs are the result of a painstakingly detailed analysis and design, full
of both meaning and intent on the side of their creators—a process Yates and
Benjamin (1991, p. 77) call the “capturing of procedural knowledge in computer
programs.” This meaning and intent is to a considerable degree preserved in the
structure and functioning of the programs. Of course, even computers running
sophisticated programs cannot be viewed as actors on par with humans (for
instance, no program has yet passed the Turing test 1), but they preserve and
display too much of their creator’s intents and interpretations to be brushed off
as inanimate in the old sense of the word. When using a program in the course
of their work, people will to a large extent be compelled to view a designated
part of the world (the problem domain that the program addresses) through the
eyes of the program’s creators. Most users will also have an acute feeling of
engaging in a kind of interaction when they work with their computers—not
only of wielding a tool, as when one uses a hammer to drive in nails. This
interaction will of course be partly self-referential, since
1The Turing test is an experimental setup where a person is put in a room with a terminal and a
keyboard, connected to a computer in another room that is either controlled by a program or by a
person. The person in the first room is then asked to determine if there is a machine or a person in
the other end by typing questions on the screen and watching the answers. No program has (to my
knowledge) been able to consistently pass as human in such tests. A particularly elegant way of
deciding the nature of the respondent that has been used is not to pose any questions at all. A
computer will wait patiently for ever (if its programmer has not anticipated such a situation),
whereas a human, after a fairly short while, will start asking if there is anybody out there, or if
something is wrong.
15 Toward the Model-Driven Organization 353
the computer’s response will regularely include feedback on the user’s own
actions, but it also constitutes, in large part, an interaction with the logical
structure created by the program’s designers.
This is, by the way, why it is so extremely important that systems analysts
and designers have a keen understanding of the meanings prospective users will
attach to the system’s responses—and not just concentrate on the organization’s
objectives, the designated problem domain and the tasks the system should
support/supplant/eliminate. That is what cognitive ergonomics is all about. (A
good discussion of many of the details in this process can be found in Eason
1988).
When we use computers in an organizational context, therefore, the result
will not only be a system of recurring patterns of live human action—which we
now should call living patterns of action—but a system where such patterns of
action are intertwined with patterns of logic residing in computer systems, logic
that represents carefully designed programmed patterns of action. In fact, the
combined patterns of action of people and computers may be so tightly
integrated and intertwined that it is difficult to conceive of them as separate
systems—which, indeed, I believe to be wrong anyway.
To me, it is therefore impossible to escape the conclusion that this logic,
these patterns of programmed action, must be regarded as an important part of
the total system of actions that constitute an organization. Likewise, the process
of program development must also be included, since it is there that the patterns
of programmed action are determined and translated into executable logic. To
underline the importance of this process and its vital role in the construction of
both present and (especially) future organizations, we may even rename the
program development process program construction.
The introduction of computer-based systems therefore creates a new level of
sophistication and complexity in organizations. It also creates a new level of
abstraction, since actions will be tied to symbols to a much higher degree than
before and the formalization inherent in the programmed patterns of action will
permeate much of the dialogues. The structure and functioning of the
organization will no longer be determined only by living patterns, created and
carried out in real time by the organization members and the members of
relevant parts of the environment. Programmed patterns, consciously designed,
will increasingly influence the structure and functioning of the organization.
They will influence them directly, because important parts of the organization’s
structure and functioning will be implemented in computer-based systems, and
they will influence them indirectly as well—because the nature of these
programs will exert strong influence even on the live actions of their users.
The Ascendance of the Active Model

When explicit, conscious design was put to use as a tool for the construction of
organizations, the conceptual model became its alter ego. The deliberate
analysis and the conscious planning and design processes opened the
organization and the work procedures to innovation and systematic
improvement, starting it on a trajectory that differentiated it more and more
from the familial-social-commercial continuum of organization that
characterized preindustrial society.
At first, the introduction of computer-based systems seemed only to be a
matter of continuing this process, especially since it all started with discrete
applications for narrow sets of tasks, such as accounting and filing, exploiting
the most basic properties of computer-based systems. When applications
expanded beyond the single task, and we learned to link them, we entered the
next level of computer use—where focus was shifted toward larger groups of
tasks or even complete processes, and where the new, computer-based
coordinating mechanisms emerged as important tools. It is this second level that
has produced most of the examples presented in the preceding chapters; and it is
on that level most of the development efforts today are concentrated, at least in
larger organizations. There is still much to do there—we are far yet from
exhausting the primary properties of computer-based systems and the basic,
computer-dependent mechanisms for coordinating and directing work.
However, as the use of information technology is both broadened and
deepened and our theoretical sophistication grows, I believe we will see a
development away from the (relatively) simple application of the basic
coordinating mechanisms toward a third level, which will be characterized by a
potentially dramatic ascendance of the conceptual model to a dynamic and
much more commanding position in the daily life of organizations. This is when
the difference between computer-based and previous tools will really start to
show.
From Passive to Active Models

The foundation of this development is the programmability of computers and
their rapidly increasing power, which provide the basis for the increasingly rich
repertory of software we have at our disposal. These programs are the result of
an analysis and design process much more detailed and deep-probing than any
earlier, with the possible exception of the design of fully automatic mechanical
machines—which also presupposes a detailed, complete, and unequivocal
description of the tasks to be executed. But mechanical automation is
necessarily so much more limited in scope that it really cannot be compared to
computer-based systems.
In the construction of computer programs, the modeling process itself has

become a conscious activity to a much larger degree than before. Although the
Machine Bureaucracies of the literate paradigm were certainly consciously
constructed, the designers were probably not aware that they were actually first
working out a (however rudimentary) conceptual model to use as a blueprint.
Today, we have taken a significant step forward, as detailed, conscious
modeling has become a normal part of systems design—which means that it is
also increasingly a prerequisite for organizational design, even though
organization people have not yet looked seriously at the methods of systems
design as a tool also for purely organizational development (which I think they
should). There are now several well-developed methodologies available, and the
three leading factions have even joined forces and created a common modeling
language called UML (Universal Modeling Language), which seems set to play
an important role in the future (for an introduction to UML, see Fowler 1997).
There is also a good number of computer-based tools for modeling available,
and there is a clear development toward a closer and closer integration between
analytical tools, modeling tools, and program development tools. Ideally, the
model should be the main focus of program construction and maintenance, and
the actual computer program code should be generated more or less
automatically by a combination of modeling tools and program development
tools. Considerable resources are today dedicated to this end, in both the
commercial and academic worlds.
Computer-based systems are therefore increasingly not only systems; they
are also much more clear-cut representations of conceptual models than
previous organization structures. We may in fact say that a computer-based
system incorporates its own model while also representing that model’s
expression—or at least a part of the expression, since there will usually be
human actors involved in a system-oriented dialogue (the exception is, of
course, purely technical systems without organizational references). Even that
dialogue will, however, be strongly constrained by the system’s inherent model,
which can only allow actions (accept input) that are defined within it. In
addition to being descriptive, therefore, the model inherent in a computer-based
system is not only a passive blueprint for design, it is also active, in the sense
that it becomes a part of the ongoing patterns of actions constituting the
organization. Its role in this web of actions is moreover not only receptive, but
even directive, in that its reservoir of programmed actions can generate
responses that guide or direct the actions of its human operators. Even the most
humble computer-based system, therefore, involves the modeling of a part of
the organization’s problem domain.
Take, for instance, Ford’s accounts payable system described earlier. This
system implicitly represents a model of the relationships between the functions
of buying, receiving, and paying for goods; and it stores the information
pertaining to those functions. In the model there are unequivocal definitions of
what an order is, what a delivery confirmation is, what a payment is, and how
they are related. When a shipment is received, information about it is no longer
communicated to another clerk for processing—when keyed in on the terminal
as a confirmation of a match with an outstanding order, the information is
instead communicated to the system (and thereby to the model). According to
the definitions built into the model, the system then automatically updates the
inventory record and generates a payment transaction. In addition, everyone
with access to the system can immediately see the status of that delivery if they
need to and act accordingly.
As long as the systems are few and they only address narrow, isolated parts
of the problem domain, the potential advantages of the single, computer-
extended coordinating mechanisms we have discussed in the previous chapters
will dominate. When systems multiply, their fields of operation will
increasingly meet or even overlap, resulting in both a need and a wish to
integrate their operations in order to reap the full benefits of systems use. In
turn, this will necessitate a more comprehensive and unified conceptual model
of a growing part of the organization’s problem domain, a model that will be
incorporated into the web of integrated systems. If this web of systems becomes
sufficiently comprehensive, we will reach a situation where the major part of the
operative actions constituting the organization (the interactions that are directly
relevant for the organization’s purposes) will be directed toward and through the
computer-based systems, and not directly toward other humans.
Somewhere around that point, we will cross a threshold: The main
constituting part of the organization will be the integrated computer-based
systems, their programmed patterns of action, and, implicitly, the conceptual
model they are based on. The coordination of the organization members will
then be mediated mainly by the systems and thereby (logically) by the model,
not by direct human communication. Such an organization will not only be
model based; it will be model driven, and the model, integrating several of the
computer-dependent coordinating mechanisms, will constitute a
supermechanism for coordinating organizations.
In my view, this development harbours a second paradigm shift in human
organization. Paradigm shifts are often proclaimed these days; there is hardly a
more misused word in computerdom, where even quite modest product
innovations are trumpeted as paradigm breakers. So, it is with some reluctance I
bring it up. However, the transition from a passive, descriptive model and the
three original technology-dependent
coordinating mechanisms to an active model and the five computer-dependent

ones constitutes a deep shift in the structure and inner workings of
organizations, a shift that in my view is at least comparable in scope to the shift
from the oral to the literate paradigm. I propose to call this emerging paradigm
the computerate paradigm.
I believe that most large organizations will reach a stage where they are
model-driven, and that the computerate paradigm will thus supersede the
literate. However, it will not completely replace it, since the literate paradigm
will still live on as the preferred frame of reference for organizations where it
proves too difficult to implement active models. We should also bear in mind
that the oral paradigm still dominates in the realm of small organizations, and is
likely to do so in the foreseable future. Kuhn's natural science paradigms replace
each other totally; if the term is to be used in organization science, we will have
to accept the notion of layered paradigms.
Early Examples
Can we find examples already? I believe we can. There are indeed organizations
that have already approached the computerate paradigm, at least for part of their
operations. Let us review a couple of the organizations discussed earlier with
this new perspective as a guideline. Perhaps the most instructive example is the
Boeing 777 case discussed in Chapter 13, since it involves the kind of model we
are most familiar with: a model of a physical object.
The really interesting part of the Boeing example—and what makes it a
prototype of model-driven organization—is that the CAD system the engineers
worked with did not contain unrelated data or data sets, but a carefully defined
conceptual model of the organization’s main problem domain, the airplane. On
the workstation screen, each engineer at all times—and at his or her own
discretion—had access to the visualized, fully updated (in real time) model of
not only his or her own design, but of all the adjoining designs and indeed the
model of the total construction (if so authorized). The evolving model also
allowed project managers at all levels to follow the progress of work in real
time.
If we believe the reports, the system had two main advantages: It eliminated
large parts of the traditional project bureaucracy needed to handle drawings and
coordinate the efforts of the many designers and design groups, and it allowed
for the integration of previously discrete steps in the design and engineering
process. There is little doubt that this advance in coordination did not come
from an improvement in direct interpersonal communication, but was rooted in
the way the project members now communicated indirectly but collectively
through their
individual interactions with the evolving model of the aircraft. By changing the
part of the model within his or her area of responsibility, the designer would
automatically communicate that change to all the other designers, who could, in
turn, respond directly to the change if it had consequences for their own work.
And, when project groups had to meet (which they still had to) to decide upon
design parameters or questions related to physical production, they had the
advantage of having a common, unequivocal, and updated conceptual model as
a basis for their discussions.
Now, the nature and advantages of models are fairly easy to understand as
long as we stick to the design of physical objects such as airplanes. Can we
imagine administrative models of this kind?
Let us return to the airline booking system: we can now reinterpret it as a
quite interesting model. Seemingly, it represents just a collection of aircraft
models—much simpler than the Boeing 777 model, of course, but still aircraft
models. In this problem domain, which concerns the sale and administration of
airplane passenger capacity, only a few of the airplanes’ properties are of any
interest—mainly the number of seats, seating arrangements, speed, and range.
However, the booking system does not exist to model individual, physical
airplanes, but to model flights—that is, particular airplanes flying particular
runs at particular times. The same physical airplane will therefore appear a large
number of times in the database, each time associated with a different set of
departure and destination points and departure and arrival times (maybe also
with different seating arrangements, if they are modified between flights—
which they sometimes are). Because locations and points in time are represented
in the model, the system is even capable of modeling something much more
complex—namely, the full web of all the flights present in the database, with all
the possibilities they offer for interconnections and transfers to cover routes not
served by direct flights. This is the part of the model that the travel agent and
the passenger see and care about. To the airlines, there are other aspects of it as
well—for instance requirements for crews, catering, and fuel.
For the purpose of seat reservation, then, and even for a number of the
airlines’ administrative tasks, the chief instrument for coordination is the system
and thereby its inherent model of that particular problem domain. For seat
reservation, the system is in fact the only coordination instrument. We can
therefore say that seat reservation is an example of an activity that is 100%
model driven. True, the users of this system do not constitute a traditional
organization, but there seems to be no fundamental reason why “proper
organizations” should not be able to base the full weight of their coordination
needs on active models in the same way. It will perhaps not be possible for all
organizations, due to the nature
of their problem domains, but in many instances I believe it will mainly be a

question of learning how to model and handle really complex domains.
Another interesting example, which we have already discussed at some
length in Chapter 14, is the cookie shop operation of Mrs. Fields. The system
called the Daily Planner was meant to incorporate as much of Debbie Fields’
own experience, selling techniques, and management principles as possible, in
order to enable the inexperienced shop manager to run a shop like a professional
(and to keep him or her within the style of operations that Debbie Fields
preferred). The system would generate directions to the shop manager based on
a number of parameters and an internal repository of rules.
We can now easily see that the people who made the system did in fact create
a fairly complete model of the cookie shop’s problem domain, and constructed a
program that expressed that model quite effectively: For the shop managers, it
must have been almost as if Debbie Fields were standing right next to them
throughout the day. To the Mrs. Fields organization as a whole, it meant that
supervision was by and large effected through this system and the model it
represented, and to implement changes in policy or in product mix would first
and foremost be a matter of changing the system (and thereby the model).
I do not have information about the flexibility of the system—if it was able to
adapt to local patterns of demand, for instance—but such flexibility is certainly
possible in principle: A general model can be built to adapt itself through
accumulation of local data (e.g., sales patterns), and thus adjust to some extent
to different local mixes of contingency factors. However, the adaptation can
never exceed the limits set by the definitions in the original model. If, for
example, the ethnic mix of the neighborhood is not defined as a parameter in the
model, it cannot be used for local adaptation, unless the local operators are
authorized to modify the model itself.
A Typology of Models
If we look at these examples, the models show clear differences in both design
and operation. The differences arise from their differences in purpose and are
manifested through how they use and combine basic system properties. Can we
discern some main types? The answer is a cautious “yes”—there are indeed
clearly distinguishable types of models, but we encounter the same problem
here as we do with Mintzberg’s (1979) coordinating mechanisms and their
corresponding organization types: organizations in real life seldom represent
pure forms. However, if we reconcile ourselves with the fact that our concepts,
theories, and models can never represent or explain the full richness of real
social phenomena, we can
nevertheless appreciate how apt archetypes can help us see and understand
important, often decisive, aspects of reality. Even if their explanatory power is
limited, they can nevertheless be of great help and make it possible for us to
analyze problems more accurately and to design more functional organizations.
I propose three basic kinds of models, then, each based on a combination of
computer-dependent coordinating mechanisms, and each representing a main
direction in the computer-based enlargement of the space of constructible
organizations. They are the regulating model, the mediating model and the
assisting model. For an informal characterization, we may nickname them
respectively the “boss model,” the “peer model” and the “sage model.”
The Regulating Model

As its name implies, the purpose of the regulating model is to direct and control
the activities in an organization. Regulating models often incorporate extensive
automation, and the organizations that have come closest to being driven by
regulating models today are probably the most advanced manufacturing
organizations, for example, a number of process industries and automobile
manufacturers. Perhaps the properties of regulating models are most visible in
operations such as the Nissan factory in Sunderland, briefly described in
Chapter 12: The production control system there manages virtually all aspects
of the assembly process, including the timing of deliveries from the key
suppliers located around the perimeter of the factory premises. Actually, we
may well view the combined production control systems of both the Nissan
factory and its suppliers as one supersystem, based on a master model, driving
the operations of the combined organizations.
The model behind the Daily Planner of Mrs. Fields is also an example of a
regulating model—it both directs the work of the personnel in every shop and
controls their performance—but it is different from the Nissan model in one key
aspect: it does not include the lock-step coordination of a production process
with numerous interdependent steps. In the Mrs. Fields organization the shops
are independent from each other, and have no need to coordinate their actions
the way the different stations on an assembly line must. The coordination here is
first and foremost a matter of regimentation—of securing a scrupulous and
uniform execution of company directives. We may therefore say that there are
two kinds of regulating models: a linked model, driving organizations where
tasks are interdependent, and an atomistic model, driving organizations where
they are independent.
The regulating model depends mainly on system-supported supervision,

programmed routines, and hyperautomation, but it also often incorporates
implicit coordination. If we look at the completed taxonomy of coordinating
mechanisms in Figure 15-1 this should indicate that regulating models imply a
combination of Mintzberg’s direct supervision and standardization of work—in
other words, a merger of real-time and programmed coordination. This is
exactly what I believe the development of sophisticated and comprehensive
regulating models will tend to effect. The richness, interactivity, and
computational capacity of computer-based systems will allow us to blend the
two modes in a way not previously possible, and thus allow us to construct
organizations that are large, extremely efficient, agile, and flexible. In the
extreme case, regulating models may allow what we have termed a joystick
organization: an organization where large parts of the activities are either
automated or directed

Tacit Explicit
Skills Skills

Computer Dependence

Mediating Regulating Model Assisting

Model Linked Atomistic Model

use of information technology (complete).
by the systems, but where key parameters and activities are controlled and
carried out in real time by the management.
The flexibility of an organization driven by a regulating model cannot, of
course, transcend the limits of the model, since all allowable actions and action
alternatives must be predefined and incorporated in it. Regulating models are
only possible to use when tasks, their execution, and their interrelations can be
defined in necessary detail beforehand. This is a tough demand, but not
impossible—it has also been a prerequisite for mechanical automation, for much
factory work, and for large parts of the work carried out in earlier white collar
bureaucracies such as banks and insurance companies. And, since computer-
based systems can accommodate much richer models and provide much better
work support than was previously possible, the prospects for building strongly
regulated organizations are now greatly improved and will continue to improve
in the future.
The Mediating Model

There is, however, work that is too complex and with circumstances too
dynamic for tasks and outcomes to be defined beforehand—or even work that
involves designing new products or processes, essentially creative work where
the process steps can be known, but not their content. This is the kind of work
where organizations are drawn toward the configuration Mintzberg calls an
Adhocracy, and where coordination must rely on mutual adjustment or an
adaptation of it. When efficiency cannot be sought primarily through pre-
planning and programming, the goal must be to achieve the best possible
exchange of information and ideas, to speed the process of mutual adjustment,
and to ensure that conflicts are resolved and agreements reached with a
minimum of effort.
This may sound like a cry for groupware and “networking”—but it is not.
That does not mean that the kind of systems gathered under the banner
“groupware” do not have a mission, or will not be part of systems built on a
mediating model—it only means that direct human-to-human communication is
very time-consuming, often inexact, and very often directed toward a set of
people that includes many who do not need the information and omit a few who
actually do. We will be much better helped if we can let the systems do as much
work as possible on their own, as well as help us make our own communication
more precise and directed toward only those who need it and only when they
need it.
An organization driven by a mediating model, then, is much more than an
organization consisting of teams communicating via computer
networks, accessing common information bases, and coauthoring electronic

documents. If we say that an organization is model driven, the model and the
suite of systems built on it must incorporate so much of the organization’s
functionality and dynamics that the organization members will work and
communicate mainly with the system and thus the model itself, and not with
particular people. This only pertains to task-related communication, of course,
not social exchanges.
The situation can be pictured as a group of people playing a board game:
They may chat and joke, but their effective contributions toward their common
goal (to have fun by finishing the game) and their effective, game-relevant
communication with each other is made solely through their separate, individual
interaction with the board and the game’s rules (which are reflections of the
game model). The consequences they suffer are partly a result of the vagaries of
the rules, and partly of the other players’ moves (their contributions). Any
player’s options at any point in time are a result of the rules (the model) and the
accumulated results of all the players’ previous moves; all players have a full
view of the situation at any time; and the information available is always
current.
The CAD model that was the centerpiece of the Boeing 777 design project is
precisely an example of such a model: The CAD system was the main tool both
for the people working at the design and for those doing the preparations for
manufacturing. The resulting structural model of the aircraft, which resided in
the system’s database, then functioned as the project’s prime communication
medium. The way it functioned is instructive. Changes were primarily not
communicated directly to those concerned and those who might possibly be
concerned; they were simply entered into the system (model). Those who were
concerned could then extract that part of the information they needed when they
needed it. Just as importantly, the information was not entered into the system in
separate, dedicated operations, it was in fact created there as a part of the
normal work process, as the designers and others used the system as a tool for
their day-to-day work. Creation and communication were thus merged into one
process, ensuring that the database always contained the latest version of
everybody’s contribution. When meetings were necessary to decide on design
parameters or other problems, all the participants could therefore draw their
basic information from the same coherent source. Moreover, the CAD system
itself could eliminate a lot of work by automatically finding and exposing
problems such as spatial conflicts, as well as helping to quickly resolve
questions where the system contained the pertinent information. The system
thus provided the main tools for work, structured the communication by acting
as medium, made the communication a lot more precise because of its criteria
for information entry and information creation, and made the communication
process
much more selective by eliminating most of the communication that takes place
“just to be on the safe side.”
The models behind the organization-like features of travel agent behavior
when coordinated by reservation systems are also mediating models. They may
seem to be regulating, in the sense that the systemsonly allow certain actions
and also contain inviolable rules on how transactions are to be effected, but the
essence of systems such as trading systems and reservation systems is not to
direct and control the actions of the users, as it is in production control systems
such as Nissan’s or Mrs. Fields’. Their purpose is to ensure that all the users
have access to the same status information at any time, and that this status
information always (in real time) reflects the accumulated sum of all the users’
system-relevant actions. In this way the users’ actions can be perfectly
coordinated by mutual adjustment without a single direct user-to-user message.
The users only have to access the information that is immediately relevant for
their own purposes and can safely ignore the rest.
The mediating model depends mainly on implicit coordination and
programmed routines, but it may also contain aspects of system-supported skills
to support professional work. The model’s revolutionary aspect is that it makes
true mutual adjustment a real alternative in much larger organizational settings
than before. With earlier technology, mutual adjustment in organizations of
more than a handful of people was only possible through representative and
consultative schemes, which often generated a lot of overhead—as in the project
bureaucracies of large design projects, with their innumerable drawings and
time-consuming modification procedures. Comprehensive mediating models
will effectively remove the theoretical upper limit for true mutual adjustment.
The model requirements will be no less stringent than for regulating models,
however—the problem domain must be accurately and sufficiently described,
and all relevant relations between the significant items in the domain must be
mapped. This is a very demanding task, but I believe we will gradually develop
the necessary skill, methods, and tools to tackle it in an increasing number of
instances.
The Assisting Model

There are tasks and organizations that do not belong to either of the two kinds
described thus far—organizations where the “product” is professional judgment,
and where there is little interdependence between tasks other than a need to
conform to professional quality standards. Those standards will moreover
typically be products of independent professional
communities, rather than intra-organizational authorities. Examples of such

organizations are universities, courts of justice, investment analysts, law firms,
and consulting firms with mainly senior personnel (consulting firms where a lot
of juniors do most of the work according to centrally produced methodologies
are more akin to the Mrs. Fields operation 2). Even organizations processing
mainly nonstandard cases—such as government departments and other political
and nonpolitical staff organizations—may belong to this class.
Such organizations have limited needs for coordination in the sense that there
is little interdependence between tasks. Their main goal is most often to secure
high professional standards and efficiency in work. Often, an important aspect
will be to produce outcomes that are as correct as possible according to
professional standards and as uniform as possible for comparable cases. The
model (and the systems) will therefore mainly aim at a best possible support for
the professional staff and their work, giving them easy access to both task-
specific and general information, professional standards as well as precedents.
The assisting model may seem to resemble the atomistic regulating model in
the sense that both aim to produce consistent outcomes. The crucial difference is
that the regulating model incorporates a “correct” behavior and a number of
“correct” action alternatives drawn up by the organization’s technostructure and
sanctioned by management. Its aim is to lead (and goad) often inexperienced
employees toward the “correct” organization behavior. The assisting model, on
the other hand, aims at supplying experienced professional employees with a
tool that allows them to exercise their professional judgment in the best possible
way.
If the model is limited to this, however, it is debatable whether it is able to
drive an organization at all in the way the other two models can. To be
complete, it must include the relatively modest coordination and control
functions even such knowledge organizations have. An assisting model will
therefore depend on both system-supported skills as well as programmed
routines. Sometimes, there will also be a small dose of system-supported
supervision—since one of the main concerns of management in such
organizations often is to ensure that cases are processed in due time and that
inquiries and requests receive prompt answers.
The assisting model does not, in my view, offer the potential for change and
increase in productivity that the two other models do, since there are
2This is not intended as a derogatory remark. Everyone with some experience in this business
knows that there are, by necessity, two kinds of consulting: the nonstandardized, “every case is
unique” type, which requires experienced consultants who can craft a suitable approach in each
case, and the standardized, volume type, where fairly rigid methods are necessary to produce
consistent results with less experienced personnel.
definite limits to how far this kind of work can be automated, eliminated, or
radically changed. As noted earlier, “soft” knowledge is generally not possible
to incorporate into systems, and wherever individual human judgment and
experience are central, computer-based systems will have limited impact.
Some Requirements for Model-Driven Organizations

Model Precision
The decisive factor for the feasibility of a model-driven organization is of
course that the model and the systems that build on it represent the
organization’s problem domain in a sufficiently detailed and correct way. This
includes the requirement that the model is unambiguous and that the
information contained in the systems has the necessary precision. The need for
such a high degree of formalization may seem to be contradicted by the use of
computers to achieve goals such as flexible automation (allowing a large
number of variants to be produced without retooling) or free-text searches in
large text bases, but this is a deceptive and superficial impression. Behind the
apparently effortless flexibility of advanced systems one will find extremely
detailed and time-consuming analysis and design processes, where all the
options and functions have been defined and described with utter precision.
Actually, the development effort needed for really large and complex systems is
often counted not in man-years, but in man-centuries.
The higher the precision is, then, both in data and in the definition of their
relationships, the better the prospects for eliminating or automating both work
and coordination. When the precision degrades, so does the usefulness of the
model. There is, for instance, a definite threshold of precision below which the
model of the Boeing 777 would be largely useless, because measurements
would not be within the necessary tolerances.
The same can be said about booking systems: If departure and arrival times
could only be specified to the nearest half hour, or if there was an error margin
of plus or minus 10% on the number of seats, its value as a coordinative tool
would be destroyed. This effect can be observed from time to time in the real
world, when delays due to bad weather, industrial actions, or heavy traffic force
reality out of synch with the plan-based model. The results are missed
connections, empty seats and lines of angry passengers.
We can also easily see that Ford’s accounts payable system would be less
than perfect if deliveries were not registered, incomplete deliveries
were recorded as complete, or vice versa—or if the system did not specify the
exact amount of money to be paid when goods were received and accepted.
Clearly, then, building such models is easiest when the relevant information
is quantitative or possible to assign to clear categories (which do not overlap)—
that is, when the information can be put into structured databases, and where
unambiguous relationships can be defined between data elements. However,
especially for models exploiting system-supported skills, considerable effects
can also be achieved with information in less structured forms, such as text and
pictures. For these purposes, the concept of hypertext may prove very valuable
and allow more extensive use of unstructured data than existing tools do.
Skill and Effort

I argued earlier that Zuboff’s contention that automation implies a decreasing
dependence on human skills is only correct in the “local” sense—in that it has
tended to reduce the skill level needed at the factory floor (with the notable
exception of some advanced machine tools and other instances of machinery
that demanded fairly sophisticated knowledge on the part of the operators).
However, both automation and industrialization have sharply increased the need
for skills in analysis, planning, and engineering. Sophisticated automation,
therefore, presupposes much more advanced skills than craft production both at
the company level and in the society as a whole.
This is even more true in the era of information technology. First, the
technology itself is extremely complex and continues to balance on the leading
edge of engineering knowledge. Indirectly, it is even heavily dependent on
advanced basic research in physics, materials, and mathematics. Second, the use
of information technology in an organization presupposes extensive knowledge
not only of the technology itself and of the target processes or tasks, but of how
to analyze and model those tasks. To develop more comprehensive systems and
successfully implement them in the organization, organization theory and work
psychology become important as well.
When we approach the model-driven organization, the demands grow further.
As work is increasingly informated, and more and more routine tasks are either
automated or eliminated, the remaining work will to a large degree be
conducted onscreen. It will require a fairly advanced ability to think abstractly,
understand symbols, and work through symbol manipulation. We will need an
advance in skills—at least in total skills, but often at all or most levels in the
organization as well. The new skills are indeed different from the old, and
almost always of a more abstract nature
(more intellective), but they are not less demanding. They often necessitate
quite sophisticated theoretical knowledge.
The parallel with industrialization also extends to the increasing need for a
professional technostructure. As noted earlier, computer-based systems always
require a higher degree of formalization and standardization than manual
procedures. This presupposes a detailed analysis of the relevant tasks and an
understanding of how they relate to each other—an undertaking that can be very
demanding in itself. Then the new combination of system and tasks must be
designed, preferably in such a way that the most powerful aspects of the
technology are exploited. This is no mean task, either (as they learned at Ford).
To build models and system suites for the model-driven organization only raises
the demands further. And, just as before, any subsequent change in the
organization or the way it works means changing the model as well as the
systems, requiring planning and analysis on the same level as the original
effort—making the need for a competent technostructure a permanent one in
every organization of some size that uses computer-based systems for more than
simple tool substitution.
In fact, since the use of computer-based systems requires significantly more
work on analysis, planning, and system construction than previous technologies,
and the use of such systems automate or eliminate large numbers of jobs, the
size and importance of the technostructure must increase in both relative and
absolute terms as computer use expands. However, just as during
industrialization, the increased efforts and resources that go into analysis,
planning, and systems construction will pay off handsomely—if the process is
soundly managed. In computer-intensive industries, therefore, competitiveness
will increasingly hinge on the competence of the technostructure and on its
ability to combine systems competence with knowledge about organizational
structuring and development. Top managers for this new combination (the title,
if drawn from a constructivist vocabulary, should probably be pattern manager,
but I suspect that something like organization design manager will sound more
attractive) should find themselves as sought after as top CEOs, and top
professionals in the field should become the brightest stars in the professional
firmament in the first half of the twenty-first century. Correspondingly, CEOs
without understanding of computer-based systems and the way they interact
with the organization will find themselves on an increasingly overgrown
sidetrack.
369
16 The New Configurations
“Life is a petty thing unless it is moved by the indomitable urge

to extend its boundaries. Only in proportion as we are desirous of
living more do we really live.”
José Ortega y Gasset, The Dehumanization of Art, 1925
The analysis in Chapter 3 concluded that coordination is the linchpin of

organization. In most of my subsequent discussions and analyses, I have
therefore concentrated on the evolvement of coordinating mechanisms,
especially on how technological innovations affect existing mechanisms and
allow new ones to emerge. These changes are the main enablers behind the
appearance of new types of organizations, whether they are genuinely new types
or just represent variations or extensions of old ones. Therefore, they also serve
as the main avenues for extending the space of constructible organizations.
Part IV outlined what I see as the significant computer-dependent
coordinating mechanisms, based on the previous analyses of how computer-
based systems enhance our capabilities for work, communication and
information storage and retrieval. I discussed their potential for extending the
space of constructible organizations, using both actual and imagined examples.
Although I hinted at possible new organizational configurations, the discussions
were focused on the separate coordinating mechanisms and their individual
potentials.
In Chapter 15, I also evoked a more integrative perspective by arguing that
the implementation of models in computer-based systems for the first time
makes it possible to work with active models rather than passive, turning
models into a kind of supermechanism for coordination. Models are no longer
paper-bound descriptions used as passive blueprints for design; they are
embodied by computer-based patterns of programmed actions, and thereby
become part of the total sum of the patterns of actions that define the structure
and functioning of an organization. When an organization model becomes
sufficiently comprehensive and
sophisticated and is implemented through a sufficiently integrated system suite

covering the essential parts of an organization’s problem domain, the active
model will begin to govern and drive the organization’s most significant
patterns of actions. What I have termed a model-driven organization will then
emerge—a new and revolutionary phenomenon in the organizational world,
which will become increasingly dominant in the realm of medium-to-large
organizations in the coming decades.
However, the picture painted so far is still somewhat fragmented, and it
would be advantageous to arrive at a more consolidated view, as Mintzberg
(1979) does with his structural configurations: “natural clusters” (1979, p. 302)
of the elements of his study (the coordinating mechanisms, the design
parameters, and the contingency factors) that seem to capture the salient
features of most organizations into five broad classes. We may ask, can
Mintzberg’s configurations be modified in any way in the computer age, and
can we see altogether new configurations on the horizon?
An interesting aspect of this analysis is Mintzberg’s (1979) proposition that
we will find in each configuration a dominant pull on the organization,
indicating the direction in which the organizational structure will develop if it is
not checked by environmental factors or control problems. If the concept of
pulls is correct and the pulls are correctly described, we may learn a lot by
looking at if and how computer-based systems will change the barriers for how
far an organization may be pulled in the desired direction.
In this chapter, I shall discuss the impact of information technology on each
of Mintzberg’s original configurations, assess their potential for modification,
suggest their possible evolutionary paths, and try to determine if they have the
potential of transforming themselves into new, computer-based variants. I will
then propose two altogether new configurations, the Meta-Organization and the
Organized Cloud.
Empowering the Simple Structure

I suggested in Chapter 5 that we could regard the Simple Structure and the
Adhocracy as the two fundamental organizational configurations, since they
represent the two basic ways of coordinating work. The Simple Structure is
perhaps the simplest of all the configurations, at least when we talk about
organizations larger than the handful of people who are able to communicate
freely all to all. To use Mintzberg’s own words (1979, p. 306, bold type in
original),
The Simple Structure is characterized, above all, by what it is not—elaborated.

Typically, it has little or no technostructure, few support staffers, a loose
16 The New Configurations 371
division of labor, minimal differentiation among its units, and a small managerial
hierarchy. Little of its behavior is formalized, and it makes minimal use of
planning, training and the liaison devices. It is, above all, organic. In a sense,
Simple Structure is nonstructure: it avoids using all the formal devices of structure,
and it minimizes its dependence on staff specialists. The latter are typically hired
on contract when needed, rather than encompassed permanently within the
organization.
Coordination in the Simple Structure is effected largely by direct
supervision. Specifically, power over all important decisions tends to be
centralized in the hands of the chief executive officer. Thus, the strategic apex
emerges as the key part of the structure; indeed, the structure often consists of
little more than a one-man strategic apex and an organic operating core.
The typical Simple Structure is the start-up, the small entrepreneurial firm
owned and managed by the founder, but even larger organizations can be
dominated by strong and charismatic leaders. This is even more common in less
developed countries, still greatly influenced or even dominated by the traditions
of oral culture, and where the literate forms of organization have
correspondingly less appeal. In the European/American sphere, it probably had
its heyday with the great American trusts in the late nineteenth century
(Mintzberg 1979). Sometimes, organizations with other configurations will
temporarily take on some of the characteristics of the Simple Structure if a
serious crisis renders their more elaborate decision-making schemes inadequate.
However, both in the latter case and in the case of the great American trusts, we
may question the purity of the configurations—if an organization is very large,
it will be impossible even for an extreme autocrat to have that kind of direct,
personal control over day-to-day operations that is the hallmark of the Simple
Structure. Such an organization will therefore have strong bureaucratic features,
but people will tend to look more toward the top manager’s apparent
preferences than to written rules, and the top manager will feel completely free
to intervene in any matter or decision anywhere in the organization.
Extending Direct Control

In the classic Simple Structure the defining feature is an extremely centralized
control over day-to-day affairs, most often concentrated in the hands of one
person. The predominant force pulling on such an organization is the pull of the
strategic apex to centralize (Mintzberg 1979)—to use direct supervision as far
as possible, and without any delegation. We find this very poignantly expressed
by Debbie Fields (Mrs. Fields cookie shops) in the statement quoted on p. 335,
where she concedes being “forced, kicking and screaming, to delegate authority,
because it was the only way the business could grow.” This statement, by the
way, also serves
to substantiate an assumption that underlies Mintzberg’s arguments though it is

not made explicit, namely, that growth is always a paramount objective—for
entrepreneurs and administrators alike—and that the desire for growth in most
cases is even stronger than the desire for control.
In the conflict between their wishes for control and growth, the
entrepreneurial managers of growing firms typically agonize over the delegation
of power to a layer of middle management. The reason is not only that they
dislike the fact that they will be separated from the direct contact with the
people in the operating core. As they see it, the problem is that the associated
growth most often also means an increased reliance on the more efficient
standardization of work as a coordinating mechanism. In turn, this means that
more power is relinquished, in this case to the professionals in the
technostructure who design and maintain the standardized work rules.
To Simple Structure managers, the most appealing aspects of the technology
will therefore be those that improve control and eliminate work, so that the size
of the organization can be kept down and direct control can be retained. I
believe they will be pleasantly surprised by the potential if they can rise above
their natural distrust of professional experts—because, if they really want to
exploit the new possibilities offered by information technology, they will also
have to accept the need for a competent technostructure to build and maintain
the new systems.
In order to increase control, the manager of the Simple Structure will of
course want to exploit system-supported supervision, which is the computer-
dependent version of direct supervision. This alone should make it possible to
extend the size of a genuine Simple Structure considerably. Further, the use of
programmed routines with a strongly regulating content will make it possible to
direct the actions of employees to a much larger extent and to a much greater
detail than before. Insofar as the top manager can supervise the content of these
routines directly, the use of this coordinating mechanism may give greater sense
of control than traditional written routines. This will especially be the case if the
systems allow fairly easy adjustments of rules and/or parameters. Such systems
may therefore also allow Simple Structures to grow larger without becoming
fully-fledged Machine Bureaucracies. However, there is a threshold here where
control will cease to come directly from the hands of the top manager and pass
into a process of rule standardization where the decisive influence is wielded by
a larger set of people. The organization structure will then topple over and
become a variant of the computer-supported Machine Bureaucracy discussed in
the next section.
Finally, extensive hyperautomation and elimination of work can allow
extensive reductions in the number of employees in an organization, while
keeping up or even increasing its economic size. This may allow Simple
Structures to expand further into the territory of mass-producing Machine
Bureaucracies than before. However, since the Simple Structure cannot easily
accommodate really large organizations, such an expansion would probably at
some point lead to either stagnation or a transition to a computer-supported
bureaucratic form.
It is of course also possible that the increased flexibility offered by computer-
based automation can be harnessed to lower the production costs of small
batches or even semi–tailor-made products to a level where they can compete
directly with the products of traditional mass-producing Machine Bureaucracies.
In such a case, a multitude of small Simple Structures may develop and
effectively replace formerly dominant mass-producers. This would be in line
with the ideas of flexible specialization on the basis of craft traditions put
forward by Piore and Sabel (1984). As they point out, such developments have
taken place before, although on a different basis. In particular, they point to the
developments in the textile industry in Italy’s Prato area in the 1930s and 1950s.
To exploit computer-based systems to extend operations without
relinquishing control, managers of Simple Structures will have to learn quite a
lot about the technology and the systems in use in the organization, since they
will have to use the systems themselves in order to achieve the control they
want. They will also have to learn to work closely with the computer
professionals in their new technostructure. In fact, a significant part of their
control efforts will have to be channeled into the supervision of systems
construction and maintenance.
In short, I believe that computer-based systems have enlarged the space of
constructible organizations considerably in the direction of allowing increased
size and economic clout for organizations configured as Simple Structures.
Technology-conscious entrepreneurs and autocrats should therefore have the
opportunity to invigorate this configuration and perhaps even increase its
importance relative to other configurations. In organizations that have grown
too large for the pure configuration, it should be possible—at least in a number
of instances—to revert back to a more clean-cut situation through significant
workforce reductions coupled with computer-supported supervision.
Does this mean that the configuration itself is modified, or is it only a matter
of an electronic invigoration of the traditional Simple Structure? I think there is
a continuum building here that will eventually pass the threshold to a
qualitatively new configuration.
Even moderate use of computer-based systems will allow a Simple Structure
to outgrow the limits of its pre-computer forebears without changing very much
in principle. However, as system use develops and covers larger and larger parts
of the operations, things begin to change.
Control is increasingly effected through the systems, and a growing part of the
top manager’s time is devoted not to supervision in the form of face-to-face
contact, but to systems design, parameter setting, and system-supported
supervision. With maximum exploitation of the technology, the result should
approach the organization described in the thought experiment with Ceramico at
the end of Chapter 14.
Emergence of the Joystick Organization

At this point, I think the threshold has definitely been crossed, and a new,
computer-based configuration has emerged. I think this new breed needs a name
of its own. I propose to keep the name suggested in the discussion of the
hypothetical Ceramico example, and will thus call this computer-dependent
version of the Simple Structure the Joystick Organization.
It will definitely be model driven and rely on a regulating model with a clear
emphasis on information aggregation (system-supported supervision) and easy
manipulation of key parameters in the controlling systems (programmed
routines). There will often be extensive automation in the operating core, even
to the point of organizational truncation. The top manager will run the
organization mainly through interaction with the systems, not with people—
other than a few close assistants. The Joystick Organization will continue to
cherish the centralization of the Simple Structure and will keep its forebear’s
organic structure and low degree of specialization and formalization. It will
thrive in the same environments (simple and dynamic), but it may grow larger
than the Simple Structure, at least in economic size. Contrary to the Simple
Structure, however, the new configuration will have a significant
technostructure, since it will need a sophisticated IS department to take care of
(and often develop) the extensive systems needed for its daily operations. The
head of this department will be one of the top manager’s closest collaborators.
Some readers may protest that such a revival of a modernized Simple
Structure flies in the face of the common prophecy that information technology
will first and foremost promote flatter organizations and a greater reliance on
teams, cooperation, and devolvement of power. My answer to this is that the
technology in itself does not promote any particular arrangement; it is an
enabler that opens up possibilities in a number of directions—just like earlier
technology. Within the constructible space available to them (the local space,
which is the constructible space restricted by local contingency factors),
individual actors will exploit the technology in the direction they prefer. People
who favor cooperation and devolvement will seek out features supporting
teamwork and decentralized decision making; people who want control will
move in the opposite direction.
As an illustration, we may note that all the while Marshall McLuhan was
writing about the democratizing effects of the telephone in the corporation (“On
the telephone only the authority of knowledge will work.” 1), Harold Geneen
was using the telephone as one of his prime instruments of control at ITT (some
would even say instrument of terror!). To keep his subordinates on their toes, he
would telephone them at any time, day or night, demanding rapid answers to
questions about their operations (Schoenberg 1985).
Perfecting the Machine Bureaucracy

The Machine Bureaucracy is the epitome of the modern organization—indeed,
many organization theorists just call this structural configuration the Modern
Organization. Its defining feature is its use of standardization of work processes
as its main coordinating mechanism. It achieves its efficiency through mass-
production of goods or services in a highly rationalized operating core. The
degree of formalization in the organization is high, and most tasks are highly
specialized. Since the operating core of the Machine Bureaucracy can only
achieve its impressive productivity through continuous production at a high rate
of facility utilization, and since changes in the production setup are costly and
time-consuming, it craves a high degree of stability in its environment. Often, it
tries to influence its environment both directly and indirectly to increase
stability.
To manage the normally quite complex organization and maintain its
operating core, the Machine Bureaucracy has both an elaborate administrative
structure and a well-developed technostructure. In fact, Mintzberg (1979) points
to the technostructure as the key part of the organization, even if the formal
power resides in the line managers (Mintzberg 1979, pp. 316–17, bold type and
italics in original):
Because the Machine Bureaucracy depends primarily on the standardization of its

operating work processes for coordination, the technostructure—which houses the
analysts who do the standardizing—emerges as the key part of the structure. This
is so despite the fact that the Machine Bureaucracy sharply distinguishes between
line and staff. To the line managers is delegated the formal authority for the
operating units; the technocratic staff—officially at least—merely advises. But
without standardizers—the cadre of work study analysts, job description designers,
schedulers, quality control engineers, planners, budgeters, MIS designers,
accountants, operations researchers, and many, many more—the structure simply
could not function. Hence, despite the lack of formal authority, considerable
informal power rests
1Marshall McLuhan, Understanding Media, New American Library, New York, 1964. Quoted in
Pool 1983.
with the analysts of the technostructure—those who standardize everyone else’s

work.
Typical Machine Bureaucracies are well-established, large organizations such as

banks, insurance companies, automobile companies, airlines, and government
services (e.g., the Customs Service or Social Security). Even the police and the
armed forces are usually configured as Machine Bureaucracies. Because it is
first and foremost a configuration for mass production, the Machine
Bureaucracy is optimized for efficiency within a quite narrow domain, and it
cannot easily adapt itself to another. It is definitely not able to live with very
dynamic or very complex environments. However, its efficiency in producing
standardized goods and services is so superior that it has become the dominant
structural configuration for larger organizations in all modern societies. The
overwhelming majority of us seem, in most instances, to prefer these
standardized, cheap products to the more tailor-made (but also more expensive)
products we could have had from enterprises with other configurations.
According to Mintzberg (1979), the main pull on a Machine Bureaucracy is
the pull of the technostructure to standardize—that is, to increase and refine the
use of standardization of work processes as the organization’s coordinating
mechanism. This reflects the inclination and training of the technostructure and
also serves to strengthen its power in the organization.
An elaboration of the standardization of work is indeed one of the main
avenues that information technology opens up, and unchecked technostructures
thus have ample opportunities to engage in their favorite pursuit. However,
dependent on the nature of the organization’s operating core and the skills and
vision of its technostructure, several scenarios are possible. The most important
are staying within tradition, truncation through automation or self-service, and
flexibilization.
Staying within Tradition

The dominant approach today is to stay within tradition, using information
technology mainly in a reinforcing way—that is, to make existing procedures
more efficient. Programmed routines are gradually substituted for explicit
routines, there is incremental automation of routine tasks, and management uses
information available in the systems to improve its control over the organization
and its operations. Even what are regarded as state-of-the-art improvements do
not necessarily bring a Machine Bureaucracy out of the traditional mold—
groupware, workflow tools and even business process reengineering (BPR) can
well be applied within the traditional structure. For example, the reengineering
of Ford’s
accounts payable function did not change Ford into an nonbureaucratic

organization. It is even doubtful whether the accounts payable function itself
became less bureaucratic through its reengineering. It can be argued that by
shifting the handling of payments over to a set of programmed routines, the task
rather became more strictly standardized than before, and the company’s
relationship with its suppliers less flexible. Instead of receiving the shipments as
they came and sorting out any problems afterward, all deviations from the
agreed delivery schedules were now immediately detected and triggered the
same, standardized response: a rejection of the whole shipment and a demand
for a corrected one. However, the change also entailed that the internal
coordination in Ford improved dramatically and that the company’s relationship
with its suppliers became far better coordinated and much more closely
controlled, since the new system allowed the control to take place in real time,
and not after the fact. This is how the benefits traditionally have accrued within
the Machine Bureaucracy: through increased and improved design, control and
regulation.
Truncation
Of course, if the process of automation and elimination is carried far enough,
most or even all of the operating core may become automated, effectively
truncating the organization as in the process industries described by Zuboff
(1988) and discussed in Chapter 12. According to Mintzberg, the configuration
then changes to a variant of the Adhocracy (Mintzberg 1979, p. 458, italics in
original):
The problem of motivating uninterested operators disappears, and with it goes the
control mentality that permeates the Machine Bureaucracy; the distinction between
line and staff blurs (machines being indifferent as to who turns their knobs), which
leads to another important reduction in conflict; the technostructure loses its
influence, since control is built into the machinery by its designers rather than
imposed on workers by the rules and standards of the analysts. Overall, the
administrative structure becomes more decentralized and organic, emerging as the
type we call the automated adhocracy.
However, there is an important difference between the traditional automation

Mintzberg refers to and the computer-based hyperautomation that is our subject.
Design and control are not laid down once and for all before installation of the
machinery. The reason is simply that computer-based systems are so much more
complex than mechanical systems—they allow much more sophisticated and
flexible automation. They may also allow a truncation by a combination of
hyperautomation
and increased self-service, which is what we will see at least in large sections of
the market for banking services and payment transactions.
In any case, the result is that an organization with a hyperautomated
operating core will need a sizable and competent technostructure not only to
look after it, tune it, and continuously improve it, but also to prepare the
extensive parameter controls that such systems allow and assist the line
organization in their use of these controls. The more sophisticated the
organization becomes in using information technology and the more it (hyper-)
automates, the more the technostructure will grow and the more important it
will become.
An organizational truncation based on information technology should
therefore result instead in an Administrative Adhocracy—the configuration
Mintzberg designates for organizations with very complex technical systems.
Just like its sibling, the Operating Adhocracy, an Administrative Adhocracy is
mainly project oriented. However, in contrast with the Operating Adhocracy, its
projects are not organized to fulfill customer needs, but to take care of the
Administrative Adhocracy’s own internal needs: the upkeep and development of
a mass-producing operating core.
If the organization is not too large, automation extensive, and the top
manager of the appropriate kind, such an organization may even revert to a
Simple Structure or become a Joystick Organization. Even if it keeps a team-
based management style, it may approach the Joystick Organization, since the
top management team in such an organization may develop a very dominating
position.
The Rise of Flexible Bureaucracy

The most exiting development, however, goes along the flexibility dimension
and can provide us with a revitalized version of the Machine Bureaucracy,
supplying the mass commodities and services of the twenty-first century. The
key here is the transition from inflexible to flexible standardization.
The Achilles Heel of Machine Bureaucracy

The Machine Bureaucracy was developed as an organization for cheap, reliable
mass production of standardized products and services. In order to maximize
productivity and minimize the need for training, it depended on a high degree of
job specialization, detailed prescriptions and rules for the execution of work,
along with fairly rigorous standards of quality and generous amounts of control.
This produced a type of organization with unsurpassed efficiency within its very
narrow problem domain, but, this efficiency was bought at the price of an
almost total inability to
tackle problems that was outside its underlying conceptual model and thereby
not defined in the implemented routines. In short, the Machine Bureaucracy is
extremely inflexible compared to other organizations, especially the Simple
Structure and the Adhocracy. In manufacturing, the costs of retooling for a new
product are considerable; in white-collar bureaucracies, it is both time-
consuming and expensive to change operating procedures and to train people to
solve new classes of problems.
This is why a Machine Bureaucracy is not suited for dynamic environments
or for products that cannot be standardized, and why Piore and Sabel (1984) and
Pine (1993) believe it needs to be relieved as the dominant configuration for
producing goods and services in advanced economies. Piore and Sabel, in
particular, argue that the main reason behind the apparent sluggishness of the
world economy the last couple of decades is precisely the mismatch between the
increasingly saturated and more rapidly changing world markets on one hand,
and the Machine Bureaucracies’ inflexibility and dependence on long,
uninterrupted production runs at full capacity on the other. The uncertainty
about market developments deters new investments, and the great costs of
renewing products—not to mention changing lines of business—impede the
ability of the economy as a whole to shift resources quickly enough between
changing demands.
Will the Answer Be Small and Nimble or Big and Flexible?

The remedy proposed by Piore and Sabel is to stimulate the growth of
technology-based flexible specialization, based on the pattern of craft
production. The companies they envisage will typically be fairly small (most of
them would probably be Simple Structures) and flexible enough to be able to
shift their production quickly between a fairly wide range of products and do so
with moderate costs.
Pine, in his turn, describes how old style mass production is giving way to
mass customization, where flexible production lines can deliver products with
great variation, where product development cycles have been shortened, and
lead times reduced to a point where even cars can be delivered to customer
specifications within a few days. Pine lists three forms of companies that make
up what he terms the “New Competition”: Japan, Inc. (the typical Japanese
manufacturing company), the Flexible Specialization firms described by Piore
and Sabel, and the Dynamic Extended Enterprise, exemplified by the renewed
American corporation. Pine says that although the three forms are clearly
different, they are all variations of the model offered by Piore and Sabel,
bringing back much of the flexibility of the craft-based firms of the American
System of Manufactures (described briefly in Chapter 7).
However, as Pine’s own examples show, it is not only the craft-based, small-
scale company that can achieve flexibility. Computer-based systems may also
allow the development of a more flexible Machine Bureaucracy, which may
answer at least parts of the challenge of more dynamic markets, and become
formidable competitors both for craft-based production and traditionalists
among their own kind. The basis for this is of course the extreme (and
increasing) richness and flexibility of computer-based systems, so dramatically
different from traditional machinery and media for automation and
implementation of routines. By relying on information technology and
appropriate reorganization, bureaucracies in both manufacturing and service
industries can become much more agile and achieve much greater flexibility in
their production.
Flexibility can be increased in three ways: by having a richer set of pre-
defined (and routinized) problem definitions and responses, by increasing
discretion at the organization perimeter—something that will improve the
organization’s capability to deal with problems within its problem domain but
inadequately provided for in established routines—and by making it cheaper
and easier to change the routines themselves. All of these will also help to
improve an organization’s agility. Agility will likewise be enhanced by more
efficient and rapid internal coordination and by the availability (especially to top
management) of more comprehensive and timely information about various key
aspects of the organization’s performance. The discussions in the previous
chapters have established that computer-based systems can make significant
contributions in all these areas.
First of all, computer-based automation can produce more complex products
and accommodate much larger variations in product types. I believe we have
only seen the beginning here, since we have only started to exploit the vast
potential of computer control in our development of production methods and
materials. Even now, car manufacturers are able to produce not only cars with
different colors and a wide selection of options on a single assembly line, but
even different models—and still operate according to the rather extreme just-in-
time principles reported from Nissan’s Sunderland operation. This means that
the mix of models can be changed dynamically not only from day to day, but, in
principle, from hour to hour. This is a considerable improvement over the
situation not too many years ago, wherein one assembly line could only produce
one model with a fairly limited number of options. Of course, not even these
factories can suddenly change their production to boats or airplanes, but this
increase in flexibility—which is still primarily built on traditional materials and
production methods—is nevertheless a harbinger of a future development where
even greater flexibility will be available. We may, for instance, imagine
materials and production machinery that do not depend on molds, but, rather,
can produce any shape designed in a CAD program
on the fly. Actually, we already have a beginning in stereolithography, where a

resin is hardened by an ultraviolet laser beam “drawing” on its surface, layer by
layer forming a three-dimensional object on a platform that is successively
lowered into the resin bath. This process is still very slow, cumbersome, and
expensive, but already it is efficient enough to replace a lot of traditional wood,
plaster, and plastic modeling in the making of prototypes.
The same principles apply to clerical work, only to a greater degree, since
offices have much less mechanical euipment (which is the least flexible part).
Both automated and programmed routines can be very complex, and allow for a
large number of predefined actions. Of course, it is also possible to prepare a
great number of alternatives in a manual environment, where routines are
documented in writing. However, the limits of human memory and the
cumbersome nature of written documentation will combine to restrict the
variation that can be sustained in practice. Well-designed computer-based
systems can easily extend the practical number of routines considerably and
assist in choosing the right one for each occasion. System-supported de-
specialization may increase flexibility further by increasing the range of tasks
that one person can execute, thus making it possible to accommodate greater
variations in task mix than before. This has been evident for a long time in
banks, where the introduction of counter terminals led to a significant de-
specialization.
We may also return to an earlier example. The changes at IBM Credit
involved automation of the larger part of the work and extensive computer
support for most of what was left. The caseworkers thus achieved a much
broader span of competence. In addition to the documented leap in productivity,
we should expect the new setup to provide greater flexibility in setting up
nonstandard deals, although Hammer and Champy do not comment on it. They
do, however, include IBM Credit among their examples of organizations that
have increased employee empowerment—something that normally entails
greater ability to tailor responses to customer requests. We have also concluded
earlier that the migration of power toward the decision-making hot spots in the
organization should lead to greater flexibility and agility for bureaucratic
organizations.
Just as important as an increased repertory of predefined routines is the
ability to change routines or establish new ones quickly and cheaply. In a
manual environment, changes are theoretically straightforward to implement,
since all it takes is to describe the change and circulate it to all concerned.
However, as all who have lived in and with such organizations know, it can be
extremely difficult to make changes take hold, and override established patterns
of action. The required effort can be quite considerable even for small changes.
One of the great advantages of computer-based systems in this respect is the

way they can be equipped with options and parameters for adjustments in their
functioning. We saw, for instance, how the people responsible for yield
management in American Airlines could use the reservation system to
implement instant changes in the prices and options available on flights
throughout the world—changes that would probably have taken weeks to effect
by manual means. And, despite the fact that many view bank systems as both
archaic and hard to change, there is absolutely no doubt that banks today have
more room to maneuver than they did 40 years ago when it comes to rapid
changes in their products—whether it is to meet attacks from competitors or to
accommodate general changes in their customers’ preferences. It has, for
instance, become much easier to vary interest rates, to let interest rates depend
on the dynamic size of deposits, to differentiate payment options, and to
combine accounts in various ways.
Looking at existing systems, one will of course find great differences in how
pliable they are, but rigid systems are more a result of poor analyses, poor
modeling, and poor design than of technological constraints. Naturally, there are
limits to the flexibility that can be achieved without renewing or replacing the
systems (which involves great cost and effort), but I believe we have a long way
to go before the potential is exhausted.
Big Will Still Be Better

Like Paul Thompson (1993), I therefore disagree with the postmodernist
contention that information technology is paving the way for a decisive break
with Machine Bureaucracy as the dominating structural configuration of larger
organizations in the advanced economies. On the contrary, I believe that
information technology is already supporting the development of a leaner, much
more flexible and much more agile type of bureaucratic organization than the
classic Machine Bureaucracy.
This new type of organization will depend mainly on hyperautomation and
programmed routines but will also draw on the other computer-dependent
coordinating mechanisms represented in Figure 15-1 (p. 361). When sufficiently
advanced, it will be model driven, relying on a regulating model with an
emphasis on programmed routines, hyperautomation, and (depending on the
type of production) implicit coordination. The middle layers in the organization
will be severely decimated, some of their power migrating upward and some
downward. Through the use of improved information access and increased
spans of competence, discretionary powers in matters related to specific tasks or
customers will be decentralized to the operating core. Top management,
however, will have much better control of the operations overall, both through
their
access to much better and more timely information and through the much
improved process control they achieve through their command over the systems
used by the operating core.
This kind of organization will be much better equipped than its predecessor
to tackle variation in its environment, because it reacts much faster to changes,
has a wider repertory of standardized responses, and has a greater ability to vary
its product mix. To phrase it in the language of Ashby’s Law of Requisite
Variety (1956), it will contain within itself a greater variety than the classic
Machine Bureaucracy, and it will therefore be able to live with more variation
in its environment. Because such an organization will be much more change
oriented in general, it will also have a greater ability to accomplish those major
changes that must come when the demands from the environment finally
outstrip its normal range of responses. I propose the name Flexible Bureaucracy
for this configuration, to denote both its strong points and its origin.
The Flexible Bureaucracy will retain most of the design parameters of the
Machine Bureaucracy, such as behavior formalization, vertical and horizontal
job specialization, large operating unit size, vertical and limited horizontal
decentralization and action planning. However, whereas the Machine
Bureaucracy usually relies on functional grouping, the Flexible Bureaucracy
will use its computer-based systems to maintain market-oriented grouping or
even matrix-like structures. The Flexible Bureaucracy will be able to thrive in
more complex and dynamic environments than the Machine Bureaucracy. It will
retain strong technocratic control, since the computer professionals required to
design and run its comprehensive systems will find a natural home in the
technostructure.
I believe that this computer-based version of the bureaucratic configuration
will prove a far more vigorous successor to the Modern Organization (Machine
Bureaucracy) than the craft-oriented alternative proposed by Piore and Sabel,
and, accordingly, that flexible standardization is a more likely solution to the
problems of the classic Machine Bureaucracy than flexible specialization. Far
from promoting the small organization, information technology (which is in its
essence an automating and coordinating technology) will invigorate the larger
organizations and make them still more formidable competitors. Indeed, of the
three forms of New Competition that Pine (1993) defines within the field of
mass customization, two of them (Japan, Inc., and the Dynamic Extended
Enterprise) correspond much more closely to the Flexible Bureaucracy than to
the craft-based type of firms envisioned by Piore and Sabel.
As a part of their metamorposis, Machine Bureaucracies are now
experiencing a period of contraction while they hyperautomate an increasing
part of their operating cores and shed organizational layers by
gradually shifting their coordination toward the computer-dependent

coordinating mechanisms. This is, by the way, a process that has really been
underway for quite some time—as early as around 1970, the employment
figures of large manufacturing companies such as General Motors, Philips, and
Unilever started to drop, whereas sales and capital expeditures continued to
grow (Huppes 1987). After they have made this transition, they may well start
to grow again—even in employees—although their most decisive growth will
still be in economic size.
The adaptability of the Flexible Bureaucracy is not limitless, however. If it is
confronted with problems not defined in its underlying model or requests
outside the available range of responses defined in its systems, even as they are
supplemented by empowered employees, it will come up against the same need
for fundamental changes as a Machine Bureaucracy. Indeed, so will all
organizations with a heavy reliance on computer-based systems. Even the IT-
based Simple Structures will find that they cannot “turn on a dime” as easily as
their noncomputerized brethren. Because of the enormous amount of analysis,
planning, and design needed to create comprehensive systems and the
conceptual models they must be based on, the required effort for major change
can indeed be large. There are numerous examples of such changes that have
turned into catastrophes when major new systems have been severely delayed,
have suffered massive cost overruns, or have even stranded altogether.
However, there are also numerous examples of successful projects of this kind,
and as our knowledge improves, our experience accumulates, and the software
tools become better, the successes will probably slowly increase their share of
the total. The Flexible Bureaucracy will, as other computer-dependent
configurations, have another advantage: the people working in them will be
more used to, and thereby probably more receptive to, change.
The Enduring Professional Bureaucracy

The Professional Bureaucracy is similar to the Machine Bureaucracy in the
sense that it is meant to produce standardized goods or services in an efficient
way. It differs from the Machine Bureaucracy in that its production process
(Mintzberg: “operating work”) is too complex to rely on low-skilled operators
working according to explicit routines. As Mintzberg says (1979, pp. 348–49,
italics in original),
We have seen evidence at various points in this book that organizations can be
bureaucratic without being centralized. Their operating work is stable,
leading to “predetermined or predictable, in effect, standardized” 2 behavior, but it

is also complex, and so must be controlled directly by the operators who do it.
Hence, the organization turns to the one coordinating mechanism that allows for
standardization and decentralization at the same time, namely the standardization
of skills. This gives rise to a structural configuration sometimes called Professional
Bureaucracy, common in universities, general hospitals, school systems, public
accounting firms, social work agencies, and craft production firms. All rely on the
skill and knowledge of their operating professionals to function; all produce
standard products or services.
For their operating cores, the Professional Bureaucracies rely on

professionals—people who have received their main training in independent
educational institutions. (The exceptions are of course these educational
institutions themselves; they tend to count many of their own graduates among
their employees.) This education not only provides them with the basic
knowledge they need to carry out their work, it also teaches them what to expect
from their professional coworkers and how it is customary to coordinate
activities with them. The education normally also serves to indoctrinate the
professionals with norms about ethical standards and proper conduct both
toward fellow professionals and customers/clients. Even in those instances
where systematic education continues after hiring (as in hospitals that educate
specialists), the content and process are fully controlled by standards set by the
larger professional community. There is little room for organization-specific
programs. In their work the professionals work relatively independent of their
colleagues but usually maintain a close relationship with the customers or
clients they serve. Their decisions and the way they carry out their work are
determined not so much by in-house rules as by their own judgment, built on the
standards of their own profession.
Whereas Machine Bureaucracies generate their own standards and rely on
formal authority, then, Professional Bureaucracies apply standards set by self-
governing professional associations and rely on the authority of recognized
expertise. The main pull in such an organization (Mintzberg 1979) is to
professionalize—to extend the supremacy of professional expertise throughout
the organization. This pull has three main expressions. Occupational groups not
yet recognized as separate professions will strive toward such recognition, the
recognized professional groups will fight for the inclusion of more prestigious
tasks into their domains and if possible secure statutory monopoly on their jobs,
and all the professionals will vigorously defend their own autonomy and join in
the effort to
2Mintzberg here quotes himself from his definition of bureaucracy in an earlier chapter, which he
refers to in a parenthesis I have left out here.
keep control of the organization and relegate the administrative apparatus

(including the managing director) to a subordinate, staff-like position.
This tendency is easy to observe, not least in hospitals (at least in Norwegian
hospitals). Almost every occupational group in hospitals has worked
systematically to establish a separate profession, complete with its own separate
education and statutory provisions for a monopoly on certain positions. The
doctors were first, followed by nurses, and later we have seen the same
development for most of the other groups, such as physiotherapists, physical
chemists, occupational therapists, and nursing assistants. The establishment of
new professions has often taken place in conflict with existing ones, since it
usually has involved staking out claims to tasks that already belonged to one or
more of the established groups. The archetypal conflict here has been (and still
is) the conflict between doctors and nurses, as nurses over the last 100 years
have fought fiercely and with great perseverance to improve their standing and
their education and to take over a significant part of the work that was earlier
the domain of medical doctors.
Hospitals are also characterized by a single-minded concentration on formal
qualifications when evaluating people for new positions, even within the
professions. To cross the border between two professions is impossible
altogether—regardless of actual knowledge and experience—without going
through the full educational program of the new profession. If the certificate is
missing, the door is totally blocked. And, as hinted earlier, the educational
programs of the different professions are totally separate, with no common
tracks or courses.
The occupational turf in a hospital is by now largely cut up and occupied by
the various professional groups, and the fight for larger domains or more
prestigious tasks increasingly amounts to a zero-sum game. The hospital
organization is therefore very rigid and extremely difficult to change. In such
organizations one would expect game theory to apply in many instances, and it
is interesting to see that alliances and conflict lines among the professions
indeed seem to comply. There is, for instance, a conflict between nurses and
nursing assistants, since the latter dislike to be supervised by nurses and
moreover want to move up toward nursing status and take over some of the
nurses’ responsibilities (and positions). The nurses, on their hand, have been
nibbling away at the doctors’ domain for a century, and these two groups still
have their skirmishes—not least in the area of administrative duties and
responsibilities in the hospital wards. What is more natural, then, than mutual
sympathy and goodwill between doctors and nursing assistants? Neither group
threatens the other, and the nursing assistants have no trouble accepting the
professional authority of the doctors. In fact, many doctors will claim that they
really do not need the (now) university-educated nurses, that they would prefer
to recruit
only nursing assistants (who receive more limited education) and then teach
them what more they need to know themselves. So far, however, the nurses
have had the most success.
This main pull of the Professional Bureaucracy—the pull to
professionalize—will receive no particular support from information
technology. Actually, the Professional Bureaucracy is probably the
configuration where information technology provides the most limited platform
for change. The reason is simply that the gist of the work in such organizations
consists of professional judgment, which typically requires the kind of “thick”
or “soft” knowledge that is (at least currently) impossible to impart to computer-
based systems. A number of expert systems that aid in tasks such as fault
finding and medical diagnosing have indeed been developed, but they cannot—
and are not intended to—replace the professionals. Rather, they are meant to
function as tools for the professionals, speeding up assessment and improving
the quality of their work.
Some Ruffled Feathers

There are of course exceptions to this general pattern. First, a growth in self-
service may become a threat to some professional groups. For instance, an
increasing number of brokerages now offer customers direct access to their
stock-trading systems, enabling customers to conclude deals directly from their
own PCs. Although the brokerages still perform the back-office functions, this
development may reduce the need for stockbrokers quite significantly and
diminish their position compared to the more routine-processing back-office
staff.
Second, not all Professional Bureaucracies are in the service sector (even if
the largest and most visible ones are). The configuration can also be found in
manufacturing in the form of craft enterprises (Mintzberg 1979), which is the
configuration favored by Piore and Sabel. They will be vulnerable to the
development of hyperautomation and flexible production and the pressure for
change implied by this development.
Traditionally, the craft enterprise depends on craftsmen who use relatively
simple tools to produce standardized goods. Because their tools are simple and
often general, it is relatively easy for them to shift their output to new products
if the markets change. This flexibility makes up for some of their lack of
productivity compared to the Machine Bureaucracies. However, in their
competition both with Machine Bureaucracies and with other craft enterprises,
they increasingly have to invest in more powerful tools and even computerized
equipment. This tendency is described by Piore and Sabel in the case of the
textile industry in Italy’s Prato district and the Japanese metalworking industry.
To Piore and Sabel, this is a proof of the vitality and adaptability of the craft
enterprise. To me, it is instead a development that will serve to reduce the
importance of the independent craft enterprise in manufacturing, since the
increasing use of computer-based equipment will tend to level out the field
between the different configurations: The necessary investments per employee
in the craft enterprise will approach those of the competing Machine
Bureaucracies (gradually becoming Flexible Bureaucracies), and the flexibility
of the Machine Bureaucracies’ production machinery will approach that of the
craft enterprise’s.
This convergence in technologies will also result in a convergence of the
required skills. Even if both the craft enterprise and the Machine Bureaucracy
have traditionally relied on action-centered skills (Zuboff 1988), they have been
very different in kind. The craft enterprise has employed highly skilled
craftsmen who perform a broad range of qualified work; the Machine
Bureaucracy has employed largely low-skilled personnel who have received the
limited, specialized training they need in-house or even on the job. However, as
the use of computer-based equipment increases, both types of organizations
increasingly need operators with the sophisticated, intellective skills required to
master the new equipment. Simultaneously, the degree of automation will tend
to rise, and computer-based systems will be employed to automate the
coordination of larger and larger parts of the production process, even spanning
organization borders, as in the Nissan Sunderland example. Craft enterprises, in
my opinion, will therefore experience a strong pull toward the Flexible
Bureaucracy configuration, or toward the Administrative Adhocracy if they are
able to automate their production completely. In some instances, information
technology even renders whole crafts superfluous, for instance, traditional
typography (as printing once eliminated the need for scribes).
We can expect such trends to continue, and Professional Bureaucracies
whose professional work can largely be automated, eliminated, or routinized
will also develop toward other configurations, such as a Machine Bureaucracy,
a Flexible Bureaucracy, or an Administrative Adhocracy—depending on the
nature of the changes they go through.
In the majority of the Professional Bureaucracies, however, the dominating
tasks fall into one or more of the three categories least susceptible to automation
(listed on p. 282): work where qualities such as creativity, judgment, artistic
skill, and emotional content are central. The professions are therefore likely to
continue their dominance in these organizations, and there is little reason to
believe that they will change configuration or that the configuration itself will
be significantly modified. As Mintzberg says (1979, p. 367),
The professional operators of this structural configuration require considerable

discretion in their work. It is they who serve the clients, usually directly and
personally. So the technical system cannot be highly regulating, certainly not
highly automated.
The professionals will therefore prefer systems that assist them in their work
and enhance their professional capacities. However, for the same reasons, they
will also take a favorable view of systems that automate or eliminate the most
routine aspects of their work, and especially systems that reduce the need for an
administrative staff. The potential of information technology in the case of the
typical Professional Bureaucracies is therefore a development toward somewhat
slimmer organizations, with a higher proportion of professionals than before,
and with sophisticated support systems both for professional and administrative
needs. There will be no dramatic changes justifying the proclamation of new
configurations, not even for the model-driven version of the Professional
Bureaucracy, which will depend mainly on the assisting model.
A New Line of Conflict

The technology does, however, hold a potential for increasing the antagonism
between professionals and administrative staff. The latter, which is usually
configured and run as a Machine Bureaucracy, tends to hold the view that
professionals are a bit on the whimsical side, and that a more “structured”
approach to work and better cooperation would improve both their productivity
and the quality of their work. They (and even the organization’s owners) may
easily see information technology as a means to improvement through injecting
more control and standardization into the professional sphere, and maybe even
automate or eliminate some of their tasks. In such a process the administrative
staff would also improve their own position in the organization. This is of
course anathema to the professionals, as Mintzberg points out in the
continuation of the preceding quote:
As Heydbrand and Noell (1973) 3 point out, the professional resist the
rationalization of his skills—their division into simply executed steps—because
that makes them programmable by the technostructure, destroys his basis of
autonomy, and drives the structure to the machine bureaucratic form.
I believe this will be an important source of conflict in Professional

Bureaucracies in the years to come. The conflict will of course become
3Heydbrand, W. V. and Noell, J. J., “Task Structure and Innovation in Professional Organizations,”
in W. V. Heydbrand (ed.), Comparative Organizations, Prentice-Hall, 1973, pp. 294–322.
most severe in organizations where significant parts of the professional work

can be automated or eliminated, and where the organization may even be posed
for a change of configuration along the lines indicated earlier. Many
Professional Bureaucracies also have strong “manufacturing” aspects, for
instance, hospitals—where the flow of patients through the wards and the
throughput in terms of the number of patients treated can be likened to the flow
of materials and output of finished goods in a factory. Such resemblances—real
or apparent—can provide platforms for attacking the traditional autonomy of
the professions, and the availability of sophisticated, regulating computer-based
systems for a variety of administrative and production-oriented purposes can
only strengthen those platforms.
The position of the computer professionals themselves will also be
interesting: Will they establish themselves as the kernel of a new
technostructure, allied with the administrative staff, or will they seek acceptance
as a new professional group? So far, the first alternative has been most common,
something that can be explained both by history (computers were usually first
brought in by the accountants in the administrative staff) and by the computer
professionals’ aptitude for logic as well as systems and efficiency engineering.
Because their work always tends to encroach on the autonomy of the other
professionals, it is also quite likely that they will have problems being accepted
as a bona fide professional group separate from the administrative staff.
Reintegrating The Divisionalized Form

The Divisionalized Form is not a configuration in the same sense as the others;
it is in a way a second-order form, a structure for the coordination of relatively
independent organizations—organizations that have their own structural
configurations and that could well exist as independent entities. As Mintzberg
says (1979, p. 381),
The Divisionalized Form differs from the other four structural configurations in
one important respect. It is not a complete structure from the strategic apex to the
operating core, but rather a structure suprimposed on others. That is, each division
has its own structure. As we shall see, however, divisionalization has an effect on
that choice—the divisions are drawn toward the Machine Bureaucracy
configuration. But the Divisionalized Form configuration itself focuses on the
structural relationship between the headquarters and the divisions, in effect,
between the strategic apex and the top of the middle line.
As noted in Chapter 7, the Divisionalized Form of the modern era was

pioneered by Edgar J. Thomson of the Pennsylvania Railroad and Pierre
du Pont and Alfred Sloan, Jr. of GM. It is primarily a configuration for

organizations that are too large or too diverse to be managed as centralized
structures organized in single tiers of functional departments. Most often, it is
an answer to market diversification either through growth or through takeovers
(conglomeration). There are a number of intermediate forms between the
Machine Bureaucracy and the Divisionalized Form, however; Mintzberg (1979)
counts four subtypes.
The Integrated Form is characterized by purely functional divisions, each
performing a step in the corporation’s total value chain, and it is only a small
step away from the departmentalized, monolithic organization. The divisions are
locked into a common planning system and generally lack the freedom to buy
from or sell to other than their sister divisions. If such an organization starts to
seek wider markets by diversifying production in its divisions and allowing
them to sell some of their output directly to outside customers, it changes to the
By-product Form. Central planning is still pervasive and the needs of the sister
divisions dominate, but some more freedom is introduced. If diversification and
growth in the by-product sector continue, the open market may at some point in
time become more important to the divisions than the corporation’s internal
market, and the organization moves on to the Related Product Form. At that
point, the demands from customers in the open market become more important
to the divisions than their internal, corporate commitments, and they require a
much more substantive independence. The end of the line is the Conglomerate
Form—the pure version of the Divisionalized Form—where the divisions are
fully independent and often totally unrelated.
In this pure form the corporate headquarters will be small—Mintzberg
mentions the case of Textron, where a staff of 30 executives and administrators
oversaw 30 divisions with combined sales of more than $1.5 billion (of late
1960s denomination). Corporate management will concentrate on monitoring
the divisions’ financial performance and on issuing policy for long-range
planning. Typically, there will also be a small legal department, and in many
instances an industrial relations office.
The process can also run in the other direction—from a conglomeration of
unrelated companies to a tight-knit, divisionalized corporation. When du Pont
took over control of GM from William C. Durant, even the monitoring and
planning functions were not established, and it was at the very most a holding
company—Durant managed it with the help of two to three assistants and their
secretaries (Chandler 1977). Du Pont and Sloan quickly established stringent
reporting and planning procedures and allocated the divisions to separate market
brackets. Over the years, control was gradually tightened, and more and more
common functions
were instituted. Mintzberg (1979, p. 405) cites Wrigley 4 to the effect that in the
mature GM, “The central office controls labor relations, market forecasting,
research, engineering, quality control, styling, pricing, production scheduling,
inventory levels, product range, and dealer relations; it decides what plants are
to be built, and what cars; it styles them, and it tests them on the corporate
Proving Ground.” There is also extensive use of standardized parts across
divisions. Mintzberg concludes that the modern GM has moved almost all the
way toward a Machine Bureaucracy, and is best characterized as exhibiting the
Integrated Form variant of the Divisionalized Form.
Differentiated Centralization
This illustrates the two pulls acting on the Divisionalized Form—one
decentralizing and one centralizing. The pull underlined by Mintzberg in his
total model of the main five configurations is the pull of the middle line
(division management) to balkanize—to decentralize, increase the divisions’
freedom of action, escape too detailed central planning, and reduce their
dependence on the sister divisions. On the other side, the Divisionalized Form is
a configuration created not for the love of decentralization as such, but as a
remedy for the mounting control problems experienced as organizations
(especially Machine Bureaucracies) become very large. The strategic apex in a
Divisionalized Form, then, delegating authority only by necessity, will almost
always be on the lookout for ways to achieve stronger central control.
Corporate management's objectives will differ, however, according to the
nature of the enterprise. The main distinction here is between companies with
totally unrelated divisions (true conglomerates) and companies where the
divisions have related products or markets, and where synergies or increased
economies of scale are conceivable (which was the case in GM). In true
conglomerates the strategic apex will concentrate on monitoring results and on
managing corporate finance; in companies with possible synergies, its ambitions
will also cover cross-divisional planning and coordination.
Information technology will also offer new possibilities for Divisionalized
Forms, and not only on the corporate level. The divisional level—the changes
possible within the divisions themselves—is also very important. However, they
are more or less equal to the possibilities for individual organizations discussed
under the other headings in this chapter and therefore do not need separate
treatment here.
4Wrigley, L.: Diversification and Divisional Autonomy, D.B.A. thesis, Harvard Business School,
1970.
Consequently, we can concentrate on the corporate level, where the main

dimension for change is the centralization/decentralization dimension outlined
earlier, and where the main differences in attitudes are dictated by the level of
affinity between divisions.
In true conglomerates, companies where there are no potential synergies
between divisions, the incentive for cross-divisional, central planning is of
course limited. However, corporate headquarters may well partake in the
planning process in the separate divisions, or at least use their plans to monitor
their progress and trigger corrective action at an early stage in the case of
underperformance. The supervision may be purely financial or may cover a
broader range of indicators. System-supported supervision offers ample
opportunity for strengthening and refining such monitoring of financial and
other quantitative information. With the proper systems in place for conducting
day-to-day business in the divisions, monitoring may even take place in real
time or almost real time, as in the reporting systems of Benetton and Hennes &
Mauritz. This will provide corporate headquarters with several options for
development: closer control and participation in the daily affairs of the
divisions, management of more divisions with the same staffing, and a reduction
of headquarters staff while maintaining or even improving control.
In addition to improved monitoring, computer-based systems will also make
it possible to elaborate the financial integration of a Divisionalized Form
considerably, regardless of subtype. With sophisticated systems, corporations
may run what amounts to internal banking systems, where internal transactions
are netted (also across borders), liquid reserves are pooled, and internal loans
and deposits are made. Many large companies, such as the Norwegian
conglomerate Norsk Hydro, have had such systems for years already. At Hydro,
which is regarded as one of the world leaders in this field, all transactions by
divisions and their subsidiaries are made toward central, internal accounts
managed by corporate finance—no division or subsidiary ever sends money
directly to another or to major suppliers, not even across borders. If a subsidiary
in Norway needs to pay a supplier in the United States, payment is made in
Norwegian kroner to corporate finance in Norway, who will then pay the
supplier in dollars from its accounts in the United States—accounts that in turn
receive payments from Hydro’s American operations for goods and services
purchased elsewhere in the world. All external loans, deposits, and currency
transactions are made by corporate finance. In this way the number and size of
external currency transactions and ordinary bank transactions are minimized.
Hydro even operates a bank for its employees in Norway—complete with
automatic teller machines installed in its offices and subsidiaries around the
country, where employees can withdraw money with their
Hydro cards. The bank accepts deposits and gives loans to employees, always at
better terms than those offered by ordinary Norwegian banks.
The advantages discussed so far also apply to the other subforms of the
Divisionalized Form—those where there are more or less clear synergies or
economies of scale to be realized by coordination across divisions. However, in
such organizations, the strong coordinating powers of computer-based systems
can also be brought to bear. In addition to system-supported supervision, both
implicit coordination, programmed routines, and hyperautomation can be
applied to overcome the information and control overload that earlier prohibited
unified coordination of the divisions. I believe that this will allow reintegration
of operations in a large number of instances, reducing the number of divisions
or even transforming Divisionalized Forms to clean-cut Machine Bureaucracies
or Flexible Bureaucracies. In so doing, they may also cross the threshold to
become model-driven organizations, depending mainly on the regulating model.
We have seen some developments lately that point in this direction—a
growing number of companies have taken advantage of computer-based just-in-
time systems to eliminate regional warehouses and coordinate distribution
nationally or even internationally. HÅG, for instance, a Norwegian producer of
desk chairs, delivers made-to-order chairs from its manufacturing plant in the
mountain town of Røros directly to customers over most of Europe within five
days of receiving the order. 5 This feat is made possible by a sophisticated just-
in-time production management system and cooperation with a forwarding
agent who runs operations with the help of a computer-based distribution
system. The condition is of course that the order stays within the range of
upholstery in stock. Dell, the American PC maker, runs a similar operation from
its plant in Ireland.
Multinational companies are increasingly lumping national markets together
in larger geographical regions and have restructured both manufacturing and
distribution along the new boundaries, supporting them with sophisticated
logistics systems. In Scandinavia we have seen quite a number of such moves
now, as companies have organized their Scandinavian operations under one
umbrella, establishing a joint headquarters in one of the capitals, often
supplying the whole region directly from a single facility.
As our experience grows and systems mature, it should be possible to achieve
a much higher integration than we experience today. The result could be larger,
faster, and more nimble multinationals, which means increased competition for
businesses who believe they are local and
5Personal communication with the project manager for the JIT implementation.
have advantages because of their small size. Still, it does not seem that the
extensions of the space of constructible organizations will contain variants of
the Divisionalized Form that amount to new configurations, neither for the
conglomerate variant nor for the more tightly knit firm—at least as long as we
maintain the condition that all the elements of the organization shall have the
same owners or at least answer to the same corporate management. When it
comes to the coordination of totally separate companies, however, we approach
something new: the Meta-Organization, which will be discussed later in this
chapter.
Transforming Adhocracy
Mintzberg views the Adhocracy as the youngest of his five basic configurations.
As a configuration for larger, formal organizations, this is probably correct,
even if it also represents one of the two primal coordination mechanisms.
However, I suspect that closer study would find that variations of it have been in
use for centuries and even millennia, especially in teams of craftsmen
constructing buildings, ships, or other large objects.
The Adhocracy comes into its own when the environment is so dynamic that
it is difficult to standardize products and perpetual innovation is necessary; and
the innovative work is so complex that it requires the efforts of many experts or
expert groups. Adhocracies must therefore bridge specialization in a much more
dramatic way than Professional Bureaucracies, where experts cooperate by
enacting their establish professional roles and adhering to their own group’s
particular standards. In Adhocracies the experts have to give and take, to
pioneer new approaches that may break with established procedures, and to
arrive at joint solutions incorporating elements from them all. Because experts
are so central to the innovations that Adhocracies live by, they must also hold
wide power—at least in in practice if not in formal designation. Describing the
design parameters of the Adhocracy, Mintzberg says (1979, pp. 432–3, bold
type in original):
In Adhocracy, we have a fifth distinct structural configuration: highly organic

structure, with little formalization of behavior; high horizontal job specialization
based on formal training; a tendency to group the specialists in functional units for
housekeeping purposes but to deploy them in small market-based project teams to
do their work; a reliance on the liaison devices to encourage mutual adjustment—
the key coordinating mechanism—within and between these teams; and selective
decentralization to and within these teams, which are located at various places in
the organization and involve various mixtures of line managers and staff operating
experts.
To innovate means to break away from established patterns. So the
innovative organization cannot rely on any form of standardization for

coordination. In other words, it must avoid all the trappings of bureaucratic
structure, notably sharp divisions of labor, extensive unit differentiation, highly
formalized behaviors, and an emphasis on planning and control systems.”
Adhocracy comes in many varieties, since organizations may border on other

configurations or have to meet special conditions. Mintzberg mentions at least
seven variants, with the Operating Adhocracy and the Administrative
Adhocracy as the most important ones. (He does not say it explicitly, but they
seem to represent two main classes, whereas the other five are subtypes). The
Operating Adhocracy is the classic form, where the teamwork is undertaken to
serve the customers’ needs directly, and where the operating core and the
administrative staff constantly mix and merge in project teams. In the
Administrative Adhocracy, however, the operating core is cut off from the
administrative part of the organization because it needs another kind of structure
since it is automated (most often it will be a Machine Bureaucracy) or even
done away with completely and contracted out to other organizations. The rest
of the organization, structuring itself as an Adhocracy, can then concentrate on
the innovative part of the work, leaving the isolated operating core to crank out
the products. Typical Administrative Adhocracies include newspapers, where
the editorial staff faces the awesome task of creating a new paper every day
(different down to the last letter), while the printing plant and the distributing
organization—always physically separated from the editorial offices, often even
organized as separate companies—can concentrate on streamlining their
repetitive duty of providing the subscribers with (from their point of view) the
same wad of printed paper every day. As we have already concluded,
information technology (especially by enabling much more extensive
automation) will make it possible to structure more organizations as
Administrative Adhocracies in the future.
The other subforms defined by Mintzberg, such as the Entrepreneurial
Adhocracy (a hybrid of Adhocracy and Simple Structure) and the
Divisionalized Adhocracy (a cross with the Divisionalized Form) will differ
significantly in the extent to which they benefit from computer-based systems.
The Entrepreneurial Adhocracy, which is really an Operative Adhocracy
(usually a high-tech start-up) with an owner/manager who is also an outstanding
professional (and recognized as such), will not benefit any more than small
Operative Adhocracies in general. The Divisionalized Adhocracy, however,
stands to gain more. As Mintzberg defines it, it is essentially a Divisionalized
Form with an environment so complex that simple divisionalization does not
suffice—it has to implement a matrix structure. A true matrix organization does
away with the unity of command that is the hallmark of Machine
Bureaucracies and conventional Divisionalized Forms, and requires close, team-

oriented cooperation between the two (or even three) dimensions found in the
matrix.
It is indeed conceivable that the strong coordinating powers of computer-
based systems—which can allow a Divisionalized Form to change into a
Machine Bureaucracy or a Flexible Bureaucracy—can help a Divisionalized
Adhocracy to collapse one of its dimensions. For instance, in a product/market
matrix (the most common one), the coordination of production and shipment of
goods may be streamlined to such an extent that the organization may be able to
collapse its matrix to essentially a market-based Divisionalized Form served by
a common product division. As noted earlier, we have indeed seen tendencies in
this direction lately, as a number of multinational companies have considerably
enlarged the geographical regions served by one organizational unit.
The Communication Bottleneck

Regardless of which subtype of Adhocracy we study, however, the central
problem is the copious communication needed to coordinate through mutual
adjustment, the coordinating mechanism required for the kind of tasks that
Adhocracies are designed to tackle. To quote Mintzberg once more (1979, p.
463),
People talk a lot in these structures; that is how they combine their knowledge to
develop new ideas. But that takes time, a great deal. Faced with the needs to make
a decision in the Machine Bureaucracy, someone up above gives an order and that
is that. Not so in the Adhocracy. Everyone gets into the act. First are all the
managers who must be consulted—functional managers, project managers, liaison
managers. Then are all the specialists who believe their point of view should be
represented in the decision. A meeting is called, probably to schedule another
meeting, eventually to decide who should participate in the decision. Then those
people settle down to the decision process. The problem is defined and redefined,
ideas for its solution are generated and debated, alliances build and fall around
different solutions, and eventually everyone settles down to a hard bargaining
about the favored one. Finally, a decision emerges—that in itself is an
accomplishment—although it is typically late and will probably be modified later.
All of this is the cost of having to find a creative solution to a complex, ill-
structured problem.
Although this is all necessary to solve one-of-a-kind problems, it is devastating

for any attempt to compete in the field of routinized work. Adhocracies are ill
equipped to handle ordinary, routine tasks, and, if they want to move in such a
direction, they must transform their structures—for instance, to Professional
Bureaucracies (for consulting based on standard methods and a repertoire of
tested solutions) or Machine Bureaucracies (for volume production of goods).
Such transformations
are, by the way, seldom made without conflict and the defection of a number of
experts. Those who prefer innovative work and adhocratic organization will
fight fiercely against the changes, and, if they lose, a number of them will
probably jump ship to join other companies more in tune with their preferences.
They may even choose to set up a company of their own. This is in accord with
the main pull on an Adhocracy, which, according to Mintzberg (1979), is the
pull of the support staff to collaborate—which translates into the experts’
insistence on organizing work in projects and participating in decision making
on all levels.
How, then, if at all, can information technology help to ease the exceptionally
heavy burden of communication that is the core problem of the Adhocracy?
Saturated as they are with communication, and with a consistent pull toward
cooperation, we should expect these organizations first of all to benefit from
systems supporting communication and teamwork—that is, they should benefit
more than other configurations from the use of groupware tools of all
denominations. I do indeed believe they will be able to use such tools profitably,
but success will not be ensured—they will need fairly firm coaching in order to
use such tools for increased productivity rather than simply for increasing their
volume of communication and probing their (always interesting) subject matters
to even greater depths.
However, as concluded earlier, groupware (e.g., coauthoring systems,
systems for meeting assistance, electronic mail and group calendars) is not
going to revolutionize any type of organization—not even Adhocracies. The
reason is that these types of systems really only support and facilitate the
various kinds of interpersonal communication that constitute the traditional
means for mutual adjustment. They do very little to reduce the required
communication volume; on the contrary, by offering improved channels they
tend to increase the total amount of communication instead.
Groupware will therefore only allow Adhocracies to become a bit more
efficient (provided the necessary coaching), to produce work of somewhat
higher quality, and to function more independently of physical proximity.
Groupware products will not have the power needed to allow structures very
different from today’s Adhocracies, since they simply do not tap into the most
powerful aspects of computers.
Ascendance of the Interactive Adhocracy

The strong points of information technology, however, offer other
possibilities—perhaps not so obvious, but much more interesting. The attack
point is, even here, the volume of communication needed for coordination.
However, the thrust is not in the direction of better tools for this
communication, but toward reducing the need for such interpersonal

communication in the first place.
How can that be accomplished? The answer lies in the mediating model and
the power of implicit coordination. If an Adhocracy can model its problem
domain to sufficient depth and with sufficient rigorousness, it can also build
systems that will shift the larger part of the coordination burden from explicit
interpersonal communication (which has to be carried out in addition to the
actual work itself) to a much more terse and efficient kind of communication,
directed toward the system (or complex of systems) and effected as an implicit
part of the actual work.
An early example is, as we noted, the CAD system used for the structural
design of Boeing 777. Before the introduction of the new system, all
coordination in the project organization had to rely on direct interpersonal
communication, meetings, and circulation of drawings—all burdensome efforts
that came on top of the actual design work. With the CAD system in place, the
need for much of that communication was simply eliminated, since the required
information could be presented through the system to anyone at any time.
Additionally, the information itself was created and fed into the system as an
integral part of the work process and required little or no separate effort.
Unlike the implementation of groupware, then, the introduction of
comprehensive systems based on a mediating model will allow quite dramatic
changes—even large organizations may achieve real-time or close-to-real-time
mutual adjustment, in some aspects comparable to the kind of coordination
achieved in small groups. Adhocracies based on mediating models should
therefore become much more efficient, since they will spend far less time and
effort on coordination than they used to. They should also be able to react and
adapt significantly faster to changes in their problem domains. This variant of
the Adhocracy is definitely different enough from its traditional forebear to
justify a new name: it could be called the Interactive Adhocracy, to connote the
way it depends on pervasive, real-time, interactive systems to sustain the
dynamic mutual adjustment that is the defining feature of adhocracies.
The Interactive Adhocracy will retain almost all the main design parameters
of the Adhocracy, such as organic structure, selective decentralization,
horizontal job specialization, high percentage of professionals/experts, and a
concurrent use of functional and market groupings. The liaison devices of the
Adhocracy will, however, be largely supplanted by the mediating model.
While keeping and even strengthening the superior problem-solving
capability of the traditional Adhocracy, the Interactive Adhocracy should be
able to come substantially closer to the other configurations in efficiency, and it
should therefore emerge as a viable alternative in a
much larger number of cases. To bring the creative power of this kind of
organization to bear on problem domains that have, as yet, not been able to
sustain the costs of an Adhocracy should be a very exciting prospect in a world
where most markets experience increasingly rapid changes and a mounting
pressure for innovation.
There are two important limiting factors for the construction of Interactive
Adhocracies. One is the extent to which the problem domain can be modeled
with sufficient rigorousness to be implemented in a suite of computer-based
systems. As we travel along the learning curve, however, and our tools and
methods improve, we will be able to do so for an increasing number of
organizations. The other major factor is the need for synchronization of goals
and objectives. The “anarchistic” nature of Adhocracies means that their
members must internalize their organization’s goals and objectives to a larger
degree than necessary in more hierarchic configurations. As Khandwalla 6 says
about the Adhocracy (quoted in Mintzberg 1979, p. 435),
The job of coordination is not left to a few charged with responsibility, but
assumed by most individuals in the organization, much in the way members of a
well-knit hockey or cricket team all work spontaneously to keep its activities
focused to the goal of winning.”
This will be even more true in an Interactive Adhocracy, where the volume of
interpersonal communication will be significantly lower than in a conventional
Adhocracy. If model-mediated mutual adjustment is to be used on a large scale,
sustained attention to the maintenance of a comprehensive team spirit and
loyalty toward common goals will be required. Without a common
understanding of the organization’s goals and objectives and of their own role in
the total picture, the members of the organization will end up pulling in opposite
directions, and will compromise the viability of the model.
New: The Meta-Organization

So far, we have only looked at the evolution of Mintzberg’s original five
configurations. However, information technology may also permit the
construction of totally new configurations. The discussions in Chapters 12
through 14 have hinted at two: one emerging from the strong coupling of
independent organizations, the other emerging from the very fringes of
organization. I will call them, respectively, the Meta-Organization and the
6Khandwalla, P. N.: “Organizational Design for Change,” Learning Systems, Conceptual Reading, 5
(New Delhi, India, 1976).
Organized Cloud. Table 16-1 (p. 402/3), lists the main characteristics of these
two and the three modified configurations discussed earlier.
As we concluded in Chapters 12 and 14, the strong, detailed, and extensive
coordination that can be achieved through the use of unified computer-based
systems makes it possible to achieve a new kind of integration between separate
organizations. Organizations such as those comprising the manufacturing
system centered on the Nissan factory in Sunderland are indeed coordinated
more closely than sister departments within most single organizations. Whether
or not the organizations involved have separate owners, their operations are so
intertwined and they depend so critically on each other for their daily operations
that it seems very reasonable to view them as a single organizational entity.
However, the fact that they have separate owners, separate economies, and
separate chains of command and are joined only in a contractual arrangement
makes it difficult to classify such clusters as Machine Bureaucracies or
Divisionalized Forms—even if they resemble these configurations in many
ways. There is also a continuum of such arrangements—from the very long-
term, inclusive, and tight arrangement of the Nissan’s Sunderland operations, to
more temporary hookups such as the ones that may be established in the
construction business to bid on a specific contract.
There are various terms in use for such arrangements, most often network
organization, networked organizations, or virtual organizations. As they are
used today, they are given quite varied interpretations, ranging from the one we
are discussing here (a close cooperation between separate organizations) to
single organizations where a large part of the members rely on information
technology to work away from the organization’s premises. The last
phenomenon does not necessarily involve any new organizational developments
at all. As Mintzberg points out (1979), physical distribution of services (or
people) does not necessarily involve any decentralization of power, especially
not when the dispersal is facilitated by much improved communication
equipment—which is precisely what makes it possible to keep the normal
chains of command regardless of distance. Instead, I will apply the term Meta-
Organization to entities consisting of two or more closely coupled
organizations. This term serves both to indicate the layered nature of such
organizational constructs and to avoid the misleading connotations that can be
attached to the other terms.
Particularly, I find the term “virtual organization” superficial and
misleading—if we extend to it the connotation of “virtual” in other computer-
related terms, such as “virtual memory,” “virtual disks,” or even “virtual
reality,” (a perfectly postmodern oxymoron), a virtual organization should be a
simulated organization—the kind you play with in
Configuration Main Main Design Main Contingency

Coordinating Parameters Factors
Mechanism
Joystick Regulating model, Centralization and organic Small to medium size,
Organization emphasizing system- structure. Little nonsophisticated technical
Early examples: supported supervision specialization, except system or hyperautomated
Possibly some and programmed sophisticated operating core, simple,
centralized routines. technostructure; little dynamic environment
franchising formalization. (possibly hostile), or
operations. strong power needs of top
manager.
Flexible Regulating model, Behavior formalization, Medium to large,

Bureaucracy emphasizing vertical and horizontal job regulating,
Early examples: hyperautomation, specialization, usually hyperautomated technical
Possibly some programmed routines, market grouping or matrix system, environment that
advanced JIT and (depending on structure, large operating is simple to moderately
producers. the problem domain) unit size, vertical complex and stable to
implicit coordination. centralization and limited moderately dynamic,
horizontal decentralization, technocratic control.
action planning.
Interactive Mediating model, Organic structure, selective Often fairly small, but can
Adhocracy emphasizing implicit decentralization, horizontal become large if problem
Early examples: coordination. Project job specialization, training domain is well suited for
Possibly some oriented. Experts (large percentage modeling. Complex and
design projects have much informal professionals/experts). dynamic environment,
using advanced power. Functional and market sophisticated and often
CAD systems. grouping concurrently. automated technical
system.
Meta-Organization Regulating model, Strongly formalized Medium to large,

Early examples: emphasizing cooperation between a regulating,
Possibly supplier hyperautomation, number of independent hyperautomated technical
clusters such as the programmed routines organizations. Vertical system, environment that
most advanced ones and (depending on product specialization and is simple to moderately
in the automobile the problem domain) functional specialization complex and stable to
industry. implicit coordination. among organizations, moderately dynamic,
vertical centralization and technocratic control.
limited horizontal
decentralization, action
planning.
Organized Cloud Mediating model, Nonmanaged, self- Small to very large,
Early examples: emphasizing implicit regulating, except for simple but extremely
Reservation coordination. operation of providing dynamic environment,
systems, trading system. Sophisticated sophisticated and fully
systems. technostructure. automated technical
system.
Table 16-1: Main characteristics of the three modified and two new struc-
tural configurations. Format adapted from Mintzberg 1979.
(This page is intentionally left blank to maintain the pagination of the printed
version, where table 16-1 occupied two pages.)
computer-based management games, an imaginary organization that does not

affect reality at all. Only real organizations can act in the real world.
I also prefer to avoid the term “network,” which has a distinctly egalitarian
connotation that does not fit many of the actual examples. Whatever the
relations are between the participants in the Nissan setup, for instance, they are
surely not equal. Also, the term “network” seem to imply that one can fairly
easily connect and disconnect to the structure; if there is one thing that is true
about setups such as the one in Sunderland, it is that it takes copious amounts of
work and long-time commitments to establish it, and, once established, it is very
expensive to change both setup and participants.
A Meta-Organization may of course also consist of equal partners, with no
single partner occupying a dominating role. The present terms are used with a
considerable degree of looseness, however—even fairly simple cooperation
endeavors such as common marketing efforts, a number of common projects, or
the use of email for coordination seem to arouse the enthusiasm of the IT
community and earn the participants a pioneer status, even if it is not at all
different from what was earlier achieved by traditional means. In contrast, to
call a construct a Meta-Organization, the activities of the different participants
should be directly and unequivocally coordinated through a common systems
infrastructure, preferably with a high level of automation.
Constructs resembling Nissan’s in Sunderland may also exist within the
boundaries of a single organization, as when a company with a number of
manufacturing sites unites the coordination of their operations with a unified
production control and delivery system, providing the sales force with
something that looks and behaves like a single source. However, when such
arrangements are set up within a single organization, it will fall into other
categories that have already been discussed earlier in this chapter—notably in
the sections on the Machine Bureaucracy and the Divisionalized Form.
In a Meta-Organization the participating organizations are closely bound
together by comprehensive systems and are usually member of only one or two
Meta-Organizations—at the very most a handful. In the case of multiple
memberships, the organization will frequently have a corresponding number of
different sites, with each physical site serving only one particular Meta-
Organization. The common systems will typically coordinate a large part or
even the total set of activities in the members' organization. The process of
setting up the Meta-Organization requires considerable efforts over extended
periods of time and is replete with planning and design in painstaking detail.
Once set up, is not easily dissolved, since the process of replacing a member is
almost as costly as setting the whole thing up in the first place. Members are
usually specialized in relation to each other and totally dependent on each other
for the combination to succeed. Because of the efforts involved, the number of
members in a Meta-Organization will typically be single-digit or double-digit,
but increasing standardization may facilitate larger Meta-Organizations in the
future.
The main purpose behind Meta-Organizations is to automate coordination of
processes across the member organizations. They will therefore depend on
regulating models, emphasizing hyperautomation and programmed routines.
Depending on the problem domain, they may also use implicit coordination.
The main design parameters will resemble the Flexible Bureaucracy’s: a strong
formalization of cooperation, vertical product specialization among members,
vertical centralization both inside the member organizations and in the
cooperative effort itself, and a preponderance of action planning. The Meta-
Organization’s technical system and preferred environment will be the same as
the Flexible Bureaucracy’s—indeed, the Meta-Organization’s goal will, to a
large degree, be to function (for production purposes) just like a Flexible
Bureaucracy.
In spite of the considerable efforts needed to establish them, I believe we will
see a growth in the number of Meta-Organizations in the future, and I believe
the most prominent form (at least for a long period of time) will be clusters of
suppliers built around dominating buyers, even though there will also be a
growth in cooperative efforts between more or less equal partners.
However, I am generally skeptical toward the very optimistic attitude many
commentators take on the prospects for imminent success of more temporary
arrangements of this kind (for example, in the construction business, which is
often subject of such discussions)). This optimism, in my view, vastly
underestimates the difficulties and efforts involved in going beyond the email
stage and setting up really close cooperation based on the use of common (or
communicating) systems in areas such as design and production control. Not
only is the level of systems standardization still far away from what is needed
for easy hookups; the organizations involved are almost guaranteed to use
different data formats and, even more important, to have different
understandings of important terms and categories. For instance, when Elkem 7
wanted to compare the performance of the different furnaces at its Norwegian
and American smelters, it turned out that the terms and parameters used to
measure performance were so different between the sites that it was simply
impossible to make a comparison. To obtain meaningful data, it would be
7The Norwegian metals company that is among the sponsors of the research behind this book.
necessary to carry out a full revision and standardization of the terms and
parameters used at the different sites—an effort so overwhelming, and sure to
meet with so much local resistance, that the project was shelved for an
indefinite period.
Supplier Clusters
The supplier cluster alternative, such as the Nissan example, is most easily
established—not because it is technically easier or requires less work, but
because a powerful buyer can demand cooperation from its subcontractors and
more or less guarantee their benefits. Such a configuration will also tend to be
more stable, since the leadership position will never be questioned. It is
interesting, however, to speculate on whether or not there will be any impetus
toward takeovers: Will the dominant buyer prefer to acquire the suppliers when
they are already functionally almost a part of its own organization?
Early in the automobile era, there was a strong movement toward such
vertical integration, with Ford’s legendary River Rouge plant as the pinnacle
(Beniger 1986). Ford’s ultimate ambition was to start with iron ore in one end of
the factory and roll out finished automobiles in the other, keeping up an
uninterrupted production flow throughout the complex. The task proved too
difficult, however, and the River Rouge plant was not competitive compared to
plants where materials, parts, and subassemblies were purchased from specialist
companies.
It may be argued that information technology now has made it more realistic
to tackle such complex coordination problems, as setups such as the Nissan
plant indeed indicates. However, a number of the old arguments are still valid:
Specialist companies will generally be more competent in their fields, since they
serve several customers and accumulate superior experience, and since they can
devote their full energy to a limited set of problems. The fact that they are not a
formal part of the buyer’s organization also means a reduced financial risk in
case of a reduced demand for cars. I therefore believe that the Meta-
Organization solution will be very stable in these circumstances, since (from the
point of view of the dominant buyer) it combines the advantages of competitive
know-how and reduced financial risks with a level of coordination fully
comparable to what could be achieved through ownership.
The Meta-Organization will also tend to be quite stable in terms of
membership, since a change of supplier can be very expensive for the suppliers
involved as well as the buyer, due to the high costs of systems development and
adaptation. In the Nissan Sunderland example, there is the added, stabilizing
requirement that the most important suppliers are physically located on the ring
road around Nissan’s main plant.
Equal Partners
Cooperation among equal partners seems to be more difficult to establish and
maintain, which should not come as a surprise. The costs involved in setting up
a Meta-Organization are high, and the benefits of cooperation can be difficult to
ascertain before hand. Moreover, a considerable level of mutual trust must be
present from the outset, since the implementation of common systems and
common procedures implies that the participating organizations will have to
reveal many of their internal functions, problems, and even company secrets to
each other. Such close cooperation will also easily put constraints on the
activities of the member organizations—it may, for instance, be considered
disloyal to do certain kinds of business with companies that are competing with
a partner.
A successful Meta-Organization consisting of equal partners will tend to be
somewhat unstable, since the partners will be fairly likely to develop different
ambitions for the evolution of the partnership. One or some of the companies
may well try to build a leading position at the expense of others; some
companies may want to proceed toward a merger; some will play brakemen;
and some may leave altogether for what they see as more exciting opportunities
elsewhere. In the event that the cooperation is successful and free of conflicts,
merger may well be a frequent outcome in the longer term.
New: The Organized Cloud

At the end of Chapter 13, I discussed briefly the organizing effects of
reservation systems. Even though we do not consider this far-flung mass of
people an organization (as it was defined in Chapter 3), the fact remains that all
these tens of thousands of travel agents and others—using, for instance, SABRE
or Amadeus—are perfectly coordinated in those aspects of their work that
pertain to reservations for the flights, hotels, and car rentals that are listed there.
In those aspects, they are even more strongly and efficiently coordinated than
most members of conventional organizations. As I remarked at the end of the
discussion, even if this arrangement is not an organization, it is certainly
organized—and I think it is unsatisfactory to dismiss this phenomenon as not
belonging to the realm of organizations just because it does not fit the traditional
definitions. I propose to call it an Organized Cloud. The cloud metaphor here is
derived more from astronomy than meteorology—picture the travel agents as a
cloud of stars, held together by the gravity of the their common database.
Organized Clouds are by and large products of the computer age, as they are
totally dependent on the powerful implicit coordination of the
database to exist—they represent perhaps the most completely model-driven

organizations we know today. Their databases represent complete mediating
models of their problem domains, and all interaction between the members
takes place via these models.
However, we may say that they have a humble ancestor in the traditional
marketplace, and somewhat more discernible progenitors in the stockmarkets
and commodity markets that developed after 1700 (the world’s first real stock
exchange was established in London in 1698)—especially since the late
nineteenth century, when traders started to use telegraphs and telephones.
However, it is the reach, capacity, speed, and the interactive nature of the
database that has made possible the formation of really significant Organized
Clouds, and it is the almost instant information dissemination and feedback
provided by the systems that breathes life into the clouds and turns them into
such powerful attractors.
Perhaps the most interesting clouds at the moment are the trading systems for
stocks, currency, and commodities, which have substantially changed the
behavior of the financial markets. Because of the almost instant conclusion of
deals and broadcast of pricing, the pace of the markets have increased
dramatically over the last decades, and in the course of the 1990s the
development of program trade (trade initiated by computers programmed to
react to certain price levels) increased the speed further. There is also a
discernible trend toward growth in the biggest clouds (the trading systems based
in London, New York, and Tokyo, the top three financial centers of the world),
and trading is increasingly done on a global basis—especially for currency. As
Yates and Benjamin (1991) point out, it will certainly be technically feasible to
organize global markets. I think we can expect to see the development of a
hierarchy of clouds on global, regional, and national bases, with the main focus
on global and regional clouds. Currencies and commodities in particular are
increasingly moving toward global integration, stocks probably toward a
regional emphasis with a number of premium stocks traded globally, whereas
small companies and local start-ups will continue to be traded mainly on a
national basis. We may also see a development of some sector clouds on a
global basis, with a single trading center (and thereby a single database)
contracting most of the business in one industry—for example, shipping or gold
mining.
The definition of an Organized Cloud is by no means clear. Clouds exhibit
some of the properties of proper organizations (as defined in Chapter 3). If we
start out with the two previous examples, it is fairly evident that cloud members
do not have a common goal in the sense that members of a normal organization
have (or should have!). Although cloud members are very interested in the
availability of clouds suitable to their purposes, and will willingly pay fees to
have access to
them, they do not have a common purpose relating to any specific cloud—they
do not look toward the interests of a cloud in the way any loyal organization
member would look toward the interests of his or her organization. However,
cloud members do have similar purposes. The travel agents all want to book
airplane seats and hotel rooms; the traders all want to trade. It is precisely these
similar interests that bring them into the cloud in the first place. In their
transactions as cloud members, however, they look only toward their own or
their clients’ narrow interests, and most of them will even be member of several
clouds simultaneously: travel agents will use several reservation systems,
stockbrokers may have access to several trading systems. They will quite
expediently use the one that offers them the best services and the most favorable
terms, just as when they shop for other products and services in the marketplace.
It would also be a little meaningless to say that clouds have a division of
labor—admittedly, the members all fend for themselves, but their activities are
not part of an overall effort to achieve a common purpose. Essentially, clouds
are non-managed: there is no central authority that can issue orders to the
members, aside from determining some basic rules. Clouds do, however, have
accepted mechanisms for reaching decisions (deals), and there is a rudimentary
power structure concentrated around the framework of rules for actions within
the cloud’s sphere of interest: There are supervisory bodies, rules about
membership, and mechanisms for expulsion or punishment in case of
misconduct. There is also a common memory, represented by the central
database, and there is definitely a communication structure. Finally, the
activities of the members are coordinated in the sense that the actions of one
member have impact on the actions of other members—within an accepted
problem domain and within an accepted set of rules.
In contrast with the Meta-Organization, the Organized Cloud typically
coordinates only one or a few of the members’ activities, and members are
typically members of several clouds. Membership is defined in terms of
subscription to a service or something similar (both organizations and single
individuals can be members), and access is usually simple, by means of a
defined (even standardized) interface. A new member can typically be up and
running in a matter of days or even hours. The Cloud is built on narrow,
standardized interaction; members all act alike in their transactions as Cloud
members; there is no specialization and no interdependence except for the
logged results of the (atomized) actions.
The Organized Cloud always depends totally on a mediating model,
emphasizing above all implicit coordination. Its main design parameters are that
it is essentially nonmanaged and self-regulating, except for the provision of the
system itself. It has a sophisticated technostructure run by
the providing organization—which is probably structured as a Machine

Bureaucracy or Flexible Bureaucracy. It can be small or very, very large—
membership in large clouds today number in the tens of thousands, some even
hundreds of thousands, and in the future clouds may consist of millions of
members. Typically, clouds exist in a simple but extremely dynamic
environment. The technical system is of course fully automated.
If we accept markets and exchanges as Organized Clouds, the configuration
has been around from time immemorial and has served very important purposes.
It may even be argued that a market-based economy as a whole can be viewed
as an Organized Cloud, although, on a societal level, it is inseparable from other
powerful organizational structures, both political and nonpolitical. Indeed,
clouds seem to be particularly well suited for market-like purposes where the
objective is to match buyers and sellers, takers and suppliers, within a
framework of open information about crucial parameters such as prices,
volume, and bookings. However, the much improved communication
infrastructures provided by information technology, coupled with the
unsurpassed coordinating powers of the database, provide an altogether new and
vastly more powerful basis for this kind of organization. As the prices for
systems and communication continue to fall, as more and more businesses and
private homes are equipped with computers and data communication links, and
as the Internet (or its eventual successor) provides standardized access and
payments, the formation of clouds will become viable for purposes with much
lower yields than airline reservations and transactions in the stock and currency
markets. The Organized Cloud—in its modern guise an organization fully
driven by a mediating model—could well emerge as one of the defining features
of future societies.
Relating Models and Configurations

In Mintzberg’s theory, each of the five coordinating mechanism gives rise to a
particular structural configuration. Even if all large organizations will have sub-
units and or pockets of deviating structures depending on different coordinating
mechanisms, there will normally be a dominating mechanism that will permeate
the organization and determine its overall structure.
In contrast with this one-to-one relationship between coordinating
mechanism and structural configuration, the analysis in this chapter has shown
that two of the three coordinating models proposed in Chapter 15—the
mediating model and the regulating model—can support more than one
configuration. However, models belonging to any of these two classes can vary
considerably in scope—that is, in how comprehensive
repertoirs of organizational actions they cover. When we include this dimension

the seeming indefiniteness is resolved, in that the comprehensive and restricted
versions give rise to different configurations. The relations between the different
kinds of active models and the new structural configurations are summed up in
Figure 16-1.
The Mediating Model is the basis for both the Interactive Adhocracy and the
Organized Cloud. Models supporting Interactive Adhocracies will tend to be
comprehensive, covering a broad set of activities, since they must support all or
the major part of the activities in a complete organization. Models supporting
Organizational Clouds, on the other hand, will tend to be quite restricted,
supporting only the narrow activity that is the business of the cloud.
The Regulating Model will probably be the most widely used, at least for
some time to come. It can support Joystick Organizations, Flexible
Bureaucracies, and Meta-Organizations, as well as aspects of Divisionalized
Forms. The Joystick Organization will tend to be simple and on the smaller side,
but the model will be comprehensive in the sense that it supports a large part of
the organization’s total activities. The Meta-Organization will tend to be large
and complex, but the model, although often quite complex in itself, will be
restricted to those parts of
Mediating Regulating Assisting

Model Model Model
Interactive Organized Professional

Adhocracy Cloud Bureaucracy
Comprehensive Restricted
Model Model
Joystick Flexible Meta-

Organization Bureaucracy Organization
Simple Complex
Comprehensive Restricted
Model Model
Divisionalized Form
Figure 16-1: Main model dependencies of the various configurations.

the activities that are involved in the relationship between the partners in the
Meta-Organization. The Flexible Bureaucracy will fall between these two—it
will usually be more complex than the Joystick Organization, but less so than
the Meta-Organization, and will usually employ models that are less
comprehensive than the Joystick Organization and more so than the Meta-
Organization. However, there will be great variation here—we may well see
large Flexible Bureaucracies that are more complex than most or all Meta-
Organizations, and there may also be Flexible Bureaucracies developing models
as comprehensive as any Joystick Organization’s.
Whereas the Meta-Organization is a configuration where certain aspects of
separate, independent organizations are very strongly coordinated, the
Divisionalized Form represents an arrangement to coordinate and control a
number of organizations that are either part of or owned by the same
corporation, and that are too complex taken together to be managed as one
intergrated organization. The kind and degree of model support here will vary
widely across the span of the different subforms of the Divisionalized Form. As
noted under the discussion earlier in this chapter, highly integrated
Divisionalized Forms may be able to use comprehensive models to reintegrate
into Machine Bureaucracies or Flexible Bureaucracies, whereas true
conglomerates may apply model support to achieve superior performance
control and streamlined, common financing.
The third model, the Assisting Model, supports only one main configuration,
one that is not among the IT-based configurations: the Professional
Bureaucracy. The reason is that it is only assisting—helping professionals to
perform their tasks better and/or more efficiently. Although it can be very
advantageous (securing much greater consistency and quality in an
organization’s products and services) and also facilitate considerable
reorganization of work in particular organizations, it cannot in my view support
genuinely new structural configurations. Even a model-driven Professional
Bureaucracy will remain a Professional Bureaucracy. Although an increased
span of competence and programmed routines make it possible to reshape work
processes to a certain degree, the structure of the organization as such will not
change much as long as the professional judgment of experts, as well as the
norms of the professional groups and of the greater professional community,
lies at the heart of the organization and decides the main features of the work
process.
413
17 Concluding Considerations
“He who bears in his heart a cathedral to be built is already

victorious. He who seeks to become sexton of a finished cathedral is
already defeated.”
Saint-Exupéry, Flight to Arras, 1942
Long on Constraints, Short on Possibilities?

Some readers may perhaps at this stage feel a sting of disappointment—missing
more spectacular technical predictions, more thrillingly novel organizational
configurations, and more splendidly liberating organization structures. They
may also feel that I have given too little attention to the more fashionable
contemporary visions and ideas debated in the industry today. For instance,
where is that virtual, networked knowledge organization based on
multidisciplinary teams, assembled on the go for a particular challenge, meeting
and working over the Internet and delivering its products in digital form directly
to the prosumer? If you have not seen it here, it may simply be because you did
not look hard enough—or because you mistook the physical topology of a
technical device such as the Internet for an organizational structure. I shall
elaborate a little on this.
Let us take a closer look at the virtual networked organization outlined
above. It would probably be a project organization put together to solve a
particular task or deliver a particular product or set of products. To do so, the
task at hand would have to be broken down into subtasks, which would be
distributed among the members of the organization. These members would then
have to coordinate their work and monitor it so that they would be able to
produce the desired result at the agreed time, with the agreed quality, and within
the agreed budget.
This coordination could be effected in several ways. First of all, one of the
project members could act as a main contractor, determine everything, hire in
the others, plan their work, and monitor and direct them as they
414 References
progressed. What kind of organization would we then have? I leave it to the

reader to decide. Then, of course, the task could be very unstructured and
pioneering, demanding a highly creative effort and involved cooperation
between a multitude of experts, all with a stake in the result. What kind of
organization would we then have? Again, I leave the answer to the reader. If in
doubt, consult Mintzberg’s short descriptions of the main properties of his
structural configurations quoted under the appropriate subheadings in Chapter
16.
My point here is simply that an organization coordinated over the Internet is
not necessarily a particular kind of organization any more than an organization
coordinated over the telephone is—or an organization coordinated via telegraph
or by smoke signals or messages speeded back and forth by horse riders. Here,
we should especially remember the crucial difference between dispersing an
organization physically and decentralizing its decision making. The nature of
the communication channels has no necessary bearing on either, although better
means of communication tend to make it easier to disperse people.
In my view, the metaphorical thinking that dominates much of the debate
about information technology and organizations today—which base
organizational concepts more or less directly on products or technologial
solutions—is not so much a result of profound insight as it is a sign of our rather
limited understanding of the deeper relationships between technological
capabilities and organizational opportunities. Organization structure and
functioning are more dependent on the nature of an organization’s main
coordinating mechanisms and decision making arrangements than on the nature
of its physical communication channels.
Does this contradict the postulation of five new information-technology–
enabled configurations earlier in this chapter? I believe not—since those
configurations are primarely based not on new communication arrangements,
but on new methods for coordination. New communication equipment may
constitute a part of the technological basis for these new coordination methods,
but its application does not necessarily create new organizational forms all by
itself.
Practical Theory
The main reason for the apparent lack of really exotic new organizational forms
in my analysis is that my goal has been first and foremost to obtain results that
have practical value. It is of course possible to make bolder predictions and
envision more breathtaking organizational structures. There is, indeed, no lack
of such prophecies. Being a consultant as well as a researcher, however, my
goal was not only to achieve an understanding of the basic relationship between
information technology and
References 415
organization, but also to produce practicable models and theories—close

enough to real life to enable me to offer better advice about how my clients’
organizations can really come to grips with this new and exciting technology. I
have therefore, throughout my analyses, striven to temper the purely
technological possibilities with the basic human constraints and preferences that
will continue to limit and shape our use of any new technology. This implied a
definitive departure from the technological stream-of-consciousness literature
represented by, for instance, Being Digital by Nicholas Negroponte and
Microcosm by George Gilder, and a concentration on the kind of technology use
that is possible and probable in a normal human and social setting.
To use a simple illustration from another technological domain: the fact that
we can easily build cars that go faster than 250 kilometers per hour, and that
most commercially available cars can go faster than 150, does not by itself
make it practical to use such speeds routinely in densely populated areas. In
fact, most countries do not deem it practical (and, hence, legal) to use such
speeds under any circumstances (except for competitions on specially
designated tracks).
Likewise, the fact that information technology makes it possible for people
who live in the most remote corners of the globe to work on common
documents, pass email to each other, and even (in a number of years) confer via
high-quality videophones does not by itself mean that organizations consisting
of only such scattered individuals will be desirable or even viable other than for
very special (and marginal) purposes. This is what the idea of the space of
constructible organizations is all about—to delineate the realistically available
alternatives for organizing work within a given culture and with a certain
technological level.
Can theory be at all helpful in practical matters? One may perhaps think that
answers to practical problems are best sought by accumulating experience, but,
in this case, I soon concluded that we most of all lacked an adequate body of
theory that could help us analyze experience and advance our understanding of
the deeper relationships between information technology and organization. The
link here is really quite straightforward. Without adequate theory, practical
questions about how to take advantage of information technology—like the
ones I mentioned in the introduction, that came from the audience when I
lectured in the late 1980s—simply cannot be credibly answered. Without theory
to help us interpret our experiences, we will not be able to understand much
about what is going on and why, let alone chart a viable course into the future
and sense potentials unrealized so far.
In order to provide the kind of practical, effective advice I wanted to be able
to give my clients, then, factual knowledge and experience is not enough. To
obtain a sound understanding of a particular field, experience,
416 References
factual knowledge and theory are all mandatory. Sometimes an unconscious,

everyday theory-in-use (Argyris 1980) may suffice, but for the large, complex
organizations of our age, explicit scientific theories are necessary as well.
As JoAnne Yates says, summing up her very interesting work on the
development of methods and technology for management control and
communication in American industry between 1850 and 1920 (Yates 1989, pp.
274–275, my italics):
Perhaps the most obvious implications concern communication and information

technology. James R. Beniger has recently argued that the “Control Revolution”
that began in the late nineteenth century contained the seeds of today’s information
society. Certainly, there are some parallels between the revolution in office
technology of the 1880–1920 period and the revolution of the last twenty-five
years. Recent innovations in computers and telecommunications have been so
spectacular that contemporary commentators tend to focus solely on the
technology, seeing it as the driving force causing changes in other parts of the
organization. The case studies in this book, however, illustrate some of the
problems with simple technological determinism. Technologies were adopted, not
necessarily when they were invented, but often when a shift or advance in
managerial theory led managers to see an application for them. Moreover,
technologies were often adopted simply to facilitate existing managerial methods;
potentially more powerful applications, such as the use of the telegraph for railroad
dispatching, were ignored for long periods. The technology alone was not
enough—the vision to use it in new ways was needed as well.
A related implication for contemporary issues concerns both communication
technology and geographical dispersion. Just as the telegraph once opened up
possibilities for wider domestic markets and more scattered production facilities to
companies such as Scovill and Du Pont, worldwide telecommunications systems
are now doing the same for international markets. The historical cases suggest,
however, that the real potential of these networks cannot be realized through a
simple extension of existing patterns of communication. Real gains await
innovative thinking about underlying managerial issues.
Therefore, when we encounter a new and uncharted territory like the interplay
between computers and organizations, “nothing will be so practical as the
development of a good new theory,” as Daft and Lewin (1993) note (with due
reference to Kurt Lewin 1). As Yates attests, the future can seldom be forecasted
by extrapolation, and to envisage potential new arrangements, it does not suffice
to make empirical investigations of the current best practice. Without theory, we
cannot distinguish between the significant and the insignificant, we cannot
easily perceive causal relationships, and we cannot predict likely outcomes in
new situations. Even
1Another, older Lewin. Daft and Lewin here refer to the article “The Research Center for Group
Dynamics at Massachusetts Institute of Technology” by Kurt Lewin, appearing in Sociometry, 1945
(vol. 8), pp. 126–135.
References 417
today we are in the infancy of computer use, and no one would seriously
propose that our results so far fathom the technology’s potential or contain the
complete blueprint for any future best practice.
So, first of all, we need a good theoretical foundation for the interplay of
information technology and organization. Equally important, this foundation
should not be built in isolation, but should relate directly to the established body
of organization theory. It is very unlikely that the introduction of a new
technology alone (albeit a powerful one) should alter the basic principles of
human interaction beyond recognition, and by segregating the study of
computers and organization from the rich body of organization research, we are
bound to forgo major insights and take on a crippling burden of parallel
research. In a field where there are many different and partly competing
theoretical approaches, it is also of significant scientific interest to test
established theory by systematically applying it to new problems. To me, this is
also a matter of practical concern; a large part of today’s managers know a lot
about organization theory and feel quite at home with the main lines of
argument. Linking a theory of information technology and organization to one
of these traditions will make it much easier for them to relate to it, to understand
it, and to use it for their own purposes. This was also one of my reasons for
anchoring my analysis in Mintzberg’s configurations. They have proved to be
very useful tools for my analyses, and I also feel they have passed their
introduction into the computer age with flying colors, showing their strength
through their adaptability.
Some Suggestions for the Practical Use of this Book

It is possible to work for improvement in an organization on several levels, from
the discharge of a single task to the structuring of the total organization.
However, single tasks are normally not very interesting targets—and tasks are,
in themselves, not tenable analytical units if you want to take full advantage of
the new possibilities offered by information technology, since any task is likely
to be a construction shaped by reigning conventions and traditional tools. Tasks
are, moreover, loci of functions, and function-oriented analysis becomes all too
easily bogged down in detailed descriptions of existing routines, obstructing our
comprehension of what the organization is really doing and restricting our
creativity in the design of new solutions.
To obtain a satisfactory level of insight and understanding, it will usually be
necessary to analyze the problem domain quite carefully in order to secure both
a tenable technological solution and a good
418 References
implementation of it, even in quite small and simple projects. If the goal is (as it
should be) to go beyond traditional arrangements and make full use of the new
technology, it is very important that the initial analysis aims at getting behind
existing work arrangements to capture the gist of the work at hand—or, rather,
its objectives, since the work (the present tasks) itself may be superfluous within
a new framework.
This is not a trivial requirement. A good example is the accounts payable
function in Ford. The key to the radical improvement achieved there was not a
system that made the work of the accounts payable department more efficient; it
was, on the contrary, the realization that the invoice itself—and hence most of
the traditional accounts payable function—could be eliminated through the
creative use of information technology. Such breakthroughs are the dramatic
goal of most reengineering projects, but they are notoriously difficult to
achieve—there are no surefire methods available, since in the end all radically
new solutions hinge on the inspired creativity of the project team members. The
best we can do in the way of methodology is to devise a number of coaxing
exercises—as Davenport suggests in his very thorough study (Davenport
1993)—and hope that creativity will manifest itself. Of course, such a process is
not entirely serendipitous; good coaxing strategies will produce much better
results than bad ones.
To avoid being trapped by existing procedures, it is necessary to employ a
top-down approach in the initial analysis, starting with the primary objectives at
the highest organizational level relevant to the project: What is, quite simply,
the nature of the products and services we aim to provide? The goal should be to
describe the desired implementation of these objectives at the level of products,
services, customers, or clients, and to chart the way they are related. I would
propose an object-oriented approach for this, since it will force us to focus
precisely on the central objects and help us avoid function analysis with its
penchant for detailed descriptions of existing tasks and routines. Consequently,
I would use object-oriented concepts and charting notations as tools in this work
and for documentation—for instance those provided in Jacobson, Ericsson, and
Jacobson (1994). There are also several comparable approaches available,
although many would say that Jacobson et al. have presented the best one so far.
One of the Jacobsons, Ivar, is also one of the Three Amigos behind the new,
unified modeling language, UML (together with Grady Booch and James
Rumbaugh).
The analysis should then proceed on one to three main levels, depending on
what is appropriate for the project in question: product-related possibilities,
process-related possibilities, and structure-related possibilities. These levels
correspond to the three levels of IT utilization discussed in this book, as shown
in Table 17-1. The boundaries between these three
References 419
levels are of course blurred, but, by and large, there is a correspondence that is
useful for both analysis and design.
Products and Services

Information technology has become one of the main enablers both for
improvements in products and services and the development of totally new
kinds of products and services. Very often, such advances hinge on one
particular aspect of the technology. For instance, the development of services as
diverse as today’s flexible and efficient airline reservation systems, automatic
teller machines, and electronic toll fee stations has been totally dependent on the
existence of powerful databases with remote access.
When we have a reasonably clear picture of the objectives of the organization
unit we are working with, its customers, their requirements, and the kinds of
products and services we would like to provide, our design work can therefore
be helped by a careful look at the discussions in Chapter 9 (“The IT-Based
Preconditions”), which is about the strong and weak points of information
technology and where it offers possibilities beyond earlier technology. Some of
the central aspects of this discussion are further elaborated in Chapter 11 (“The
Individual
Level of Analysis Level of IT Support
Products and Services Direct utilization of information technology

properties
Changes in products and services or the
devlopment of totally new ones made possible by Discussions mainly in Chapter 9, some in
the use of information technology. Chapter 11, emotional defenses in Chapter 10.
Processes Computer-dependent coordination methods

Cross-organization coordination and integration Discussions mainly in Chapters 12 through 14,
of tasks involved in the production and delivery of some in Chapter 11, emotional defenses in
particular products and services or classes of Chapter 10.
products and services
Structure Active models

Integration and information-technology–based Discussions in Chapters 15 and 16.
coordination on the level of the total
organization.
Table 17-1: The relationships between levels of analysis and levels of

IT support.
420 References
and the Group”), which also brings up the subject of self-service—a very
important factor in several business areas in the future, not least financial
products and services. To avoid being unduly constrained by the contemporary
technological level, and to plan more realistically for some years into the future,
Chapter 8 (“Information Technology Characteristics”) should also prove useful.
Finally, to temper the techno-optimism and avoid the emergence of an unbridled
“chip-chip-hurrah” mentality, I would recommend Chapter 10 (“Emotional
Barriers and Defenses”).
Processes
Even though processes and services usually rest mainly on one aspect of
information technology and thus depend directly on specific hardware and
software products, their provision often involves more complex organizational
processes spanning several organization units. Or, to view it from the opposite
perspective, practically all medium-to-large organizations will have a number of
processes that are central to their operations—some of which will produce and
deliver products and services to outside customers/clients, and some of which
will serve vital administrative needs. The key to these processes is
coordination—of the efforts of those who are part of the process and of the
customers, suppliers, and other parties inside and outside the organization.
To fathom the new possibilities for coordination provided by information
technology, it should be useful to look at the chapters in Part IV. Chapter 11
(“The Individual and the Group”) is perhaps the least interesting here, but it
offers some ideas on the usefulness and limitations of information technology in
the coordination of workgroups and teams. Chapter 12 (“Routines and
Automation”) should offer inspiration in the field of automation and task
elimination, and Chapter 13 (“Coordination by Default”) provides important
signposts for forays into the exiting realm of implicit coordination—a
coordinating mechanism with huge potential. Chapter 14 (“Comprehension and
Control”) outlines the possibilities arising from the greatly improved availability
of information provided by information technology, especially the possibilities
for centralization and decentralization and the concomitant migration of power
within the organization. Finally, Chapter 10 (“Emotional Barriers and
Defenses”) is useful for avoiding the worst pitfalls of too technocratic
approaches.
Structure
Today, most of the focus in the literature on information technology and
organization is on the process level. Compared to the situation a couple of
decades ago, when systems were still viewed mainly as specific tools for
References 421
rather narrow functions in the organization, this represents a significant step

forward. It is also a step up onto a higher level of complexity, perhaps the
highest level we can presently handle with some confidence. However, as we
gradually integrate our processes, even they will become candidates for closer
integration and coordination. We then reach a level where the whole
organization—and often a part of its environment as well—must be described in
the same model and served by a set of intergrated systems. This involves rising
to yet a higher level of complexity—today barely within reach and only for
organizations with a well defined and fairly narrow problem domain. Tackling
integration at the organization level will require thorough understanding of the
relationship between work, technology, and organization, and we will need
advanced methods for analysis, description, and modeling. At this level,
organization structure becomes one of the paramount issues.
Organization structure is a subject of considerable interest already at the
process level, as key processes can involve large numbers of people and many
organization units. To achieve the best possible results, it is always important to
choose structures that match our objectives and the nature of the required
processes and systems central to those processes. Sometimes, we have a choice
between process designs calling for different structures, and it is important to
know the strengths and weaknesses of those structures if the desired results are
to be achieved. However, structure first becomes a paramount concern when we
do not simply approach a single process, but try to go one step further and
integrate processes, support functions, systems, and system use across the total
organization.
When we work at this level, the matters discussed in Part IV are still
important, but the most significant contributions should come from the two first
chapters in Part V. Chapter 15 (“Toward the Model-Driven Organization”)
should serve to increase our understanding of the potential of conceptual
modeling, and point the way to how such models can form the basis for really
comprehensive computer-based systems—and thus allow organizations to
achieve new levels of integration and coordination. Chapter 16 (“The New
Configurations”) discuss in greater detail what the potential is for different types
of organizations. This should help make us more aware of the potentials of our
own organization and the ways in which it could (and could not!) be
transformed, maybe even into a totally different configuration.
Limits to Flexibility—But Not to Costs?

Throughout this book I have tried to maintain a prudent attitude to the potential
of information technology, especially to the possibilities we have
422 References
for actually reaping its promised benefits through practical implementations. It

is a very complex technology, difficult to master even in itself, and when it is
inserted into such complex social constructs as our modern organizations, it is
indeed a challenge to understand and manage the compounded ramifications.
I would therefore like to temper the fairly upbeat tone of the last two chapters
with a few words of caution here at the end, with reference to my own
discussions and to those taking place within the field of information technology
and organizations. I will consider two issues in particular: flexibility and cost.
They are, in fact, intimately related.
First of all, I believe that the case for flexibility has been overstated in the
general debate on these matters. I do agree that information technology will
allow us to build more flexible organizations that can respond more quickly and
more accurately to changes in their environment and challenges from their
competitors. Hyperautomation can accommodate much more variation than
traditional automation, and increased spans of competence will have similar
effects. I have detailed my views on this subject in Chapter 16, particularly in
the section on the Flexible Bureaucracy. However, as indicated in that section,
this flexibility is highly circumscribed—if the organization comes up against a
problem that requires responses outside the functional scope of its systems, it
will often prove much less flexible than an organization without any systems at
all. The reason is simply that in order to respond it will have to modify its
systems, which is necessarily a thorough and costly process for any significant
change.
In theory, a computer-based organization can of course resort to ad hoc
manual solutions just as easily as a non-computerized organization could, but
that will hardly be true in practice. The technology-dependent organization will
neither have the workforce nor the culture to revert effortlessly to such
traditional means of doing business, and it will often experience great trouble in
the process. Not that it would necessarily help if it did revert—the reason for its
predicament would probably be that one or more of its competitors had
developed the more comprehensive systems capable of meeting exactly that
kind of challenges. If so, even an adequate manual solution would be of little
use, since it would be too expensive to maintain over an extended period of time
compared to the more efficient systems deployed by competitors.
If the new demands are such that they can be met by small adjustments or
additions to the existing systems, these problems are manageable—a solution
can be in place in a matter of days, weeks, or a few months (depending on the
scope of the changes, the technical nature of the system, the quality of the
underlying models and the competence of the IS staff). However, from time to
time, organizations that depend heavily on
References 423
information technology will face a situation where the underlying structure of

their systems and/or the basic technological solutions they employ cannot
accommodate the necessary changes. They will then have to renew one or more
of their systems completely, or face a situation where they may have to leave
the field in question altogether, ceding it to the competition.
Total renewal of principal systems is a major operation involving
considerable risk, high cost, and a significant period of time. The thorough
analysis, the modeling, the creative efforts involved in designing new solutions
that will have to last for quite a number of years, the painstaking labor to work
out the design in sufficient detail, then the challenging process of rendering the
design into executable program code, of testing the code, and finally the
implementation of the new system in the organization, all this combine to make
such projects major undertakings. Often, the great cost and considerable time
required to do a thorough job on analysis and design tempt people to take
shortcuts, frequently with catastrophic results: The systems either end up
without critical functionality or the development process gets bogged down in
endless and ill-structured modifications.
When talking about such major changes, it is also important to realize that
after the introduction of new systems and routines, the people working in the
organization will need considerable time to bring their performance up to par
with the new tools and within their new work environment. No system can
deliver peak performance without able and experienced operators, and all the
important new adjoining routines must also be mastered. This is a fact often
overlooked by the champions of perpetual change. Because the very fabric of
any organization consists of well-established patterns of action, it stands to
reason that one cannot rearrange those patterns too extensively too often and at
the same time maintain an efficient organization. I started my career in the
personnel department of a shipyard, and I still remember my experienced
superior telling me that it would normally take a newly hired hand without
previous experience a full year to reach the expected normal performance for an
unskilled laborer there. It would be strange if it should take very much less time
to develop the proficiency needed to make full use of advanced new computer-
based systems in a new organizational setting with significantly redesigned
tasks.
My message here is not that the problems are insurmountable, nor that they
are so severe that they will pose a permanent threat to the IT–based
organization. I merely want to underscore the fact that the increased
productivity and flexibility provided by information technology come at a cost,
as do all other improvements, and that the use of this technology increases the
height of the steps in the stairs leading upward—in terms of
424 References
cost, efforts, and rewards—just as the transition from craft production to the
modern organization did. Figure 17-1 illustrates the relative merits of these
three generations of organizations when it comes to flexibility, productivity, and
cost of change.
It is of course possible to prepare the organization for the system renewals
that has to come at certain intervals. Apart from maintaining adequate financial
capabilities and an organization-wide realization that change is necessary
(which are general measures important for any type of change), minor changes
as well as major new development efforts can be speeded up and achieve much
higher quality levels if the organization and its functional requirements are
analyzed and modeled on a continuous basis—not just when a specific need
arises. As I mentioned at the end of Chapter 15, any enterprise with aspirations
to become a model-driven organization should have a pattern manager, and one
of the duties of this office would be to continuously maintain and update the
conceptual model of the organization and its problem domain. All changes to
existing systems, as well as all new systems, should have this model as their
foundation.
Traditional
Information
Machine
Craft Technology–
Bureaucracies
Organizations Based
(Manual Routines/
Mechanical Organizations
Automation)
Flexibility
Productivity
Cost of Change
Figure 17-1: The relative merits of three generations of organizations. The areas
of the circles do not correspond to exact values; they are only
meant to illustrate the relative levels of performance and resource
requirements of the different generations.
References 425
If such a model is sound, it will ensure that the systems that are implemented
will be better prepared for later changes; the changes will be consistent in their
basic features, and new systems will be much easier to integrate with older ones.
In a large organization the maintenance of such a model may require the full-
time work of several persons, but the cost involved will be recouped many times
over through reductions in development costs and increases in systems quality.
The Knowledge Factor

Throughout the pages of this book, I have repeatedly stressed the complexity of
large computer-based systems and their intricate interplay with the
organizations that use them. I have also touched upon the considerable skills in
analysis, planning, organization, and engineering that is necessary to create,
implement, maintain and use these systems and design the organizations that
can make the most of them. This is a fact that cannot be stressed too strongly. I
do not think there is any other wide-spread technology where relative
differences in knowledge and skills are so decisive for the success of an
investment. Two similar firms can therefore invest the same amount of money
in the same systems and experience radically different results.
In a business world hunting for competitive advantage, this is an often
neglected fact—even though it is widely acknowledged in IS circles that the
best programmers work several times faster than the average, and produce far
better programs with fewer errors to boot. Some of the same is true for systems
analysts, and even on the purely technical side, the difference between a good
engineer and a mediocre one can translate into hours and days of system
downtime. In a future where organizations are becoming increasingly model-
driven, superior organization designers and Organization Design Managers
(“pattern managers”) will also become very important—even crucial—for the
competitiveness of large organizations.
The firm with superior knowledge and skills both in the systems and the
organization area, and with these skills combined in a cross-disciplinary
professional organization with a management team that can also think business,
will therefore have a significant competitive advantage, which will be more
important the more IT-intensive a business is. Today, both hardware, quite
sophisticated standard software and software tools are freely available to all,
and news about radical new and profitable ways of exploiting IT get around
quickly. Experience has also shown us that the barriers to entry that are erected
by early adopters of innovative systems tend to erode over time even if they are
initially successfull, as customers
426 References
and competitors gradually find ways of circumventing the defenses and break
the lock-in.
However, teams with superior knowledge and skills take time to build, are
fairly inconspicuous, and very hard to replicate. Experience across diverse
organizations, from football clubs to universities, shows that such superior
teams exhibit a significant tendency toward self-reinforcement, and that
established superiority in skills and knowledge is possible to maintain over long
periods of time. For those of us who look for an unobtrusive competitive edge,
this should be a very interesting option, and for those of us who really love
computers and their intricate interplay with organized humans, it is indeed a
heartening conclusion to this book.
References 427
References
Agor, Weston H. (ed.) (1989): Intuition in Organizations: Leading and Managing

Productively, Newbury Park, CA, Sage Publications.
Anderson, John R. (1990): Cognitive Psychology and its Implications, Third Edition, New
York, W. H. Freeman and Company.
Argyris, Chris (1980): Inner Contradictions of Rigorous Research, New York, Academic
Press.
Ashby, W. Ross (1956): “Variety, Constraint and the Law of Requisite Variety,” in Ashby, W.
Ross: An Introduction to Cybernetics, London, Chapman and Hall. Reprinted in
Buckley, Walter: Modern Systems Research for the Behavioral Scientist, Chicago,
Aldine Publishing Company, 1968.
Ashby, W. Ross (1962): “Principles of the Self-Organizing System,” in Foerster, Hans von and
Zopf, Georg W. (eds.): Principles of Self-Organization, New York, Pergamon Press.
Reprinted in Buckley, Walter: Modern Systems Research for the Behavioral Scientist,”
Chicago, Aldine Publishing Company, 1968.
Asimov, Isaac (1975): Asimov’s guide to Science, Vol. 2, The Biological Sciences, London,
Pelican Books.
Augarten, Stan (1984): Bit by Bit: An Illustrated History of Computers, New York, Ticknor &
Fields.
Baran, Nick (1991): “Breakthrough in Holographics Memory Could Transform Data Access,”
Microbytes column, Byte, January 1991.
Barber, Paul (1988): Applied Cognitive Psychology: An Information-processing Framework,
London, Routledge.
Bartlett, Christopher A., and Ghoshal, Sumantra (1989): Managing Across Borders, Boston,
Harvard Business School Press.
Beals, Carleton (1970): Stories Told by the Aztecs Before the Spaniards Came, London,
Abelard-Schumann.
Beniger, James R. (1986): The Control Revolution: Technological and Economic Origins of
the Information Society, Cambridge, Harvard University Press.
Berger, Peter and Luckmann, Thomas (1967): The Social Construction of Reality, London,
Penguin Press.
Bertalanffy, Ludwig von (1973): General Systems Theory, Harmondsworth, UK, Penguin
Books.
Bijker, Wiebe E. and Law, John (1992): Shaping Technology/Building Society. Studies in
Sociotechnical Change, Cambridge, MA, The MIT Press.
Bing, Jon, and Harvold, Trygve (1977): Legal Decisions and Information Systems, Oslo,
Universitetsforlaget (University Press of Norway).
428 References
Boyd, Robert, and Richerson, Peter J. (1985): Culture and the Evolutionary Process, Chicago,
University of Chicago Press.
Booch, Grady (1991): Object Oriented Design, Redwood City, Benjamin/Cummings
Publishing Company, Inc.
Boulding, Kenneth E. (1956): “General Systems Theory—The Skeleton of Science,”
Management Science, 2, pp. 197–208. Reprinted in Buckley, Walter: Modern Systems
Research for the Behavioral Scientist, Chicago, Aldine Publishing Company, 1968.
Braverman, Harry (1974): Labor and monopoly Capital: The Degradation of Work in the
Twentieth Century, New York, Monthly Review Press.
Bright, James R. (1985): “Evolution of Automation: Electric Lamps,” in Rhodes, Ed, and
Wield, David: Implementing New Technologies: Choice, Decision and Change in
Manufacturing, Oxford, Basil Blackwell (in cooperation with The Open University),
1985.
Buckley, Walter (1967): Sociology and Modern Systems Theory, Englewood Cliffs, NJ,
Prentice Hall.
Buckley, Walter (1968): Modern Systems Research for the Behavioral Scientist, Chicago,
Aldine Publishing Company.
Buckmann, Peter (producer) (1991): “Hansatid,” a series of 3 television programs about the
Hanseatic League produced by the Norwegian National Television and broadcast in
January 1991.
Bush, Vannevar (1945): “As We May Think,” Atlantic Monthly 176 (1), June 1945. Reprinted
in: Greif, Irene (ed.) (1988): Computer Supported Cooperative Work: A Book of
Readings, San Mateo, CA, Morgan Kaufmann Publishers, Inc.
Campbell-Kelly, Martin (1997): A History of the Information Machine, New York,
HarperCollins.
Ceruzzi, Paul E. (1998): A History of Modern Computing: 1945-1995, Cambridge, MA, MIT
Press.
Chandler, A. D. Jr. (1977): The Visible Hand: The Managerial Revolution in American
Business, Cambridge, MA: Harvard University Press.
Chinnock, Chris (1997): “Digital Ink Gives New Meaning to Paper Recycling”, Byte, August
1997.
Christopher, Martin (1992): Logistics and Supply Chain Management—Strategies for
Reducing Costs and Improving Services, London, Pitman Publishing.
Clarkson, Mark A. (1991): “An Easier Interface,” Byte, February 1991.
Clegg, Stewart R. (1990): Modern Organizations: Organization Studies in the Postmodern
World. London, Sage Publications.
Coad, Peter, and Yourdon, Edward (1991): Object Oriented Analysis (Second Edition),
Englewood Cliffs, NJ, Prentice Hall.
Cohen, John (1980): The Lineaments of Mind—In Historical Perspective, Oxford, W.H.
Freeman and Company.
Crozier, Michel (1964): The Bureaucratic Phenomenon, Chicago, University of Chicago Press.
Daft, Richard L., and Lewin, Arie Y. (1993): “Where Are the Theories for the ‘New’
Organizational Forms? An Editorial Essay,” Organization Science, 4 November 1993.
Dahl, Helmer (1984): Teknikk, kultur, samfunn: Om egenarten i Europas vekst, Oslo,
Ingeniørforlaget.
Davenport, Thomas H. (1993): Process Innovation: Reenginering Work through Information
Technology, Boston, Harvard Business School Press.
References 429
Drucker, Peter (1988): “The Coming of the New Organization,” Harvard Business Review,
1988 (1).
Eason, Ken (1988): Information Technology and Organizational Change, London, Taylor and
Francis.
Ellis, Henry C., and Hunt, R. Reed (1989): Fundamentals of Human Memory and Cognition
(Fourth Edition), Dubuque, IA, Wm. C. Brown Publishers.
Eriksen, Trond Berg (1987): Budbringerens overtak, Oslo, Universitetsforlaget (University
Press of Norway).
Fineman, Stephen (ed.) (1993): Emotion in Organizations, London, Sage Publications.
Flam, Helena (1993): “Fear, Loyalty and Greedy Organizations,” in Fineman, Stephen (ed.)
(1993): Emotion in Organizations, London, Sage Publications.
Fowler, Martin (1997): UML Distilled. Applying the Standard Object Modeling Language,
Reading, MA, Addison-Wesley Longman, Inc.
Galbraith, Jay R. (1977): Organization Design, Reading, MA, Addison-Wesley Publishing
Company.
George, Joey F., and King, John L. (1991): “Examining the Computing and Centralization
Debate,” Communications of the ACM, July 1991.
Geschwind, Norman (1979): “Specializations of the Human Brain,” Scientific American,
September 1979.
Gleick, James (1988): Chaos—Making a New Science, London, Sphere Books.
Goffman, Erving (1959): The Presentation of Self in Everyday Life, New York, Doubleday &
Co.
Goffman, Erving (1970): Strategic Interaction, Oxford, Basil Blackwell.
Goffman, Erving (1974): Frame Analysis, Cambridge, MA, Harvard University Press.
Goldberg, Philip (1989): “The Many Faces of Intuition,” in Agor, Weston H. (ed.) (1989):
Intuition in Organizations: Leading and Managing Productively, Sage Publications.
Goody, Jack (1986): The Logic of Writing and the Organization of Society, Cambridge,
Cambridge University Press.
Granovetter, Mark (1985): “Economic Action and Social Structure: The Problem of
Embeddedness,” American Journal of Sociology, 91 (3).
Graves, Robert (1960): The Greek Myths: 1, Revised Edition (first published 1955),
Harmondsworth, UK, Penguin Books.
Greif, Irene (ed.) (1988): Computer Supported Cooperative Work: A Book of Readings, San
Mateo, CA, Morgan Kaufmann Publishers, Inc.
Hamilton, Gary G., and Biggart, Nicole Woolsey (1988): “Market, Culture, and Authority: A
Comparative Analysis of Management and Organization in the Far East,” in Winship,
C., and Rosen, S. (eds.): “Organizations and Institutions: Sociological Approaches to the
Analysis of Social Structure,” supplement to American Journal of Sociology, 94.
Hammer, Michael (1990): “Reengineering Work: Don’t Automate, Obliterate,” Harvard
Business Review, 1990 (4).
Hammer, Michael and Champy, James (1993): Reengineering the Corporation—A Manifesto
for Business Revolution, London, Nicholas Brealey Publishing.
Hannan, Michael T. and Freeman, John (1977): “The Population Ecology of Organizations,”
American Journal of Sociology, 82 (5).
Hannan, Michael T. and Freeman, John (1984): “Structural Inertia and Organizational
Change,” American Sociological Review, 49 (April).
Harvey, David A. (1990): “State of the Media,” Byte, November 1990.
430 References
Hassard, John, and Parker, Martin (1993): Postmodernism and Organizations, London, Sage
Publications.
Hatvany, J. et al. (1985): “Advanced Manufacturing Systems in Modern Society,” in Rhodes,
E., and Wield, David: Implementing New Technologies: Choice, Decision and Change
in Manufacturing, Oxford, Basil Blackwell (in cooperation with The Open University),
1985.
Havelock, Eric (1986): The Muse Learns to Write—Reflections on Orality and Literacy from
Antiquity to the Present, New Haven, CT, Yale University Press.
Heller, Martin (1991): “AI in Practice,” Byte, January 1991.
Hesselholt, N., Ockens, E., Christensen, E., Elling, F. and Mikkelsen, M. (1992):
“Identifikation af hunde med microchips,” Dansk Veterinærtidsskrift, 75 (14). English
summary.
Hochschild, Arlie R. (1983): The Managed Hearth, Berkeley, University of California Press.
Hochschild, Arlie R. (1993): “Preface,” in Fineman, Stephen (ed.) (1993): Emotion in
Organizations, London, Sage Publications.
Hofer, Myron A. (1981): The Roots of Human Behavior. An Introduction to the Psychobiology
of Early Development, San Francisco, W.H. Freeman and Company.
Hollway, Wendy (1991): Work Psychology and Organizational Behaviour—Managing the
Individual at Work, London, Sage Publications.
Hopper, Max D. (1990): “Rattling SABRE—New Ways to Compete on Information,” Harvard
Huppes, Tjerk (1987): The Western Edge—Work and Management in the Information Age,”
Dordrecht, Netherlands, Kluwer Academic Publishers.
Jacobson, Ivar, Ericsson, Maria, and Jacobson, Agneta (1994): The Object Advantage:
Business Process Reengineering with Object Technology, ACM Press Books/Addison
Wesley Publishing Company.
James, Peter, and Thorpe, Nick (1994): Ancient Inventions, New York, Ballantine Books.
Johansen, Robert (1988): Groupware: Computer Support for Business Teams, New York, The
Free Press.
Kahneman, Daniel, Slovic, Paul, and Tversky, Amos (eds.) (1982): Judgment under
uncertainty: Heuristics and biases,” Cambridge, Cambridge University Press.
Katzenbach, Jon R., and Smith, Douglas K. (1993): The Wisdom of Teams—Creating the
High-Performance Organization, Boston, Harvard Business School Press.
Keen, Peter G. W. (1991): Shaping the Future: Business Design through Information
Technology, Boston, Harvard Business School Press.
Kepner, Charles H., and Tregoe, Benjamin B. (1965): The Rational Manager—A Systematic
Approach to Problem Solving and Decision Making, New York, McGraw-Hill.
Kitto, H. D. F. (1951): The Greeks, Harmondsworth, UK, Penguin Books. I have used the
Norwegian translation Grekerne, published in 1971 by Gyldendal Norsk Forlag, Oslo.
Koolhaas, Jan (1982): Organization Dissonance and Change, Chichester, UK, John Wiley &
Sons.
Kuhn, Thomas S. (1970): The Structure of Scientific Revolutions, Chicago, University of
Chicago Press.
Leach, E. R. (1970): Political Systems of Highland Burma, London, The Athlone Press—
University of London. First edition published in 1954 by G. Bell & Son.
Leavitt, Harold J. and Whisler, Thomas L. (1958): “Management in the 1980’s,” Harvard
References 431
Lie, Merete, and Rasmussen, Bente (1983): Kan “kontordamene” automatiseres, Research
report no. STF82 A83004, IFIM/SINTEF, Trondheim.
Lincoln, James R. (1982): “Intra- (and Inter-) Organizational Networks,” Research in the
Sociology of Organizations, Vol. 1.
Lloyd, Peter C. (1965): “The Political Structure of African Kingdoms,” in Banton, Michael
(ed.): Political Systems and the Distribution of Power, London, Tavistock Publications.
Long, Richard J. (1987): New Office Information Technology: Human and Managerial
Implications, Beckenham, Croom Helm Ltd.
Lundberg, Nils H. (1991): “Kjemometri i kjømda,” Teknisk Ukeblad (Technology Weekly,
Oslo), 1991 (36).
Lysgaard, Sverre (1961): Arbeiderkollektivet, Oslo, Universitetsforlaget (University Press of
Norway).
March, James G., and Simon, Herbert A (1958): Organizations, New York, John Wiley &
Sons.
Maus, Arne, and Espeli, Tron (1981): EDB i kontoret—produktivitet og sysselsetting, research
report no. 699, Norwegian Computing Center, Oslo.
McKersie, Robert B., and Walton, Richard E. (1991): “Organizational Change,” chapter 9 in
Scott Morton, Michael S. (ed.) (1991): The Corporation of the 1990s: Information
Technology and Organizational Transformation, New York, Oxford University Press.
Meyer, John W., and Rowan, Brian (1977): “Institutionalized Organizations: Formal Structure
as Myth and Ceremony,” American Journal of Sociology, 83 (2).
Meyer, Marshall W., and Zucker, Lynne G. (1989): Permanently Failing Organizations,
Newbury Park, UK, Sage Publications.
Mintzberg, Henry (1973): The Nature of Managerial Work, New York, Harper & Row.
Mintzberg, Henry (1979): The Structuring of Organizations, Englewood Cliffs, Prentice Hall.
Mintzberg, Henry (1989): Mintzberg on Management: Inside Our Strange World of
Organizations, New York, The Free Press.
Mishkin, Mortimer, and Appenzeller, Tim (1987): “The Anatomy of Memory,” Scientific
American, June 1987.
Moeller, Mike (1994): “Boeing Goes On-Line with 777 Design,” Computer-Aided
Engineering, 13 (8).
Morgan, Gareth (1986): Images of Organization, Beverly Hills, CA, Sage Publications.
Morris, Desmond (1967): The Naked Ape, London, Jonathan Cape.
Murra, John W. (1986): “The expansion of the Inka state: armies, war, and rebellions,” in
Murra, John W., Watchel, Nathan and Revel, Jaques: Anthropological History of
Andean Politics, Cambridge, Cambridge University Press, 1986.
Nadler, David, and Tushman, Michael (1988), Strategic Organization Design, Glenview, IL,
Scott, Foresman and Company.
Nash, Manning (1966): Primitive and Peasant Economic Systems, Scranton, PA, Chandler
Publishing Company.
Negroponte, Nicholas (1995): Being Digital, New York, Alfred A. Knopf.
Nelson, Theodor H. (1988): “Managing Immense Storage,” Byte, January 1988.
Noyce, Robert N. (1977): “Microelectronics,” Scientific American, September 1977.
O’Hara-Deveraux, Mary and Johansen, Robert (1994): Globalwork: Bridging Distance,
Culture and Time, San Francisco, Jossey-Bass Publishers.
Ong, Walter J. (1982): Orality and Literacy, London, Methuen & Co.
Palme, Jacob (1995): Electronic Mail, Boston, Artech House.
432 References
Parish, Tom (1990): “Crystal Clear Storage,” Byte, November 1990.

Pehrson, Robert N. (1964): The Bilateral Network of Social Relations in Könkämä Lapp
District, Oslo, Norwegian University Press.
Pine, Joseph B. II (1993): Mass Customization—The New Frontier in Business Competition,
Boston, Harvard Business School Press.
Piore, Michael J., and Sabel, Charles F. (1984): The Second industrial Divide: Possibilities for
Prosperity, New York, Basic Books.
Polanyi, Michael (1967): The Tacit Dimension, London, Routledge & Kegan Paul.
Pool, Ithiel de Sola (1983): Forecasting the Telephone—A Retrospective Technology
Assessment of the Telephone, Norwood, Ablex Publishing Corporation.
Porter, Michael (1985): The Competitive Advantage: Creating and Sustaining Superior
Performance, New York, The Free Press.
Psaltis, Demetri, and Mok, Fai (1995): “Holographic Memories,” Scientific American,
November 1995.
Putnam, Linda L., and Mumby, Dennis K. (1993): “Organizations, Emotion and the Myth of
Rationality,” in Fineman, Stephen (ed.) (1993): Emotion in Organizations, London,
Sage Publications.
Quick, J. C., Murphy, L. R., Hurrel, J. J., and Orman, D. (1992): “The Value of Work, the Risk
of Distress, and the Power of Prevention,” in Quick, J. C., Murphy, L. R., Hurrel, J.J.
(eds.): Stress & Well-Being at Work, Washington, DC, American Psychological
Association.
Rachlin, Howard (1989): Judgment, Decision and Choice—a Cognitive/Behavioral Synthesis,
New York, W.H. Freeman and Company.
Rasmus, Daniel W. (1991): “Putting the Experts to Work,” Byte, January 1991.
Rhodes, E. and Wield, David (1985): Implementing New Technologies: Choice, Decision and
Change in Manufacturing, Oxford, Basil Blackwell (in cooperation with The Open
University).
Robinson, Mike (1991): “Through a Lens Smartly,” Byte, May 1991.
Rockart, John F. and Short, James E. (1991): “The Networked Organization and the
Management of Interdependence,” chapter 7 in Scott Morton, Michael S. (ed.) (1991):
The Corporation of the 1990s. Information Technology and Organizational
Transformation, New York, Oxford University Press.
Ross, Randall R., and Altmaier, Elizabeth M. (1994): Intervention in Occupational Stress,
London, Sage Publications.
Rowan, Roy (1989): “What it is,” in Agor, Weston H. (ed.) (1989): Intuition in Organizations:
Leading and Managing Productively, Newbury Park, CA, Sage Publications.
Scientific American, special issue on microelectronics, September 1977.
Sandars, N. K. (1964): The Epic of Gilgamesh, Harmondsworth, UK, Penguin Books.
Schmidt, Kjeld and Bannon, Liam (1992): “Taking CSCW Seriously,” Computer Supported
Cooperative Work, 1 (1–2).
Schoenberg, Robert J. (1985): Geneen, New York, W. W. Norton & Company.
Scott Morton, Michael S. (ed.) (1991): The Corporation of the 1990s: Information Technology
and Organizational Transformation, New York, Oxford University Press.
Scott, Richard W. (1987): Organizations: Rational, Natural and Open Systems, Englewood
Cliffs, NJ, Prentice Hall.
Sharp, Lauriston (1952): “Steel Axes for Stone Age Australians,” in Edward H. Spicer (ed.):
Human Problems in Technological Change, New York, Russel Sage Foundation.
References 433
Shotwell, James T. (1961): The Story of Ancient History, New York, Columbia University
Press. Reprint of The History of History, Volume I, originally published in 1939 (same
publisher).
Silverman, David (1970): The Theory of Organisations, London, Heineman Educational
Books.
Simon, Herbert A. (1976): Administrative Behavior, Third Edition, New York, The Free Press
(first published in 1945).
Simon, Herbert A. (1977): The New Science of Management Decision, Englewood Cliffs, NJ,
Prentice Hall.
Simon, Herbert A. (1981): The Sciences of the Artificial, Second Edition, Cambridge, MA,
MIT Press.
Simon, Herbert A. (1989): “Making Management Decisions: The Role of Intuition and
Emotion,” in Agor, Weston H. (ed.) (1989): Intuition in Organizations: Leading and
Managing Productively, Newbury Park, CA, Sage Publications.
Sobel, Alan (1998): “Television’s Bright New Technology,” Scientific American, May 1998.
Stein, Richard Marlon (1991): “Browsing Through Terabytes,” Byte, May 1991.
Stinchcombe, Arthur L. (1965): “Social Structure and Organizations,” in James G. March
(ed.): Handbook of Organizations, Chicago, Rand McNally.
Stix, Gary (1991): “Plane Geometry: Boeing Uses CAD to Design 130,000 Parts for its New
777,” Scientific American, March 1991.
Strassmann, Paul A. (1985): Information Payoff: The Transformation of Work in the
Electronic Age, New York, The Free Press.
Sundstrom, Eric (1986): Work Places: The psychology of the physical environment in offices
and factories, Cambridge, Cambridge University Press.
Thé, Lee (1995): “Workflow Tackles the Productivity Paradox,” Datamation, 41 (15).
Thompson, James D. (1967): Organizations in Action, New York, McGraw-Hill.
Thompson, Paul (1993): “Postmodernism: Fatal Distraction,” in Hassard, John, and Parker,
Martin: Postmodernism and Organizations, London, Sage Publications.
Thompson, Tom (1996): “What’s Next?” Byte, April 1996.
Thorud, Geir I. (project mgr.) (1991): Norsk veiledning i bruk av EDIFACT Versjon 1.0,
Sekretariatet for Norsk TEDIS.
Walton, Richard E. (1989): Up and Running: Integrating Information Technology and the
Organization, Boston, Harvard Business School Press.
Weber, Max (1968): Economy and Society, New York, Bedminster Press. Originally published
in Tubingen by J. C. B. Mohr (Paul Siebeck) in 1922 under the title Wirtschaft und
Gesellschaft. There are numerous translations. I have also used a Norwegian translation
of some passages, printed in Egil Fivelsdal (ed.): Max Weber: Makt og byråkrati, Oslo,
Gyldendal Norsk Forlag, 1971.
Weick, Karl E. (1979): The Social Psychology of Organizing, Reading, MA, Addison-Wesley
Publishing Company.
Weiss, Joseph (1990): “Unconscious Mental Functioning,” Scientific American, March 1990.
Wiener, Norbert (1954): The Human Use of Human Beings: Cybernetics and Society, New
York, Doubleday Anchor.
Williamson, Oliver E. (1975): Markets and Hierarchies: Analysis and Antitrust Implications,
New York, The Free Press.
Williamson, Oliver E. (1985): The Economic Institutions of Capitalism, New York, The Free
Press.
434 References
Wilson, Peter J. (1988): The Domestication of the Human Species, New Haven, CT, Yale
University Press.
Winograd, Terry (1988): “A Language/Action Perspective on the Design of Cooperative
Work,” Human Computer Interaction, 3 (1). Reprinted in: Greif, Irene (ed) (1988):
Computer Supported Cooperative Work: A Book of Readings, San Mateo, CA, Morgan
Kaufmann Publishers.
Yates, Francis A. (1966): The Art of Memory, London, Routledge and Kegan Paul.
Yates, JoAnne (1989): Control through Communication—The Rise of System in American
Management, Baltimore, MD, John Hopkins University Press.
Yates, JoAnne, and Benjamin, Robert I. (1991): “The Past and Present as a Window on the
Future,” in Scott Morton, Michael S. (ed.) (1991): The Corporation of the 1990s:
Information Technology and Organizational Transformation, New York, Oxford
University Press.
Zuboff, Shoshana (1988): In the Age of the Smart Machine: The Future of Work and Power,
New York, Basic Books.
435 435
Author Index
Agor, Weston H. 72 Clegg, Stewart R. 28, 29, 33, 111, 149,

Altmaier, Elizabeth M. 237 150-1, 153, 158, 171-3, 328
Anderson, John R. 62, 63, 67, 68, 71, 134 Coad, Peter 211, 213
Appenzeller, Tim 62 Cohen, John 64, 84
Argyris, Chris 52, 84, 168, 169, 240, 416 Crozier, Michel 173
Ashby, W. Ross 31, 42, 383
Asimov, Isaac 72 Daft, Richard L. 416
Augarten, Stan 28, 86, 120, 123 Dahl, Helmer 115, 125, 182
Davenport, Thomas H. 418
Bannon, Liam 287 Drucker, Peter 272
Baran, Nick 194
Barber, Paul 60, 61, 75 Eason, Ken 244, 353
Bartlett, Christopher A. 239, 249 Ellis, Henry C. 60, 62, 63, 66-8, 71
Beals, Carleton 101 Ericsson, Maria 418
Beniger, James R. 25, 112, 126, 406, 416 Eriksen, Trond Berg 77, 114, 119, 121
Benjamin, Robert I. 122, 123, 352, 408 Espeli, Tron 287
Berger, Peter 28, 110, 247
Bertalanffy, Ludwig von 30, 31, 42 Fineman, Stephen 79, 80
Biggart, Nicole Woolsey 100, 171, 172 Flam, Helena 80
Bijker, Wiebe E. 110 Fowler, Martin 355
Bing, Jon 196, 346 Freeman, John 159
Booch, Grady 211, 212, 213, 418
Boulding, Kenneth E. 26, 42 Galbraith, Jay R. 32, 40, 41, 44, 49
Boyd, Robert 110 George, Joey F. 325, 344, 415
Braverman, Harry 158 Geschwind, Norman 62
Bright, James R. 222 Ghoshal, Sumantra 239, 249
Buckley, Walter 42 Gleick, James 221
Buckmann, Peter 119 Goffman, Erving 29, 84, 96
Bush, Vannevar 208 Goldberg, Philip 72
Goody, Jack 117, 118, 150, 303
Champy, James 262, 279, 280, 288, 290, Granovetter, Mark 171, 172
294, 310, 381 Graves, Robert 101
Chandler, A. D. Jr. 25, 78, 112, 126, 148- Greif, Irene 272
51, 153-5, 159, 161, 171-3, 391
Chinnock, Chris 201 Hamilton, Gary G. 100, 171, 172
Christopher, Martin 299 Hammer, Michael 262, 279, 280, 288, 290,
Clarkson, Mark A. 210 294, 310, 381
Hannan, Michael T. 159
436 Systematic Table of Contents
Harvey, David A. 194 Mishkin, Mortimer 62

Harvold, Trygve 196 Moeller, Mike 310, 312
Hatvany, J. et al. 281 Mok, Fai 194
Havelock, Eric 114, 132, 135, 167, 207 Morgan, Gareth 81, 153, 155, 158, 335,
Heller, Martin 221, 258 338
Hesselholt, N. 345 Morris, Desmond 79, 90, 93, 109, 143, 235
Hochschild, Arlie R. 79, 143, 238 Mumby, Dennis K. 238
Hofer, Myron A. 82, 84, 111 Murra, John W. 102, 103
Hollway, Wendy 153, 236, 245
Hopper, Max D. 305, 306, 340 Nadler, David 42, 44
Hunt, R. Reed 60, 62, 63, 66-8, 71 Nash, Manning 100
Huppes, Tjerk 384 Negroponte, Nicholas 184, 194, 218, 415
Nelson, Theodor H. 208
Jacobson, Agneta 418
Jacobson, Ivar 418 Ong, Walter J. 103, 117, 129, 132-5, 167,
James, Peter 39, 41, 112, 124, 416, 418 207
Johansen, Robert 138, 231, 267, 268, 271,
272, 274 Palme, Jacob 228
Parish, Tom 194
Kahneman, Daniel 69, 70 Pehrson, Robert N. 65
Katzenbach, Jon R. 266 Pine, Joseph B. II 153, 154, 379-80, 383
Keen, Peter G. W. 272 Piore, Michael J. 373, 379, 383, 387-8
Kepner, Charles H. 67 Polanyi, Michael 52, 168, 240
Khandwalla, P. N. 400 Pool, Ithiel de Sola 137-9, 182, 325, 326-7,
King, John L. 101, 106, 325, 344 336, 344, 375
Kitto, H. D. F. 103 Porter, Michael 299
Koolhaas, Jan 153 Psaltis, Demetri 194
Kuhn, Thomas S. 85, 357 Putnam, Linda L. 238
Law, John 110, 196, 346, 383 Quick, J. C. 237

Leach, E. R. 94, 95
Leavitt, Harold J. 325, 326, 344 Rachlin, Howard 70, 110
Lewin, Arie Y. 416 Rasmus, Daniel W. 258
Lie, Merete 287 Rasmussen, Bente 287
Lincoln, James R. 51, 274 Robinson, Mike 225
Lloyd, Peter C. 101 Rockart, John F. 272
Long, Richard J. 63, 224, 230, 231, 259, Ross, Randall R. 42, 237
287, 294, 413 Rowan, Brian 36, 83
Luckmann, Thomas 28, 110, 247 Rowan, Roy 72
Lundberg, Nils H. 288
Lysgaard, Sverre 241 Sabel, Charles F. 17, 373, 379, 383, 387,
388
March, James G. 27, 41, 50, 163-4 Sandars, N. K. 26, 101
Maus, Arne 287 Schmidt, Kjeld 287
McKersie, Robert B. 220, 269 Schoenberg, Robert J. 127, 375
Meyer, John W. 36, 83 Scott, Richard W. 41
Meyer, Marshall W. 30 Sharp, Lauriston 64, 92
Mintzberg, Henry 6, 40-53, 60, 70-2, 76, Short, James E. 272, 413
96-8, 104-5, 157, 160, 163, 164, 297, 326, Shotwell, James T. 89
331, 370, 371-2, 375-7, 384-5, 388-9, 390- Silverman, David 28, 31, 100, 168, 351,
2, 395-8, 401, 403, 410, 417 352
Systematic Table of Contents 437
Simon, Herbert A. 11, 27, 41, 50, 69-70, Weber, Max 156, 163, 262
71-2, 163-4, 217, 224 Weick, Karl E. 28, 30, 62, 67
Slovic, Paul 69, 70
Smith, Douglas K. 68, 266
Weiss, Joseph 72
Sobel, Alan 201 Whisler, Thomas L. 325, 326, 344
Stein, Richard Marlon 218 Wiener, Norbert 186
Stinchcombe, Arthur L. 83 Williamson, Oliver E. 158-9, 171-3
Stix, Gary 310, 311 Wilson, Peter J. 90-1, 92, 93, 115,
Sundstrom, Eric 156, 245 142
Winograd, Terry 225
Thé, Lee 270
Thompson, James D. 96 Yates, Francis A. 116
Thompson, Paul 343, 382 Yates, JoAnne 122, 123, 352, 408,
Thompson, Tom 194 416
Thorpe, Nick 124 Yourdon, Edward 211, 213
Thorud, Geir I. 307
Tregoe, Benjamin B. 67 Zuboff, Shoshana 240-3, 246, 278,
Tushman, Michael 42, 44 296, 297, 319, 322-4, 328, 331, 333,
Tversky, Amos 69, 70 339, 367, 377, 388
Zucker, Lynne G. 30
Walton, Richard E. 220, 229, 241,
258, 269, 333-5, 336
Subject Index
aborigines automatic machines 141–2, 164, 352

of Australia see Yir Yoront automatic teller machines (ATM) 202, 264,
of Cape York Peninsula 64 285
abstraction 240–2 automation 141-2, 153, 155, 165, 168, 170,
action-centered skills 240–1 174, 185, 219, 221–3, 256-7, 275, 312
active file see file action theory and 351-3
active model 354–65 consequences for society 300–2
Adhocracy 47, 49, 96, 97, 98, 99, 162, 362, extent and type 280–3
370, 377 hyperautomation and 298-300
transforming 395–400 organizational truncation and 295-
see also under types 8
Administrative Adhocracy 298, 378 potential 283–6
air traffic controllers 61 routine 277–80
airline reservation systems 17–18, 316–17, shrinking the organization and
339–40, 358, 363 294–5
airplanes 309–12 skill and 323-5
algorithmic decomposition 211 automatization 61
algorithmic strategy 68 automatized control 60
Altair 192 automobiles 309–12
Amadeus 305, 407 see also Ford Motors; General
American Airlines 306, 382 Motors; Nissan
American Hospital Supply Corporation 4 availability of information 319–20
American Springfield Armory 153 Aztecs 101, 103
American System of Manufactures 153,
154, 379 Ballmer, Steve 73
amphitheater 77 bandwidth 54, 59, 75, 223, 226, 268, 271
animated simulation 226 banking 284–6, 295, 309–12
animation 209–10 future of 290–3
Apple II 192 Bantus of Africa 65
Aristotle 135 Benetton 328–9, 347, 393
ARPANET 201 bidirectional hypermedia links 208
artificial intelligence (AI) 217, 221, 258–9 Boeing 777 223, 310–12, 313, 338–9, 357–
artificial memory (AM) 217, 258 8, 363, 366, 399
assisting model 364–5, 389, 412 Bolsheviks 158
astronomy 112 bookkeeping machines 142
AT&T 73, 143 bounded improvements in productivity 256
atomistic model 360 buildings 115–16
attention barrier 60 bulletin board function 227
business process engineering (BPR) 14, competence, span of 257–60, 262, 279, 381
376 complexity 54, 149, 240-2, 421
By-product Form 391 computers and 185, 191, 214, 220-
2, 261, 319-25
calculators, mechanical 120, 142 decision making and 70
canned actions 141, 352 organization and 48, 50, 104, 156-
capacity for work 53–4 62, 167-70, 325-6, 338-42
Carty, General John J. 182 society and 237
causal relationships 322 writing and 117, 156
cell automation 256–7 computer(s), computer-based systems
cellular radio 127 References are too numerous to list. The
centralization 325–6 reader is advised to use the Table of
automation and 295-8 Contents or to look in the Index for terms
vs decentralization 44 referring to the desired context.
differentiated 392–5 computer-aided design (CAD) 256, 269,
by hyperautomation and 338–9, 357, 363, 380, 399
elimination 331–2 CAD/CAE system 310
by informating 328–31 computer conferencing 202, 268–9
possibilities for 326–36 computer modeling 209, 321
by remote control 332–6 computer programs, status of, in
telephone and 137, 326-8 organizations 351–3
Ceramico 40, 331–2 computer-supported cooperative work
chaos theory 220 (CSCW) 272, 273, 287, 294
chemometry 288 computerate paradigm 357
Cicero 114 Computing-Tabulating-Recording
Clarke, Arthur C. 190 Company (CTR) 123
coaxing strategies 418 concentration camps 35
collective accumulation of knowledge 26, concentration of information 140
27 conceptual model 163, 167-70, 354-65,
communication 41-6, 51, 53-4, 74-8, 114, 421, 424
123-7, 136–41, 144, 197–206, 223–31, 271 Condit, Philip M. 311
ad hoc 46 conferencing 138, 202, 203, 226–31, 249,
Adhocracies and 397-8 268-9, 271, 273
centralization and 326-8 Conglomerate Form 391
databases and 304-9 constructible space 33–4, 175, 415
group work and 272-4 boundaries of 34–6
in computer-based systems 14 biological characteristics 34, 36–7
input and output 198–201 coupled organizations 315–16
I/O bottlenecks 223–6 extensions to 261–71, 272–6, 294–
model mediated 356-9, 362-4 302, 309–16, 325–47
new channels 226–31 local 175
range of 136 psychological characteristics 35,
routine 46 37
simple organizations and 99-107 single organization 309–15
social isolation and 245-6 social and cultural factors 35, 37–8
speed of 136 tools and methods 35, 38
virtual organization and 246-9 contingency theory 32, 55
communication bandwidth 54, 252 control 344–7
communication bottleneck 74–6, 397–8 Cooperative Computing 266
communication range 54, 76-8 coordinating mechanisms 45, 414
communist paradigm 86–7, 104 in simple organizations 98
programmed 52-53 vs centralization 44

real-time 50-2 by de-specialization 337
taxonomy of 47-53, 165-7, 315, by increasing the depth of control
330, 361 337–42
conceptual models and 356-70 by information availability 337
coordination possibilities for 336–42
as a basic element in organization decisions, mechanisms for reaching 46
structuring 42-6 definition of organization 45
by feedback 50, 163 delegation of authority 99, 107
need for 42, 43-5 Dell 394
in organizations based on pre- depth of control 329
computer technology 149-62 design, significance of 244–5
in organizations using computers de-specialization 262–5, 337, 381
309-18, 333, 337-41, 356-65 digital money 292
see also the various configurations diplomacy 327
by plan (program) 50, 163, 164 direct control, extending 371–2
in simple (preliterate) direct supervision 47, 48, 50, 52, 96, 100,
organizations 97-107 163, 164, 166, 170, 177, 330, 361, 371-2
automation and 141 directing work 98, 164, 354
bureaucratic organization and 162- displays 201, 203, 209, 296
7 distance education 228–9
“buzzword” organizations and distribution networks 137
413-4 division of labor 40, 42–3, 45
conceptual models and 356-65 divisionalization 48, 105, 160–1, 390-2,
culture and 174 396
databases and 217, 289, 303-18 Divisionalized Adhocracy 396, 397
explicit design and 167-70 Divisionalized Form 48, 104-6, 160, 162,
groupware and 272-4 175, 390–5, 397, 411, 412
our communication bottleneck and domestication 93–5
74-6 du Pont, Pierre 159, 160, 390–1
telephone and 137 Durant, William C. 159, 160, 391
costs 421–5 Dynamic Extended Enterprise 379, 383
Councilcracy 98, 99, 147
couriers 123–5 economic action, theories of 171
cross-referencing 122 EDIFACT 307
cross-unit committees 44 EDVAC 211
customs systems, electronic 198 electronic data interchange (EDI) 205,
cyberspace 218 307–8
Electronic Fund Transfer at Point of Sale
Daily Planner 334–5, 359, 360 (EFTPOS) 285
database 191, 195, 210 electronic mail 137, 197, 200, 201–2, 220,
capacity 305 225–9, 273
common 204, 269 electronic meetings 268–9
multiple 306–9 see also videoconferencing
reach 304–5 electronic white boards 268
speed 305–6 Elkem 249
structured 196, 303–6 embedded knowledge 259–60, 328, 337
text 196 emotion-controlling technology 143
groupware and 273 emotions 54, 78–82, 144
organizations and 309-17 augmentation by technology 142–3
decentralization 45, 313–14, 326
barriers to virtual organizations Flexible Standardization 378, 383

246–9 Ford, Henry 28, 153, 154, 156
information technology and 239– Ford Motors 28, 289
50 accounts payable system 293–4,
organization effects on 236–9 356, 366, 376–7
organizational constraints and River Rouge plant 406
238–9 Fordist systems 154
encapsulation (of information, complexity) frames 29
102, 107, 151, 159, 161, 164, 213, 214 Frederick the Great 106, 155
Engelbarth, Douglas 267 Fujitsu 281
ENIAC 184, 189, 191, 200, 209 functional decomposition 211
entities 211 functional specialization 149, 152-7, 158,
Entrepreneurial Adhocracy 396 211, 257-8, 262-3
Entrepreneurial Organization 49
Epic of Gilgamesh 26 Galileo 112, 305
Equal Partnership 18, 407 Gates, Bill 73
equifinality 173 gender 248
ESPRIT 292 Geneen, Harold 28, 127, 375
European Central Bank 292 General Motors 159, 384, 391, 392
explicit design 167-9 general systems theory 41–2
explicit model(ing) 165, 168, 169 grapholect 129
explicit routines 165, 278, 315, 335, 376, graphs, graphics 75, 140, 220, 225-6, 256,
384 268, 320
explicit skills 53, 167 Greek creation myths 101
externalization of information processing grouping of task 42
186, 232 group work 129, 130, 138, 265–71, 272,
Eyde, Sam 29 276
groupware 266, 267, 269–70, 272, 274,
face-to-face contact 140, 247, 267, 269, 275, 294, 362, 376, 398
271 Adhocracies and 398
family as organization 100 Machine Bureaucracies and 376
family business 149 gut feeling 73, 74
fax machine 200, 201
feeling 78–82 H&M (Hennes & Mauritz) 329, 393
Feldreich, Bengt 72 Hadza 92
Feudal Form 105 HÅG 394 (accent on A - Norwegian Å - in
feudal organization 95, 101, 102, 103-5, Norwegian, last letter of the alphabeth!)
116, 151 handwriting recognition 199
divisionalization as a descendant Hanseatic League 119
of 160-1 health care systems 340–2
Fields, Debbie 229, 333–5, 359, 360, 364, heuristic strategy 68
371 Hitler, Adolf 344, 345
flexible automation 366 Hollerith, Herman 122–3
file 46, 128, 157, 163, 165-6, 191 holographic displays 209
active 118, 163, 165-6, 204 Homer
compared to database 195-6, 210- Iliad 89, 101, 132, 135
11 Odyssey 89, 101, 132, 135
see also database and records Horty, Professor John F. 196
Flexible Bureaucracy 378, 383-4, 388, 397, hot spots 343, 344
402, 405, 414 hunter/gatherers, present-day 90–3
Flexible Specialization 373, 379, 383
see also preliterate societies; Yir see the various configurations

Yoront pre-computer technology and 129-
Hurd, Cuthbert C. 86 35
hydroelectric power 112 as precondition for organizing 46,
hyperautomation 298–300, 315, 316, 324, 53-4
331-2, 361, 372, 377, 382, 387, 394, 405, satisficing 69–71
422 in simple (preliterate)
hypermedia 207–9 organizations 99, 102, 105, 106
simplification 69–71
IBM 28, 86, 123, 305 196, 306 unconscious processing and
IBM Credit 279–80, 381 intuition 71–3
IDC 181, 183, 184 writing and 129-30, 132-5, 156-7
implicit coordination 118 information sharing 164
conceptual models and 361, 364, information technology References are too
420 numerous to list. The reader is advised to
coupled organizations and 315-16 use the Table of Contents or to
database and 217, 289, 303-4, 308 look in the Index for terms referring to the
decentralization and 337 desired context.
Divisionalized Form and 394 infrastructural support (for work groups)
Flexible Bureaucracy and 382 270–1
Interactive Adhocracy and 399 Innovative Organization see Adhocracy
Meta-Organization and 405 Innuit Eskimos 142
as an expression of mutual Integrated Form 391
adjustment 314–15 integrating managers/departments 44
Organized Cloud and 407, 409 Intel 187, 188
single organizations and 309-13 intellective skills 240–1, 388
writing and 163-6 intelligent agents 219
Incas 102–3, 119 Interactive Adhocracy 17–18, 398–400,
Index Librorum Prohibitorum 121 402, 411
indexing systems 122 internal contracting 150–1
indirect supervision 164 Internal Revenue Service 44
Industrial Revolution 114, 132, 162 Internet 3, 182–3, 197, 202, 210, 217-18,
informating work 322–5, 329, 337-8 226, 247, 266, 282, 286, 291-3, 340, 410,
information concentration 225-6, 320–1, 413-14
329 Iraq, air strike on 78
information dissemination 64, 120-2, 140, ISDN 201, 202
226-31, 408 ITT 28, 127, 375
information presentation 206–7, 225
information processing 41, 42, 65–74, Japanese watchmakers 86
129–35, 142, 174-5, 218-22 Joystick Organization 332, 335, 361, 374–
capability 54 5, 378, 402, 411-12
cordination and 43-5 just-in-time production 306,322, 380, 394
delays of deliberation 73–4
elements in problem solving 66–9 Kachins of Burma 95, 97, 98, 101
externalization of 184-7 kanban system 305, 306
information technology and 184- keyboards 199
91, 218-22, 280 Khan, Genghis 78, 106, 119
maximizing 69–71 khipu 103
in organizations based on pre- kinship 91-4, 100, 102, 116, 118
computer technology 153, 156, 157 knowledge support 262–5
in organizations using computers !Kung San 92, 142
computer memory 188, 191-7, 223

Lapps of Northern Scandinavia 65 emotions and 81
Large Electron-Positron Collider 112 externalization of 114–20
law of arm's length 218 literate mindset and 133-5
law of requisite variety 31, 383 long-term 63
Lenin 158 messengers and 77
liaison devices 44 oral mindset and 133-5
liaison individuals 44 organizational 46
lineage 65, 102-3, 147, 177-8 schemas and 84
linked model 360 short-term 62
literacy 114, 120–2, 131, 135, 147-8, 154, simple (preliterate) organization
156, 167 and 99-107
literate mindset 133, 144, 167, 178, 240 writing and 127-9, 133-5, 165
literate paradigm 170, 351, 355, 357 mental sets 83-7, 93, 138, 207
logarithms 119 messaging 199, 202, 204-5, 307-9, 316
long-term memory 63 Meta-Organization 370, 395, 400–7, 409,
Luria, A.R. 132 411-12
Luther, Martin 121 Metro Tel 242, 333
microelectronics 112
Machine Bureaucracy 48, 49, 52, 156, 162, microfilm 128, 195
168, 261, 297, 298, 335, 343, 372, 373, Microsoft 73
397 Microsoft Windows 181, 183
constraints 170 military
coordination and 166-7 fire control systems 321
culture and 171-5 organization 155–6
flexible bureaucracy 378–84 situation room 130
literate paradigm and 168-70 mission control centers 130
as the modern organization 154-7 Missionary Organization 49
perfecting 375–84 mnemones, civic office of 116
staying within tradition 376–7 mnemonics 114, 115, 116
truncation 377–8 model-driven organization 16, 356-65,
Mafia 35 370, 424, 425
mail services 123–5, 136 model precision 366–7
mailing lists 141 requirements for 366–8
Manchester Mark I 185, 192 skill and effort 367–8
market-based theories 171–2 see also new configurations under
mass literacy 114, 120-2, 131, 154 names
matrix organization 43, 44, 396 modeling systems 209, 321
Mayas 103 models, typology of 359–65
Mazda 289–90, 294, 310, 312 see also under names
McDonald's 238, 335 Modern Organization 152-70, 375
McLuhan, Marshall 336, 375 modes of rationality 29, 172
mediating model 361–4, 399, 408, 409, moral philosophy, development of 113
410, 411 multifinality 173
Interactive Adhocracy and 399 multimedia 187, 189, 209–10, 235, 266,
Organized Cloud and 408-10 283
meeting support 267–9 mutual adjustment 47, 50–2, 98-100,
Memex 208 149,159, 163, 165-6, 170, 397-400
memory 34, 37, 61–5, 127–9, 144, 215–18 implicit coordination as an
artificial 217, 258 expression of 314-15
as basic precondition 54
mediating model and 362-4, 397-

400 palmtop computers 189
myths and legends 89–90, 101 Pandam 91, 92, 97
paper-based information 128, 211, 286,
Naiken 92 311, 314
NASA Apollo project 50 constraint of 157
network organizations 316, 401, 414 paradigm 85
networking 14, 266, 362 paradigm shift 168, 356
Neumann, John von 211, 212 partnership 147, 149
New Competition 383 passive model 354–7
Newcomen, Thomas 112 Pasteur, Louis 112
newspapers 122, 126, 131, 136, 154, 281, pattern manager 368, 425
396 patterns of action 30-2, 39, 52, 168, 351–
Nielsen, A.C. 304 3, 355-6, 381, 423
Nissan 299–300, 310, 312, 332, 360, 364, patterns of logic 351–3
380, 388, 401, 404, 406 Pennsylvania Railroad 160–1, 390
nonliterate see preliterate Pentium chip 187, 188
Norse religious myths 101 Philips 384
Norsk Hydro 29, 393 Picasso 110–11
numerical information 127, 187, 225–6, pictorial information 128, 224-5
245, 256, 269 pin production (“trade of the pin maker”)
152-3
object-oriented decomposition 213–14 Plato 79–80, 117, 135
office automation, debate on 286–8 Political Organization 49
Olson, Larry 312 polling, sampled 131
Operative Adhocracy 36, 378 pornography 143, 235
oral languages 129 power, migration of 342–4
oral mindset 132–3, 135, 167, 240 power structure 28, 46
oral paradigm 168, 351, 357 Preece, Sir William 139
oral society 116, 132–4, 153 preliterate organization 89, 99–107
organism–organ–cell hierarchy 212 basic principles 106–7
Organization Design Managers 368, 425 circumventing the barrier of
organization structure 41, 45, 173, 272, cognitive capacity 101–3
355, 372, 413, 421 feudal type organization 103–5
basic elements in 42-6 military organization 105–6
size and 39-41 organization domains and their
tools and 25-7, 33-8 structuring 100–1
see also coordination see also under names
see also the various configurations preliterate (nonliterate) societies 64–5, 98,
organizational memory 46 99-107
organizational truncation 295–8, 300, 374, primal space 33, 34, 95
377-8 printing 114, 120–2, 131, 141, 148, 200,
organizations as systems 30–2 281
Organized Cloud 370, 401, 402, 407–10, printing press 121–2
411 problem domain 168-70, 213, 289, 325,
organizing mode 41 355-56, 364, 366, 417, 421
Orwell, George: 1984 345 problem solving 66–9
OS/2 183 generating hypotheses about
overpopulation 67 solutions 68
oversocialized theories 172 procuring information 66–7
Oyama, General 182, 327
testing and evaluating the solutions Russian empire, collapse of 104

69
understanding the problem 67–8 SABRE reservation systems 305, 306, 407
process orientation 19–20 SAGE computers 204, 328
processing see information processing satellite telephone 127
processing power (computer) 187–90 schema 84-5
Professional Bureaucracies 49, 384–90, Scientific American 137, 138, 187, 193
412 screen-fax 201
program construction 353, 355 screen sharing 227, 268-9
programmed routines 278, 335, 361, 372, see also conferencing;
381 videoconferencing
programs (computer) 184, 185-7, 199, 210, self-service 264–5, 281, 282, 283, 310,
211-14, 222, 278 312, 378, 387, 420
embedded rules in 259-60 sensory registers 62
cost of 423 sets see mental sets
future of 190-1 Shan 95
modeling and 354-7 shared external memory 118
status of in organizations 351-3 shareholder democracy 131
see also application types and short-term memory 62
areas Siemens 333, 336
Prudential Insurance Company 304 Simonides 114
psychopharmaca 143 Simple Structure 47, 96–9, 106, 147, 150,
punch card equipment 122–3, 128–9, 142 153, 162
putting out (contracting out) method 150 computer-based transformation of
370–5, 384
quantitative information 75, 186, 219, 220– simplest structure 96
1, 232, 281, 303, 319-22, 328, 366 situation room, military 130
Sloan, Alfred P. Jr 159-60, 161, 213, 391
radio 126 small arms manufacture, US 152–3
railroads 114, 125–7, 137 smart cards 292
administration and 155 Smith, Adam 152, 153
divisionalization in 160–1 social isolation 240, 245–6
RAMAC 196 Socrates 135
reading 75, 132, 134 software see programs
screens and 201, 210, 281 Soviet Union 87, 158
records, organization of 103, 116, 117-19, space and time, constraints of 76–8
122–3, 165, 195-6, 217, 303 space of constructible organizations see
see also database and file constructible space
regulating model 360–2, 365, 410, 411–12 span of competence 257–60, 262, 279, 381,
Related Product Form 391 412
relentlessness (of computer systems) 242–5 span of control 105
response assistance 260 specialization (of work) 106, 150, 151-7,
retrieval of information see records and 163, 174, 257-8, 261, 263
database speech recognition 199
return channels 136 Stalin 345
Richman, Tom 334 standardization, constraints of 170
rigidity (of computer systems) 242–5 standardization of outputs 48, 52, 105
role conflicts 241–2 standardization of skills 48, 49, 52, 98,
Roman empire 119 165, 166, 260
routine jobs, elimination of 261–2, 301, standardization of work 48–9, 52, 98, 164,
324 165, 170, 278, 361
steam engines 112 see also under types and

storage 54, 122-3, 157, 184, 191–7, 207, technologies
210, 216-18 transaction costs 171
trends 193–4 transistor 112, 188, 193
see also database and records transportation (of goods/people) 123, 125,
structural linking 44 136-7, 197-8
structured decomposition 211 travel time 124, 125, 126, 136
structuring information 210–14 Trevithick, Richard 112
functional approach 211-12 Turing test 352
object-oriented approach 212–14 TVINN 198
sudden insight 71
Supplier Cluster 406 UML (Universal Modeling Language) 355,
SWIFT 308 418
Swiss watchmakers 85–6 undersocialized theories 171–2
synthetic organizations 96 Unilever 384
system auditors 259 UNIVAC 192, 304
system-supported skills 260, 262, 364, 365, Universal Machine 186
367
system-supported supervision 330–1, 332, verbal information 74-6, 117-18, 224–5,
335, 361, 372, 374, 393 228, 231, 328
system-to-system communication 203–6, videoconferencing 138, 230–1, 233, 268,
306–9 271, 273
videophones (picturephone) 138, 202–3,
Tabulating Machine Company 123 229–31, 233, 268, 271, 273
tacit knowledge 52, 98, 168, 240, 263, 337 vidwams 271, 273
tacit skills 52, 166 Vinge, Vernon 190
task-continuous hierarchy 149, 153 virtual organizations 246-7, 271, 316, 401,
task elimination 288–90 413
taxonomy of coordinating mechanisms 47– emotional barriers to 246–9
53, 165, 298 voice mail 203, 229, 334
classes of coordinating voice-recognition technology 75
mechanisms 50-3 vubicles 271
figures 51, 165, 315, 330, 361
Taylor, Frederick W. 154, 156, 158 Wagner, Rudolf Clemens 346–7
Taylor system 158 Wang 346
teams 265–71 Watson, Thomas, Sr 28, 86
technical space 34 Watson, Tom 28, 86
telegraphs 125–7, 137 Watt, James 112
telephones 125–7, 137-9, 166, 182, 197, Work Force Supervisory System (WFSS)
202–3, 223, 229–31, 325–6, 336, 346 242–4, 245, 333, 335
conferencing and 138, 202-3, 268 work support 269–70
centralization/decentralization and workflow tools 270, 376
137, 325, 326-8, 336 working memory 54, 62-3, 65, 71, 129,
effect on jobs 138 135, 209, 218-19, 320
future of 203 World Wide Web 207, 208, 209, 235
in warfare 182 writing 75, 113, 114, 139, 156
teleworking 150 as an administrative technology
Textron 391 117–20
theories-in-use 52, 84, 168, 169, 240 and communication 123-5
Thomson, Edgar J. 161, 390 importance of numerals 119–20
tools 27, 35, 38, 109–14 and information processing 129-32
invention of 115
and the literate mindset 132-35 Xanadu hypermedia 208, 209
and memory 127-8 XCEL 258
and organization 148, 153-4, 153, XCON system 258
156, 167-8 Xerox PARC 210
in printing and mass literacy 120-2
significance of a shared memory yield management 340
118–19 Yir Yoront 92, 93, 115, 134
written records see records

Future Organizational Design by Lars Groth - Creative Commons Edition 1.1

Uploaded by

Copyright:

Available Formats

Future Organizational Design by Lars Groth - Creative Commons Edition 1.1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Future Organizational Design by Lars Groth - Creative Commons Edition 1.1

Uploaded by

Copyright:

Available Formats

FUTURE

As originally published by John Wiley & Sons, Ltd.

Electronic Edition v. 1.1

In order to view this electronic edition with correct left-right

Future Organizational Design

is licensed under a Creative Commons

This means that:

For the full license, please go to http://creativecommons.org/licenses/by-nc-

This is an electronic edition of the book

“Future Organizational Design

The author can be contacted at [email protected].

“Man is not the sum of what he has already,

I A PLATFORM FOR THE INVESTIGATION 1

... and to Use Tools 27

3 THE BASIC PRECONDITIONS FOR ORGANIZING 39

II INDIVIDUAL CAPACITY AND ORGANIZATION BEFORE THE

6 THE POWER OF TECHNOLOGY 109

III IT AND THE PRECONDITIONS FOR ORGANIZING 177

Three Basic Characteristics 184

9 THE IT-BASED PRECONDITIONS 215

IV EXTENDING THE SPACE OF CONSTRUCTIBLE ORGANIZATIONS 251

11 THE INDIVIDUAL AND THE GROUP 255

On the Fringes of Organization 316

V THE NEW ORGANIZATIONS 349

TRANSFORMING ADHOCRACY 395

"Sometimes a scream is better than a thesis."

The Present Volume

heretofore were unimaginable. New structural configurations, made possible

Wiley Series in Information

RICHARD BOLAND Department of Management Information and Decision

NIELS BJORN-ANDERSEN Copenhagen Business School, Denmark

"Gratitude is the poor man's payment.”

he heard about my doctoral work, he expressed interest and offered to read my

“A man’s behavior is the index of the man, and his discourse is

Sustained Success Grows from Knowing Why

A Quest for Practical Directions

If I could answer these questions, I felt I would also be in a much better

tell if their particular organizations could benefit from intensive use of

Venturing Beyond Prescriptions

our mores. We have literally millennia of experience with organizations, with

Table 1-2: Overview: The basic contribution of pre-computer tools, the

building and how they constrain us in establishing and maintaining

Table 1-3: Overview: The basic contribution of information technology, the

Organization—A Human Endeavor

division of tasks, hierarchical supervision, detailed rules, precision, clarity and

The Contributions of Information Technology

human cognition and of the distinctive properties of information technology. In

The Joystick Organization

The Flexible Bureaucracy

The Interactive Adhocracy

The Organized Cloud

From Analysis to Action

A Key to this Book

Part I: A Platform for the Investigation

Part II: Individual Capacity and Organization before the Computer

Part III: IT and the Preconditions for Organizing

Part IV: Extending the Space of Constructible Organizations

Part V: The New Organizations

2 Organization and Tools—the

“Man is a tool-using animal . . . Without tools he is nothing, with