Ramakrishna Akella’s Post

Matrices … With the AI boom and, for wireless folks, the MIMO boom - understanding and handling matrices (sparsity, quantization, conditioning, rank, eigenvalues and eigenvectors, algorithms to invert, algorithms to decompose etc.) is an important skill to have. Of equal importance, as one starts to lead technical and product teams, is understanding the Eisenhower matrix. Most specifically, the ability to ignore the “urgency trap” (and especially the urgent but not important quadrant) and to learn to dwell (time and attention) in the second quadrant (important but not urgent) is an extremely important skill. It is, like many good things in life, easy to state and hard to follow. And yet, it is the path to real success. Leads who stay in the “urgent quadrants” exhaust themselves and their teams. One can cry wolf only so many times. Or to put it differently, per the popular cliche, if everything is important, then nothing is important. A second matrix that is important is the 2x2 matrix of systems. {simple, complex} X {small, big}. Here ones need to avoid the “complexity trap”. One must avoid building complex-small systems like the plague. The expected value is certainly negative. But the real interesting one to understand is the “big” systems column. One might think, rightly in my opinion, that a system like Starlink is big. But is it also complex? Does it have to be complex? What does complexity even mean and what is the threshold. Here is where you differentiate the visionaries from the wannabes. The best systems - in terms of impact, satisfaction and expected value - are sometimes the big systems. Sometimes big systems can come from simple small systems that grow organically (AWS as an example) and sometimes they come whole (GPS, Qualcomm CDMA, Starlink etc.). In the latter case, the biggest and best skill that the original and visionary leads bring to the table is to keep the “bigness” of the bold vision and then relentless simplify everything else. To do this they are clearly connected to the vision, the system definition, its execution and to the teams. After the system is defined, the leads plant less and prune way more. They act less like bosses (who tend to be invisible and operate by diktat) and more like gardeners who are pulling the weeds constantly (hence sitting in the sun, with a sun hat and visible). That is the only way to pull off a simple-big system. So if you are in a project where the leads are not constantly simplifying and path-clearing for their teams, raise you hand and ask for simplification. And in anything you have agency and where you have control, in the words of a currently highly visible lead, “delete a part, process or step” with great energy. Big systems are worth building. They can be built simply. Indeed if one wants to build a big system, one can mostly likely only build its simplest version at the beginning. The condition number of your product is dependent on which quadrant it is occupying.