Skip to main contentSkip to navigationSkip to navigation
The algorithm predicted that Jon Snow had just an 11% likelihood of death, compared to Stannis Baratheon’s 96%.
The algorithm predicted that Jon Snow had just an 11% likelihood of death, compared to Stannis Baratheon’s 96%. Photograph: Helen Sloan/AP
The algorithm predicted that Jon Snow had just an 11% likelihood of death, compared to Stannis Baratheon’s 96%. Photograph: Helen Sloan/AP

Game of Thrones algorithm finds Jon Snow should not have died

This article is more than 8 years old

Students in Munich have crunched data on every Game of Thrones character to predict their likelihood of death – turns out Jon Snow’s was a statistical shocker

  • Spoiler alert! This story is about people being killed off in Game of Thrones

It is 2016, and nothing can escape data analysis. Not even Game of Thrones.

Students at the Technical University of Munich have applied their skills to helping us understand the likelihood of a character’s death over the course of Game of Thrones, the sixth series of which begins on 24 April.

Their project A Song of Ice and Data analyses as much online data as it can find about both the book and the TV series, regularly scraping and updating information from the vast, fan-based Wiki of Ice and Fire, the Game of Thrones wiki, Wikipedia and Twitter. The algorithm categorized each character by more than 20 features including age, title, gender, number of dead relations and their popularity, based on incoming and outgoing links on the Wiki of Ice and Fire.

There are 2,028 characters in the full Game of Thrones world, typically with more than 30 characters in each episode of the TV show and more than twice the number of male characters than female.

Men are more likely to play noble characters, whereas women are more likely to play peasants – but are are also less likely to be killed off. The Munich team said they had developed a machine-learning algorithm to predict the likelihood of death for characters, and found it is 33% for men and 23% for women.

They extended the analysis to the age of characters killed, which only really tails off at the age of 70 and is most likely between 31 and 40. Identifying various other predictors of risks, the tool could then predict which characters were most likely to be, perhaps quite literally, axed. And some have already been axed, not least the beautiful Jon Snow who – and maybe we didn’t need data to tell us this – suffered an unexpected death. He didn’t deserve to die!

Among the findings, ordered by likelihood of death:

Tommen Baratheon - 97%

The boy king currently sitting on the Iron Throne looking slightly vulnerable and very easy to kill.

Stannis Baratheon - 96%

The wannabe king who murdered his own daughter in his desperation to make it to the throne. Not a very nice man, and actually already dead. And deservedly so.

Daenerys Targaryen - 95%

Queen of the Andals and the Rhoynar and the First Men, Khaleesi of the Great Grass Sea, Breaker of Shackles/Chains, Queen of Meereen, Lady Regnant of the Seven Kingdoms, Mother of Dragons. (Friends call her Dany.)

Davos Seaworth - 91%

Former smuggler and Stannis’s right hand man. It might be a shame to kill off Davos, as he’s been doing so well with his reading homework.

Petyr Baelish - 91%

Snakey Petyr Bealish is a master spy whose ruthless ambition has earned him many enemies. Also his accent has changed quite a lot, so maybe kill him off for that?

And as for Jon Snow?

Please oh please Lord of Light – could Jon Snow be, like, not dead? Let’s cling to the hope that producers may have acknowledged his popularity on Twitter, which showed a huge spike of outrage on the night his death was broadcast.

And there is computer-determined reason why, it says here, because their system has determined he only had an 11% likelihood of death. What were they thinking?!

Comments (…)

Sign in or create your Guardian account to join the discussion

Most viewed

Most viewed