Is Jon Snow Dead? Artificial Intelligence Software Says No

Israeli scientist uses JavaScript course on artificial intelligence, machine learning and big data to predict that the Mother of Dragons will win the Game of Thrones, before dying.

Jon Snow dying in the Game of Thrones season finale.
Screenshot from YES Binge

The most important question for Game of Thrones fans belongs to quantum mechanics: Schrödinger’s Jon Snow. Or in other words, is one of the last remnants of the House Stark, Jon Snow, who ended the last season of the hit television show with what seemed to be dozens of stab wounds, dead or alive.

The show’s producers have repeatedly said he is dead, like the parrot in the Monty Python skit, but it seems no one is really convinced and not a week goes by without long articles explaining the miracle of how he is still alive.

The “Jon Lives” camp has received unexpected support recently from an Israeli scientist and students at the Technical University of Munich. Israeli Dr. Guy Yachdav, an expert in bioinformatics in the university’s computer science department, mentored a project as part of a JavaScript course on artificial intelligence, machine learning and big data – and the results were that Jon Snow has only an 11 percent chance of being dead.

Is that good enough for you? After all, the research was based on data taken from the works of George R. R. Martin, and also shows that the Mother of Dragons, Daenerys Targaryen, has a 95 percent chance of dying in the show – but will also win the Game of Thrones.

Davos Seaworth, one of the more heart-warming characters in the series, is estimated to have a 91 percent chance of dying.

A Song of Ice and Data, as the project is called (and it has its own website: https://www.got.show) includes tools that follow all the Twitter feeds about each character, a map that follows their journeys all over Westeros and Essos, the world built by Martin; and is filled with data predicting the future – in the Seven Kingdoms.

The Twitter-following bot also conducts a sensitivity analysis on the tweets about each character. For example, Lord Renly Baratheon, the gay claimant to the throne who is killed in the second season, suddenly shot up in Twitter between the fourth and fifth seasons. Yachdav says he tried to figure out why, and it turns out it happened on the day that the U.S. Supreme Court recognized gay marriage. The person who runs Renly’s Twitter feed tweeted about it, and it went viral – even making the news.

The idea began during Christmas vacation, when Yachdav was looking to fill his free time. “I was sitting bored about a year and a half ago, and started to work on it myself. I started gathering the information and making all the attributes of the characters, but I needed to go back to work, and as part of my doctorate I had to teach this course,” he says. So he decided to continue the project as part of the JavaScript and data course.

The data behind the research is taken from various wiki sites based on the books, which differ somewhat from the television series. Each character has his or her own profile, from the simple facts to more complicated ones like the number of deaths in the families and the various relationships among the characters. An artificial intelligence program they developed, standard algorithms, then tries to predict what will happen to each character.

So far, they have a 74 percent success rate in their predictions, but they will have to wait longer and watch the new season to see how right they are – and who else will lose their head.