A.I.-Generated Map Predicts Who Will Die Next in 'Game of Thrones'

Central European University/ Milan Janosov

According to sophisticated machine-learning calculations, the Queen of Dragons will soon meet her demise.

In early July, a computer scientist at Central European University’s Center for Network Science, Milan Janosov, predicted the fates of the Game of Thrones characters that have yet to be impaled, poisoned, or have their brains bludgeoned. He quantified their social significance on the show, and then entered these numbers into a machine-learning algorithm.

To create his algorithm, Janosov first built a network of the realm’s social system based on how often characters interacted with each other. He pulled this data from the show’s subtitles, which provided nearly 600 scenes worth of information about how often characters appeared in the same scene. This data illustrated which characters had the strongest ties to other characters, as shown in the realm’s social network below. As expected, the Starks and the Lannisters take their rightful place at the central nexus of deception, alliance, and lust.

Milan Janosov/Central European University

After calculating which characters are the most important, Janosov then related “network position to survival.” Of the 94 characters he considered, 61 had already died, providing a wealth of data about the social position and level of importance of the deceased. This allowed Janosov to calculate their fates, as he described in his paper:

Based on this knowledge we can form an educated guess of who is going to die in the near future in the following way: which of those people still alive have similar features to those who have already passed away?

After entering this data into a machine-learning algorithm, the model accurately predicted the fates of about 75 percent of the 94 characters that were considered. Though it generally performed well, the model incorrectly predicted that eight characters would live, but they actually died, like Margaery Tyrell, who perished amidst a shocking, explosive wildfire event that took hundreds of lives. Janosov notes that incorporating more data, like gender and membership in a noble house, could increase the accuracy of the predictions.

But for the characters left standing, the machine gave its best estimates for their probabilities of death. Daenerys Targaryen, the Queen of Dragons, (who has three dragons, a horde of 100,000 Dothraki fighters, 8,000 elite eunuch warriors in the Unsullied, and alliances with Dorne and the Tyrells) has a 91 percent chance of death.

Grey Worm, one of the realm’s most skilled soldiers and commander of the Unsullied, doesn’t fare much better, with a 90 percent chance of death.

Jorah Mormont has just a five percent chance of death, according to the algorithm. The Queen of Dragons, Daenerys Targaryen, has an almost certain death, at 91 percent. 

Flickr / Peter23394

In brighter news, the algorithm deduced that Jorah Mormont, the vicious swordsman who once fought for and guarded Daenerys Targaryen, has just a five percent chance of death, even though he is in the later stages of the grim greyscale disease. Janosov’s intelligent program made this prediction before the onset of Season 7, when Samwell Tarly discovered a potential cure for the ghastly skin malady and began peeling the reptilian, pus-laden scales off of the tortured knight.

Theon Greyjoy, like Jorah, has just a five percent chance of death. After years of torture, paternal abuse, and having his genitals cut off, the machine might be showing Theon a bit of mercy. Or not.

Here’s Janosov’s final list:

Central European University/ Milan Janosov

Related Tags