# Statistician Mathematically Predicts Which Game of Thrones Characters Will Die in Last Two Novels

Thursday, 25 September 2014 - 2:52PM
Thursday, 25 September 2014 - 2:52PM

Fans of the Game of Thrones novels will be waiting on tenterhooks until George R.R. Martin finishes the last two books of the Song of Ice and Fire series, the first of which will be called The Winds of Winter and will be released in 2015 at the very earliest. Martin is known for ruthlessly killing off beloved characters, and left several of the characters' plotlines on cliffhangers in the previous installment, which will make the wait all the more excruciating. Luckily, a statistician from the University of Canterbury has performed a Bayesian analysis in order to estimate how many POV chapters each character will have in the next two books, and therefore how likely they are to be killed off.

Here's the abstract of the paper, which awesomely includes a spoiler alert: "Predictions are made for the number of chapters told from the point of view of each character in the next two novels in George R. R. Martin's A Song of Ice and Fire series by fitting a random effects model to a matrix of point-of-view chapters in the earlier novels using Bayesian methods. SPOILER WARNING: readers who have not read all five existing novels in the series should not read further, as major plot points will be spoiled, starting with Table 1."

[Credit: Richard Vale]

[Credit: Richard Vale]

"How many chapters will each character get in 'Winds of Winter'? The height of the bar over each number represents the probability of a character getting that number of chapters. If the bar reaches the '0.5' mark, the chance is 50 percent. If it reaches '1' the model is certain of that number of chapters."

Vale loosely correlated the probability of having zero POV chapters with the probability of death, although he acknowledged that this was a not a perfect correlation, by any means:

[Credit: Richard Vale]

"One of the most compelling aspects of the Song of Ice and Fire series is that major characters are frequently and unexpectedly killed oﬀ. The probability of a character having zero POV chapters... is therefore of interest... Note that although a character who has been killed oﬀ will have zero POV chapters, the converse is not necessarily true."

In his overall analysis, he concludes that (spoiler!) readers can reasonably expect to see characters who have featured prominently in previous books, such as Arya and Daenerys, to appear in the next two books, while characters like Jaime or Davos might not be so lucky.

He freely admits that there are several issues inherent to his model. He does not take into account the plots of the novels, only the distribution of chapter POV in previous books. So in the case of Jon Snow, he predicts that the character survived based purely on a statistical analysis of the previous books, without integrating the actual cliffhanger into the analysis. He also does not account for characters who have not had POV chapters in other novels, or the appearance of entirely new characters. There also may not be enough data to perform a statistical analysis in an optimal manner.

Ultimately, he concludes what we all do when we make predictions: that they will either be disproven or vindicated when the books are published: "We could continue to make predictions in the hope of getting one right, but there is no merit in this. We hope that it will be possible to review the model's performance following the publication of [book] six."