Forget about advanced artificial intelligence killing off humanity, we have more immediate concerns! Like, say, dodging machine-generated Game of Thrones spoilers. Thanks to Milan Janosov at the Central European University, a future could exist where you're having to warn your toaster off revealing the next death in the popular (and bloody) TV series.
Before we continue, this post contains, uh, predicted spoilers for the show. Who knows how accurate they are, but it's worth pointing out now if you're worried.
OK, so maybe appliance spoilers are a little far-fetched, but Janosov has managed to generate a list containing death probabilities for the show's vast array of characters using subtitles of all things.
For this, he was able to build an "aggregated network of the realm's social system":
In this network each node represents a character of the story and the weight of the link between each pair of characters symbolises the strength of their social interaction. I considered scenes to be the elementary units of the social interaction (an average episode contains about twenty of them) ... In other words, scenes are complete graphs, or cliques, increasing the tie strength between all pairs of people present by one. By calculating these scene-level complete networks and then aggregating them, we arrive to the global social network of Westeros.
Yes, you read that correctly, he built a social network for Game of Thrones.
So, how can this be used to predict future deaths? As Janosov explains, we know who's already died (61 characters apparently), so we can use their social ties and conversations to look for similarities and consistencies.
Without further ado, here's the list he came up with.
It's not bulletproof — we're talking about story and narrative here, which has the potential to defy any sort of prediction — but it's an interesting look at what you can do with enough time, data and computing power.