An Algorithm Can Quickly Find the Killer in an Agatha Christie Mystery

By Andrew Liszewski on at

She’s still widely regarded as one of the greatest mystery writers ever, but coming up on what would be her 125th birthday, a group of academics has managed to create an algorithm that can accurately predict which character will turn out to be the killer in an Agatha Christie novel.

A panel of academics, including data analysts and researchers from Queen’s University in Belfast, analysed 27 of the 83 novels published by Agatha Christie during her lifetime and discovered that the story’s location, modes of transport, and the nature of the victim’s death were all important clues to solving the mystery.

It was also discovered that how the culprit was introduced served as a key clue to the mystery. When the killer was a female, Christie tended to use a more negative sentiment when introducing the character. But when the killer was male, the word patterns used to mention or address the character were more neutral or positive.

Even where and how the victim was murdered revealed patterns into who the killer was. If they were strangled, 75 per cent of the time it was by a man. But if they were murdered out in the country, there was a 75 per cent chance a woman was responsible for the act. And it was these patterns in the data that allowed the panel to develop an algorithm that could, and with a surprising degree of accuracy, solve many of Agatha Christie’s mysteries.

That being revealed, it’s still far more enjoyable to just read the books and try to solve them yourself. [Guardian]