This Algorithm Claims It Can Pick Bestselling Novels

For the last few years, we've been waiting for the algorithm that can pick bestselling novels, and now it might have arrived, albeit with a questionable first prediction. Jodie Archer and Matthew Jockers, the authors of the forthcoming The Bestseller Code, have built a program that "looks at themes, plot, character, setting, and also the frequencies of tiny but significant markers of style" to predict how well a manuscript will sell.

But the bestseller-identifying algorithm picked Dave Eggers' under-performing 10th novel, The Circle, as "the single most paradigmatic bestseller of the past 30 years," leading some to criticize its effectiveness. There's also an element of irony in the selection. Eggers' pseudo-Luddite novel centers on a tech employee working for a company that plans to capitalize on mass-publicization of data, including day-to-day interactions, words, and thoughts.

Even if Archer and Jockers' program mucked up this time, it's not impossible to imagine that an algorithm might predict bestselling novels someday soon. A tool created by scientists at the University of Antwerp successfully chose the hit dance songs of 2015. Music is much more mathematical — and therefore, perhaps, easier to predict — than prose and poetry, but the current system of manuscript readers and slush piles might soon be a thing of the past.