Science Has Discovered Whether Your Book Was Worth Writing


What would you do if you knew you wouldn't fail?

If you're a writer, you might soon be able to find out — quite realistically — if you can "publish a bestseller." Recently, researchers from Stony Brook University claimed that they had discovered the algorithm that will predict the success of a book. Their study, according to assistant professor Yejin Choi, is the first of its kind.

Image Credit: Brew Books

The project was based on writing style, involving such variables as parts of speech and the amount of "extreme" and "negative" words in each sample. Choi and her team selected works from a variety of genres and measured success by the number of downloads a book had received from the Project Gutenberg website. She claims that she can predict literary success with 84% accuracy.

Theoretically, writers could use this study to tailor their work to those styles that have been most successful. Choi also suggests that the algorithm could help publishers go through slush piles, a strategy that would quickly and easily separate better manuscripts from less successful ones. "I would expect so, especially with all the plethora of published books and unpublished manuscripts that they would have access to, one could really build good [sic] statistical models than what we had in this study," she told the Guardian.

Image Credit: Steve Greer

But while the study and its applications seem useful — and even accurate — its goals are fundamentally wrong. It predicts success — but not the kind that writers or readers want.

A project like this assumes, first, that readers want a singular, streamlined type of writing. While there are trends among bestsellers, there are always breakout books that do something different from their predecessors — or do the same thing other writers have done, but better. Jennifer Egan incorporated a PowerPoint presentation into her bestselling novel, A Visit from the Goon Squad. Would an algorithm have rejected her brilliant but wholly unconventional manuscript if it were in a pile with more familiar styles?

Image Credit: Anchor

Widespread use of an algorithm like this one also removes the most important aspect of writing for writers, publishers, and readers alike: it's personal. While they might theoretically jump at the chance to reduce their slush piles, publishers want to stand behind the books they choose, having been personally affected by what they read. Sarah Savitt, an editor at Faber & Faber, said, "It sounds like a fascinating academic experiment but for me choosing books for publication is such a personal process, and I can't imagine applying this algorithm to my reading pile."

At its core, success should be measured by whether a writer moves a reader, changes a heart, opens a mind or earns a spot on a well-loved bookshelf.

The composition process is deeply personal for writers themselves, and many just don't write with monetary or conventional success in mind. Of course, there are some who see trends in literature and write to them, but there is a large number who simply fell in love with storytelling or poetry, and writing what they personally need to write is success enough. Plenty want to publish as an extension of that, but few would want to achieve success by having their manuscript mechanically chosen. It seems more likely that writers would want their work chosen by an editor who was moved by it, not by a computer that identified it as having more of the "correct" types of words.

Ultimately, this study misidentifies what success means to writers. At its core, success should be measured by whether a writer moves a reader, changes a heart, opens a mind or earns a spot on a well-loved bookshelf.

The key to that kind of success is simpler than any algorithm could understand: living in the world and recording it meaningfully.