Writers who fear that computers will someday displace them may shudder to learn that the machines won’t just write the books, they will read them too.
In recent months, both researchers and literary critics are harnessing computational power to “read” books in an effort to divine qualities human writers and readers haven’t the bandwidth to discover themselves (“The Taxonomy of Titles in the 18th Century Literary Marketplace” anyone? Anyone? Bueller?).
Among them are a trio of computer scientists at New York’s Stony Brook University who created an algorithm to predict the success of literary styles that boasts an 84 percent rate of accuracy when analyzing previously published works.
“In a paper published by the Association of Computational Linguistics, Vikas Ganjigunte Ashok, Song Feng, and Yejin Choi said the writing style of books was correlated with the success of the book,” writes Live Science contributor Joel N. Shurkin. Using a process called “statistical stylometry” to analyze literary stylings in an array of books across genres, the team identified the “characteristic stylistic elements more common in successful tomes than unsuccessful ones.”
It’s only a matter of time before researchers team up with an agency like Narrative Science, whose artificial intelligence algorithms pair data with “natural language communication” to produce written content, resulting in bestsellers by bots.
But who would read it? Other computers, thanks to Franco Moretti, who founded Stanford’s Literary Lab so that digital dalliances with texts could have a room of their own. His essay collection, “Distant Reading,” recently won the National Book Critics Circle Award for criticism, and he has been lauded by Wired for “his data-centric approach to novels, which he graphs, maps, and charts … if his new methods catch on, they could change the way we look at literary history.”
At first glance, this may make you want to cue up the famous, “Understanding Poetry” scene in “Dead Poets Society,” wherein Robin Williams goads his class to aggressively edit a couple dozen poetry primers of their analytic assumptions.
Actually, Moretti’s work is fascinating – look over his “pamphlet” on “Network Theory, Plot Analysis,” which teems with graphs on the intricate relationships in Hamlet. However, some in lit-crit circles aren’t enthused with Moretti’s critical approach and suggest putting words into a numbers cruncher can only result in damage to both.
“He thinks that literary criticism ought to be a science,” writes Joshua Rothman for the New Yorker’s “Page-Turner” blog. “The basic idea in Moretti’s work is that, if you really want to understand literature, you can’t just read a few books or poems over and over … Instead, you have to work with hundreds or even thousands of texts at a time.”
This is precisely what the algorithms at Amazon and Apple’s respective bookstores do, which can sometimes produce discomfiting results that say more about readers’ proclivities than perhaps we care to know.
For example, we can forgive readers their fascination with fan-fic-turned-softcore, but why the hell is “Mein Kampf” a bestselling ebook? Vocativ contributor Chris Faraone asks, “Is this what happens when Mein Kampf becomes available in the privacy of our own iPads? Could it be a cultural curiosity, much like what’s happened with sleazy romance novels, which surveys show are increasingly consumed in more clandestine e-form?”
What’s almost more chilling are the results of Amazon’s recommendation engine, which aggregates information from millions of purchases in an effort to upsell consumers on additional product. Click-through the titles on the “Customers Who Bought This Item Also Bought …” prompt, for a glimpse into the data-driven future.
So, worse case scenario, Ashok, Feng and Choi’s stylometry breaks down the secret bestseller recipe, Narrative Science bots implement it and, later, Franco Moretti can explain how we became enslaved to a bestselling Nazi computer overlord via our Kindles.
Somewhere Philip K. Dick is crying.
• • •
Daedalus Howell is read by robots at DHowell.com