On January 8, Inside Science reported that computer scientists at Stony Brook University had designed an algorithm allowing them to determine what makes a novel a success. The results are eerily precise. Among the traits most likely to make a book well reviewed and widely read are an unadorned, journalistic style; higher numbers of nouns and adjectives; and lower numbers of adverbs and verbs.
Thankfully, literature is not a science. Yet the writing and selling of literature increasingly is. Thanks to a proliferation of analytics, it’s easier than ever for publishers to track, graph, and therefore do their desperate best to predict market trends. Judged on that cold scale of downloaded units, Mein Kampf—which has come roaring back recently thanks to a high volume of e-book sales—might now be considered a good book.
I won’t go so far as to say that reducing the richness of books to ones and zeroes, and then judging them on such a scale, is tantamount to literary eugenics. But it does raise a question about what it means for a book to be formulaic, and whether that’s a good or bad thing. Or whether those kinds of questions even mean anything anymore.
Trying to decrypt the subatomic mechanics of literature is nothing new. Most recently, mystery writer James W. Hall devoted an entire non-fiction book, 2012’s Hit Lit, to boiling down the vast and intricate possibility of the written word into a dozen features that make books sell. At least Hall relies on some softer science than Stony Brook’s prose-crunchers. Rather than feeding reams of text from the Project Gutenberg database (along with some more recent classics and bestsellers) into their algorithm, Hit List touches on themes and tropes that are more likely to appeal to the largest readership: timeliness, topicality, and archetypes like romantic outsiders and unresolved schisms within societies and ourselves.
Hall’s list of qualities is pretty broad, but at least they allow for a little creative interpretation. And that’s where the Stony Brook project becomes more problematic. Publishers have always used whatever tools at its disposal to try to sell books and make money. They’d be remiss if they didn’t. But what happens when authors begin to think this way—or actually start plugging their own work into such an algorithm?
Not that there’s anything wrong, at least theoretically, with authors using analytic methods to maximize the potential of their books. As a novelist myself, I can attest to the fact that such a massive and complex undertaking requires some bridge building between the right brain and the left. Accordingly, any decent creative-writing teacher will offer time-tested tips and tendencies—if not outright rules—that might push a novel-in-progress toward a more sellable, or at least readable, form.
In November, a guide called Wonderbook quietly hit the shelves. Written by science-fiction and fantasy author Jeff VanderMeer—and featuring contributions by everyone from Neil Gaiman to George R.R. Martin—Wonderbook is a how-to resource that seeks to teach people how to write speculative fiction. At first glance, it may seem to be just another aggregate of rules for writers to navigate, another formula to follow. It isn’t. What makes Wonderbook so brilliant is not just its layout—a tumbling flow of unpretentious advice and gorgeous illustrations that unlocks the mind rather than narrows it—but in the way it stimulates intuition and encourages unconventionality. It also gives solid, practical, tried-and-true writing advice, the kind you might find in far less inspiring form in any basic writing class. If anything, Wonderbook is more of a “learn the rules before you break them” kind of thing, an instructional device that celebrates the virtue of literary exploration, both as a reader and a writer—commerciality be damned, or at least downplayed.
VanderMeer’s own work bears this out. His dense, lush early novels like City Of Saints And Madmen and Veniss Underground made a huge splash in the science-fiction and fantasy community, but they in no way threatened to become bestsellers. That may change with Annihilation, the first part of a new trilogy due in February. Although it’s unmistakably a VanderMeer book, its Ballardian strangeness and Lovecraftian scope are carefully measured and controlled—including the concise prose, which probably would rate quite highly on the Stony Brook algorithm. VanderMeer’s career arc can be seen as the literary equivalent of, say, Peter Gabriel’s, who started out with his baroque work in Genesis before greatly editing and streamlining his sound for his solo career. Gabriel became a household name only after he contained his experimentalism in more direct, disciplined forms. The same can be said of VanderMeer—although the verdict for Annihilation, as strong as it is, has yet to be delivered.
But VanderMeer is just one example. Thousands of aspiring novelists, some of them the next VanderMeer, Dan Brown, or Dickens, currently toil away at their keyboards in hope of becoming successful. Obscure or famous, successful or not, literature depends on them. All of them. The reason they’re necessary—even when the vast majority of them are destined to be failures, in the false binary of a literary algorithm—is that they contribute momentum, enthusiasm, and the kind of innovation that can only come from those who scribble outside the lines. And no formula can measure that.
VanderMeer himself took to social media last week with a similar sentiment, posting on Facebook that “book culture is an indispensable part of who we are as a society, and part of keeping it that way is having as much diversity as possible. And I mean ‘diversity’ in every possible way, including not just offering up to readers the kinds of commercial fiction that are easily described and devoured.” He concluded by warning that the increasingly analytics-happy book industry currently risks going “against its own best interests in allowing everything to be commodified, with ‘worth’ measured solely in dollars and cents.” Or in adverbs and verbs.
(Image: Jeremy Zerfoss, from Jeff VanderMeer’s Wonderbook)