I’m an old-school English major. I spent two semesters studying Shakespeare, two semesters studying the Romantic poets, and one grueling semester learning Chaucer from a lecturer who mainly taught at Columbia and reminded us of it every class. Among my electives were courses in Greek mythology, Chinese and Japanese novels, and 19th century Russian literature. (Tolstoy or Dostoevsky? Tolstoy, hands down.) I took a course in American short stories with Maurice Baudin, who reputedly gave Joseph Heller a C+ for the first six chapters of Catch 22, and a colloquium on literature and democracy with Ralph Ellison, who intimidated us so much we hardly ever spoke.
I wasn’t keen on deconstruction, postmodernism, and all that other trendy bullshit borrowed from French literary critics — at least one of whom, it turned out, concealed his collaboration with the Nazis during World War II, explaining why those theories so prize perception over truth — so I didn’t go to grad school. And I’m glad I didn’t, not only because it’s more important for me to write than to criticize writing, but because had I gone the academic route, today I’d be facing the challenge of digital humanities, i.e. the computer-driven analysis of texts and literature.
The shift from qualitative to quantitative analysis is the latest in the academic humanities’ ongoing effort to show they can be scientific (and obtain grants). Sometimes it works. The highlight of the week for me — and the inspiration for this post — was a New York Times article about freelance scholar Dennis McCarthy’s use of plagiarism software to identify a previously unknown source of material for Shakespeare, a treatise called A Brief Discourse of Rebellion and Rebels written in the 1570s by fringe courtier George North. There’s convincing evidence that Shakespeare pulled significant pieces of Richard III and Henry VI Part 2 from North’s work, and that he may even have used the North manuscript as the basis for a soliloquy by Lear’s Fool (Act III, Scene 2, 78–95).
More often, though, data mining takes the human out of the humanities. It’s about filtering millions of words through a cyber-sieve to see what falls where and reporting back the numbers. As the eminent critic Stanley Fish wrote in 2012, “Whatever vision of the digital humanities is proclaimed, it will have little place for the likes of me and for the kind of criticism I practice: a criticism that narrows meaning to the significances designed by an author, a criticism that generalizes from a text as small as half a line, a criticism that insists on the distinction between the true and the false, between what is relevant and what is noise, between what is serious and what is mere play.”
And then there are uses that, from my self-interested perspective, are pernicious. A decade ago, researchers at Stanford embarked on an effort to determine what makes a work of fiction a best-seller. After crunching more than 20,000 novels written over the last thirty years, they came up with an algorithm for success, essentially suggesting that novels could be written by computers, although the researchers resisted that conclusion, saying their algorithm prescribes the characteristics of, but not a formula for, popular novels. A more likely use, they and some editors said in a 2016 Atlantic Monthly article, would be for publishers to apply the algorithm to manuscript submissions, eliminating external biases like the writer’s track record from editorial decisions and thus giving novices and previously poor sellers (like me) a better chance.
I don’t believe it. There have already been a few experiments with computer-written literature. Just as trucking firms may not need drivers in a few years, big publishing houses may not need writers. Plug in a genre, story line, characters, and locations, and run the algorithm.
Which leads me to wonder: if Shakespeare were to come of age today, would anybody publish him? It’s hard to imagine an algorithm based on thousands of recent novels recognizing the value of Two Gentlemen of Verona.