Hemingway and Versioned Writing

Ernest Hemingway wrote at least 47 different endings to A Farewell to Arms. He gave some of them great, semi-fanciful names: “The Nada Ending,” the “Live-Baby Ending,” the “Fitzgerald Ending.” A newish edition of the book, published by Scribner in 2012, includes them all. It’s rare glimpse into the hidden work that goes into a novel. Is it getting rarer? In a New York Times article about the Scribner release, Julie Bosman suggests that it is:

[S]ince modern authors tend to produce their work on computers, the new edition also serves as an artifact of a bygone craft, with handwritten notes and long passages crossed out, giving readers a sense of an author’s process.

I’ve been keeping an eye out for similar laments — not because they’re wrong, but because they seem to capture the current perception about how computers affect writing and writerly archives.

Toward the end of Joan Didion’s The Year of Magical Thinking,¹ she’s reading through a Microsoft Word file her late husband used for note-taking:

According to the computer dating the file called “AAA Random Thoughts” was last amended at 1:08 p.m. on December 30, 2003, the day of his death, six minutes after I saved the file that ended how does “flu” morph into whole-body infection. He would have been in his office and I would have been in mine. […] The file called “AAA Random Thoughts” was eighty pages long. What it was he added or amended and saved at 1:08 p.m. that afternoon I have no way of knowing.

A Wall Street Journal article describes Ken Lopez as “a broker—one of a dozen in the country—who deals in the flotsam of authorship,” such as writers’ old grocery lists, record collections, and correspondence. He, too, laments the digitization of the literary process. From the Journal:

Typescripts that used to be cut, pasted and blue-penciled are spotlessly digital now. Word processing has obscured the process of writing. “The archaeology of it, seeing paths not taken,” Mr. Lopez said early one Monday. “At this point, that’ll all be lost.”

Hannah Sullivan, a professor at Oxford, has just published what sounds like a fascinating book about the history of revisions in literature. Speaking with the Boston Globe, she argues she argues that the shift away from paper-based workflows also changes the way writers revise:

While [computerized word-processing] makes self-editing easier, Sullivan thinks it may paradoxically make wholesale revision, the kind that leads to radically rethinking our work, more difficult.

“The ideal environment for revision is one where you can preserve several different versions of a text,” Sullivan says. With only one in-progress draft on a computer, we lose the cues that led the Modernists to step back from their work and to revise it. “It’s that moment of typing things up that led to the really surprising and inventive changes,” Sullivan says. “The authors came back to their text, but it seemed estranged.”

Didion, Lopez, and Sullivan’s concerns are entirely valid, but they’re strongly influenced by the current state of digital word processing. Today’s popular word processors are terrible at handling successive versions of a document. The emergence of new, versioning-savvy tools such as Draft and Editorially, suggests that we’re getting closer to a mode of writing that reaps the efficiencies of computers without losing the archival benefits of paper.

I look forward to the day Scribner publishes The Collected Track-Changes of your favorite modern novelist. Or, even better, a Git repository of the entire evolution of their magnum opus. In theory, those could be magnitudes more revealing than handwritten sheafs of paper. But until then, it’s hard to shake the perception that we’ve lost those “paths not taken.”