Skip to main content
Watch Me Write This Article

If you could watch a replay of James Somers building a program that lets us see our own writing played back to us, you’d need to start seven years ago. You’d catch a glimpse of him sending drafts of his essays to a college professor, who said the best way to write was through immediate feedback on revisions in progress. You’d look over his shoulder as he excitedly read an essay by venture capitalist Paul Graham, who persuaded a company called Etherpad to install a timeline slider in a word processor so Graham could watch as the computer showed him his writing process keystroke by keystroke. You’d sit by Somers’s side as he played around with Etherpad and saw it didn’t do everything he wanted. And then you’d have sympathy for a guy who spent years trying to reinvent the wheel before he realized the hub and spokes were at hand all along.

In November, Somers, a developer for Genius, released an app called Draftback.1 It’s a fascinating experiment that treats writing like data. After years of trying to build a program, Somers realized that Google Docs was already saving every keystroke we enter. So he hacked Google Docs to play documents back to their authors, materializing on the screen with every stutter-step inherent to the writing process. In its latest form, Draftback is a Google Chrome extension that can reach deep into the archives of any Google Doc you have editing rights to, make sense of all that writing and rewriting you innocuously poured into it, and beam it right back to you, backspaces and all. It doesn’t matter if your document was created before or after you installed Draftback — the keystrokes have been buried the whole time. Draftback can unearth any fossil.

In practice, it looks something like this:

It’s a program that acknowledges how we write — in a word processor, staring into the maw of a blank screen — and then turns the computer into a camera. What can we learn if we rewind and press play?

Somers started all this because he thinks the way we teach writing is broken. “We know how to make a violinist better. We know how to make a pitcher better. We do not know how to make a writer better,” Somers told me. In other disciplines, the teaching happens as the student performs. A music instructor may adjust a student’s finger placement, or a pitching coach may tweak a lefty’s mechanics. But there’s no good way to look over a writer’s shoulder as she’s writing; if anything, that’ll prevent good writing.

As a result, finished work has become a writer’s currency. It’s what she hands in to her editors, what she publishes as a book, what she’s assessed on. The process of writing — the masochistic act of choosing what to put down on the page — is merely what she complains about to friends. It’s a hidden act, and self-conscious writers (as if there were any other kind) prefer it that way.

Somers wants to use Draftback to peek over somebody’s shoulder — ideally somebody really good. His personal goal is to get A.O. Scott, the film critic for The New York Times, to write a review or essay in Draftback. “He’s a beautiful prose stylist (diction, cadence, etc.), his writing is accessible and unpretentious but world-class, and he seems to always put his finger on the essence of whatever it is he’s talking about.” Somers is curious about whether all that comes naturally to Scott.

But if not Scott, he’ll take any sensational writer. “I am not going to let a tiger go by the tail until somebody really, really great writes something important and then people can break it down. Because that’s going to be an artifact that’s valuable to every single high school teacher, high school student, college teacher, every literate person in the world.”

There’s a whiff of startup utopianism in there — the idea that one hack of Google Docs could somehow have a global effect. Somers, a talented writer who chose to code instead of toil in the inky depths with dopes like me, admits that the idea of tracking every keystroke springs from a coder’s mind-set. “I’m a programmer, and no programmer does any work any more without a revision control system, like Git. You start to wonder, what is the revision control equivalent for writing?”

This obsession with revision is what separates Somers and Draftback from the way we usually hear about the writing process. The way great writers work is often portrayed as an almost alchemical recipe: Combine a 4 a.m. alarm with 8 ounces of coffee, all while playing the same gamelan recording every morning, and then if you’re lucky you can create a heartbreaking work of staggering genius.2

We’re suckers for that stuff. It’s like a “lose 40 pounds with one easy trick” scam but for artists., Maria Popova’s hugely popular paean to the creative life, has published so many writers’ advice about writing that she collected it in one central list. And a post on writers’ routines proved so popular that Popova turned it into an infographic showing the writers’ waking hours, productive hours and number of major awards. Similarly, Sarah Stodola, in her new book, “Process: The Writing Lives of Great Authors,” details 18 writers’ rhythms. She ends each chapter with a cheat sheet of a writer’s process — she calls it “A Day in the Writer’s Life” — but it’s just a description of the day. Joan Didion’s discrete revisions remain a mystery.

Of course, Didion didn’t make her name in the age of Big Data. Embedded in Draftback’s ingenuity is also a certain kind of inevitability: that writing, like any commodity, is at the mercy of a technology that never forgets.

Somers has laid out the utopian case for what that can mean: People learn how to write in a way they never could before. But there’s a dystopian one, too. Because of the way Draftback works, anybody with editing rights to a Google Doc can see that document’s past. As I wrote this article, a piece from a FiveThirtyEight freelancer landed in my inbox — along with its 22,256 keystrokes. After a few clicks, I could watch that whole article materialize before me, free of charge. Is that my right as the editor, or does it violate some tacit agreement between the writer and me? As I researched this piece, I admit I glanced at a few pieces I edited years ago, and it felt like eavesdropping on a TED talker’s rants before he went on stage. I was used to the self-assured performance, but behind the curtain was an ocean of self-doubt.

(When asked for a comment about Draftback, a Google spokeswoman declined.)

While watching the playback of this piece, I swam through such dark waters I still can’t shake the neurosis out of my ears. (You, too, can witness this horrifying journey in the embed at the end of the article.) At one point, this article started with a reference to VCRs — VCRs! — even though it’s about Big Data and modern technology. I may as well have led with Betamax. Still, that’s better than a passage that once included the words “Github” and “warts” in the same sentence. And let’s not mention the tortured analogy between Bach’s Cello Suite No. 1 and the writerly process.

But, surprisingly, I extracted some nuggets of insight from the replay. I already knew I used a ton of “TKs” as placeholders in my work,3 but I didn’t realize they were a sign I wasn’t committed to a piece’s embryonic structure. The replay showed me that I build a story’s skeleton first, so I can see where all the pieces fit before I put any flesh on the bones.

The first section of this piece changed three times, and it wasn’t until 40 percent of the way through the process that I arrived at one that worked (editor willing).4 Once that was out of the way, it took me nearly a week of half-hearted efforts to hone a thesis that felt compelling rather than perfunctory. Then it took another week and my editor’s prodding before I really got it right. It wasn’t until a few rewrites in that I understood that Somers wasn’t a full character in this article so much as he was its catalyst. Once I pared down his and Draftback’s origin story, the pace quickened; I got to the thesis faster and established the tension with more oomph.

The data, in other words, told me something I didn’t already know. That was surprising. When I wrote about data and online dating last year, I concluded that the huge databases at places like OKCupid were really just articulating things we already know about ourselves. We know we are a little racist, that opposites can attract, and that men get older but the women they want never do.

Data on an individual level, though, holds more promise, as any Fitbit-er5 can attest. When I look at an OKCupid chart, I can only assume what was going through the minds of the guys who copied and pasted pickup lines to dozens of women at once. But while looking back at my behavior — whether it be for exercise, writing or some other thing — the director’s commentary is easy to find. It’s in the same brain making sense of the data.

But Somers wants more: He wants you to learn from me. Which, good luck.

Understanding the decisions behind my (and my editors’) nearly 16,000 keystrokes is going to be difficult, at least until Somers can add the ability to annotate decisions I made while writing. He said he’s interested in adding that capability (he does work for a site that annotates the Web, after all), but until then, another person’s process is likely to be inscrutable. Watching a great writer’s work replayed can’t articulate the almost divine inspiration a writer needs to write something great.6 You might see that Thomas Jefferson changed, “We hold these truths to be pretty darn clear” to “we hold these truths to be self-evident,” but that wouldn’t tell you why he made that change.7

Even with annotations, there’s no way to tell whether a writer will get spooked by knowing her process is fair game for her readers. For Graham, the venture capitalist, the idea that his whole process could be made public was too much. “Incidentally, I didn’t write any more essays using Etherpad because it was distracting thinking that people would be able to see everything I typed. When I write an essay, I’m usually so worried about whether I’ll be able to do a good job that I feel I can’t afford any distractions.”

Even when the process can be made public, it’s still easiest to keep it to ourselves.

Press play below to watch me write the article you just read. Click here if you want to punish me and watch it in full-screen.

CORRECTION (March 6, 2015): An earlier version of this piece originally misidentified A.O. Scott as a Pulitzer Prize winner. He was a Pulitzer finalist in 2010.


  1. Fittingly, he unveiled it alongside a blog post chronicling its development cycle. One can’t release a replay app without a replay of its origin story.
  2. Or “A Heartbreaking Work of Staggering Genius.”
  3. “TK” is an old journalistic trick to mark a spot that the reporter has to fill in before an article goes to press. It’s the single thing that’s most changed my writing over the past TK years. Whoops, make that “the last eight years.”
  4. My editor’s response: “It’ll work perfectly if you accept my suggestions.” His edits start around Jan. 28 in the replay; the copyeditor’s edits start around Feb. 11.
  5. FitBitter? Fitbiter?
  6. I stole this line from my editor.
  7. My editor stole this line from an old “Family Ties” episode.

Chadwick Matlin is FiveThirtyEight’s senior editor for culture.

Filed under

Comments Add Comment