The Simple Software That Could—but Probably Won't—Change the Face of Writing


It took T. S. Eliot about a year to compose his masterpiece epic poem, "The Waste Land," and by the time he was done he had left a substantial paper trail. He wrote his triumph of modernism in a distinctly modern way, as a kind of bricolage, by stitching together some fifty short fragments ten to fifteen lines apiece. These draftlets he would then assemble, type up, and send in carbon copy to a few friends and prospective publishers.

We should be grateful. "The Waste Land" is long and hard to understand, deeply allusive and annoyingly (if brilliantly) fragmented. So any record of its production -- in the form of drafts, letters, typescripts, annotations, and the like - could help us unravel its many convolutions.

Imagine what happened, then, when in 1971, forty-seven years after the poem's publication and six years after the Nobel prize-winning poet's death, his widow released a volume entitled The Waste Land: Facsimile and Manuscripts of the Original Drafts.

Eliot scholars went nuts. They scoured his prepublication manuscripts trying to figure out which of its five parts were written first, how they were arranged into a coherent whole, even what Eliot was reading while he wrote them. To help with the chronology they conducted an almost forensics-level analysis of his typewriters. They discovered -- among many other ridiculous things -- that one model he used produced letters 0.02mm wider than the other, a minute detail which went a long way toward dating his many drafts.

Perhaps their most important finding was also the most obvious. Some of Eliot's typescripts had marks all over them, marks which were known to be the notes of Ezra Pound, Eliot's champion in the U.S. and a well-known literary critic. He had made massive changes to the original manuscript. Example: that famous opener, "April is the cruellest month," used to be buried under a section some hundred lines long before Pound cut the whole thing. All told his edits shrunk the poem in half. As a result it became more cryptic, rhymed less, and in some ways mutated into a bleaker, more biting critique of the modern world.

Which is to say that Pound completely transformed "The Waste Land." And the scary thing is that we might have never known -- we might have lost our whole rich picture of the poem's creation -- had Eliot not been such a bureaucrat, typing up and shuffling around so many snapshots of his work in progress.

Fast-forward to the present day. Some people worry that with the advent of the word processor, early drafts of important work no longer survive. When a writer hits "Save" he creates a snapshot, sure, but a snapshot which overwrites his earlier work. So unless he's particularly conscientious, the only draft he'll end up with is the draft, the one he publishes.

Things started looking up 663 days ago, when the well-known entrepreneur and venture capitalist Paul Graham, on his popular social news site Hacker News, submitted a story with an atypically dramatic title: "The most surprising thing I've seen in 2009, courtesy of Etherpad."

He linked to The bulk of the page was taken up by a basic text editor, and on top of that a slider, the kind you might use to move through a song in iTunes or adjust the treble on your hi-fi. As you moved the slider the text changed.

Graham explained what you were looking at:

I've been wanting to play back the writing of an essay for years. Since Etherpad saves every keystroke, I convinced the founders to add a way to play them back. "Startups in 13 Sentences" was the first essay I wrote on Etherpad. Now I'm going to write all of them on it.

Playback is just one little feature of Etherpad, but think of the implications of this alone. Among other things it will make cheating impossible in classes where students write papers, because now you can finally "show your work" in writing the way you do in math.

Forget about saving drafts -- Etherpad promised (or threatened) to save every keystroke: every note and idea, every version of a phrase, every snag and breakthrough. It would all be recorded, and labeled, and automatically backed up as you typed.

Imagine the consequences. As Graham said, students using Etherpad in English class would in effect be "showing their work." It could do wonders for teaching. But that's nothing compared to the idea of our great poets and novelists using such a tool to record the minutiae of their creative process, the dynamic history of their work, and bequeathing it all to their readers. It'd be like Eliot letting us lean over his shoulder.

Before we get carried away, though, let's keep in mind that Graham's announcement was 663 days ago -- an eternity in software time -- and that you probably know precisely zero people who use Etherpad or, for that matter, anything like it. What's going on?

For one thing, nine months after Graham's post, Etherpad was acquired by Google and folded into the Google Wave team, another project that, while no doubt a high-caliber technical achievement, is now basically defunct. If you go to you'll see that the service has been shut down.

But really that shouldn't matter. Etherpad's "Time Slider" was an afterthought, something they hacked together at Graham's request. It's fairly simple -- as a matter of fact, when they released it there were already several other tools that did the same thing. What's more, before closing up shop Etherpad open-sourced their entire codebase (a fantastic public good, by the way), making it trivial for just about anyone to make an exact replica.

So something else must be going on. We must not want to write using a tool that tracks our every move. Because when people want something, and the tech to make it happen is readily available, it tends to happen.

I have a few theories, but they all start with the fact that writing is fundamentally about the final draft. It's not like writing code, say, where recording one's every change is standard practice. (Ask any coder worth her salt whether she uses a "version control system." If she says "no," well, she's not worth her salt.)

That's because code is so fragile, and simple changes can propagate in complex and unpredictable ways. So it would be stupid not to keep old versions -- i.e., versions that worked -- close at hand.

Writing is different. A writer explores, and as he explores, he purposely forgets the way he came.

I'm reminded of how the word "essay" derives from the French "essayer," a verb meaning "to try." It was coined in the late 16th century by Michel de Montaigne, in many ways the father of the form. Montaigne wrote as a kind of maieutic exercise, a way of drawing his thoughts into the light of day, of discovering what he wanted to say as he said it.

No need, then, to drop so many breadcrumbs along the way. Especially when such a trail could do more harm than good. Readers could use it to find places where you massaged the facts; they'd be able to see you struggle with simple structural problems; they'd watch, horrified, as you replaced an audacious idea, or character, or construction, with a commonplace.

This is not to mention the legal ramifications (teasing out someone's "intention" just got a whole lot easier...) nor the mere fact that working under this kind of surveillance could drive you crazy with self-consciousness.

I should know: I wrote the article you're now reading using Etherpad's software. You can watch how I fumbled along, start to finish, by clicking the big "play" button on this page.

Main Eliot source: Rainey, Lawrence S. "Eliot Among the Typists."