The Simple Software That Could--But Probably Won't--Change the Face of Writing | Hacker News Reader

The Simple Software That Could--But Probably Won't--Change the Face of Writing

theatlantic.com

226 pointsjsomers69 comments

This is linked in the article, but you can watch the author write this article here: http://ietherpad.com/ep/pad/view/somers-atlantic/latest

It's really interesting to watch someone else's writing process and see how that compares to mine. (Especially when that someone does a lot more writing than I do.) I find that when I'm writing a mid-length essay like this I tend to get very disorganized very quickly. I think it'd be valuable to try to record myself and see what, exactly, I'm doing, and then try to use that to improve.

Wow! This makes me feel so much better about my writing. I always thought good writers just busted out an article. I didn't think they put in so much scaffolding. Being a good writer now seems almost attainable.

I'd like to see the same thing done with math.

That's a hell of an idea. One thing it took me a long time to learn was that being good at math is mostly about working really hard. It's hard to know that, because until you're a graduate student the only messy incomplete work you ever see is your own. Everything you find in textbooks is pristine, and worse, it all looks like it occurred quite naturally to the author.

"because until you're a graduate student the only messy, incomplete work you ever see is your own"

That's assuming, of course, you never ever try to solve a problem with anyone else before grad school. That seems possible but highly unlikely.

There's a big difference between seeing a peer struggling on a problem and seeing an instructor/professor struggling on a problem.

A few weeks ago I was in a maths lecture, and the lecturer was, as is fairly standard, writing proofs from his notes as he went. He then mentioned one of the exercises, explained the proof strategy for it, and then started to actually write out the proof.

After five or six lines he stopped, scratched his head, looked at the proof, and crossed out the last line he'd written. Then he looked at it some more. After a couple of minutes he looked up, apologised, and said he'd got stuck and that we'd go back over it in the next lecture.

It was a great relief to me that my own struggles were not the result of laziness or stupidity: some things are just hard, and take a while to get right, and it's rare to be able to write down something in its correct form on the first pass.

>After five or six lines he stopped, scratched his head, looked at the proof, and crossed out the last line he'd written. Then he looked at it some more. After a couple of minutes he looked up, apologised, and said he'd got stuck and that we'd go back over it in the next lecture.

He probably just proved the Riemann Hypothesis in his head and got distracted.

"The worst thing you write is better than the best thing you didn't write."

Oh, yes. One of the things I learned as a writer is to write crap. Write drivel. Sitting there trying to get it Just Right, staring at the page or screen as you attempt to dig up that bon mot just kills it.

Write whatever, get it down. Out of order, out of the blue? Stick it in the Do Something With This section. Just get it out of your head and saved someplace.

Then you can go back, read it over, rewrite stuff, fix it up and, ideally, cut the shit out of it to make it tight on a serious murder-your-darlings rampage.

Hence NaNoWriMo, of course.

Oh, I don't know about that. I didn't write Don Quixote, but it's pretty good.

I don't consider myself a great writer, but for anything complex I work on it will be full of bits like "Benefits include X, Y and [needs third benefit]" or "[NEEDS TO EXPLAIN WHY WE ARE MORE SECURE CHOICE]" that I throw in so I can get my framework down without getting bogged down in details.

It didn't hit me until you in particular said it. This sort of scaffolding reminds me of how I write code.

   def foo(a,b,c):
      h = bar(a,b) + bar(b,c)
      if h > a + b + c:
         h = dothatotherthing(h,a,b,c)
      return h

where neither bar, nor dothatotherthing are written yet. (clearly, they have better names than that)

When I write anything longer than about a paragraph, I always start with an outline. I used to hate outlines; I'd just bust out a 10-15 page paper and write up the outline afterward (who didn't do that?).

Now though, for a 1000-1500 word piece, I spend at least an hour outlining it. Even the outline itself goes through a few drafts before I start trying to add complete paragraphs. I'd rather have a solid grasp on the ideas I want to communicate than on the specific turns of phrase I'll communicate them with. Turns of phrase are a dime-a-dozen. A clearly communicated idea is much harder to find.

A lot of that has come from practice, but I also owe a lot to PG's essays on rapid iteration from mush to presentable product.

My teacher once told me of a newspaper journalist who, when using his computer, would just turn off the monitor and write his piece from start to finish. He would then, presumably, proceed to polish it afterwards. How is that for distraction-free work flow?

There are so many ways to approach writing, it sometimes scares me.

For essays, a revision database is useful. They're constructed of elements, single arguments for a collaborative effort to prove an argument or illustrate a point.

When I write fiction, I don't want to remember where I came from. If I remember my source of an idea too well, I risk full out plagiarizing an idea instead of making it my own and building on it.

I can't write a story about a boy going to a magical school and focus too heavy on the obvious source without making a fanfic of Harry Potter, and you can't focus on the less obvious source without making a fanfic of Earthsea.

An essayist needs to remember their source to stay on point. A fictionist needs to forget the source and get lost in the literary woods. You build a universe and have an adventure.

Stephen King for example finds an idea and simply asks "what happens now?" Asimov had the problem that he played the formulaic method and some of his stories were extremely weak because of the lack of freedom. He dedicated his whole work to the source, and only really succeeded because his ideas were at least original. However, Asimov is my personal least favourite of the great SF writers. I'll drop dead for any Arthur C. Clark or any Heinlein, but I'll get bored with the vast majority of Asimovs work because it's so A,B,C formula that it kills the fun.

I'm sorta the opposite way - when I write fiction, I really want something that remembers everything I wrote before, because then I can forget about what I just wrote.

I used to have a real problem with deleting code, because I worked hard on it, damnit, and a non-trivial number of times the stuff I thought was the wrong approach turns out to be the right one. That fear went away once I started using version control, because I know that if I ever need the code, it's a rollback away.

The same goes for fiction. I find it hard to discard ideas and whole pages of prose, because damnit, I worked hard on that, and the story might shift again in the future to require it. If I'm using a word processor with revision history, I can just trust the computer to keep the old stuff, and if I ever need it again, I can pull it up.

Back when I wrote reviews I developed an all-or-nothing approach. The few times my editor said "this one needs major changes" I would simply delete and rewrite the whole thing from scratch - one time did a total rewrite of a 1,500 word review inside an hour, and I know for me I couldn't have done it without 'destroying' (there's always a copy somewhere) my old work.

I suppose if having a revision history gives you the bravery to take a hatchet to your work, then it's certainly beneficial to your writing.

If anything I would need a version control because I edit like a Roman general; I decimate. I'll take out huge swathes because it doesn't fit, but once or twice I find myself going back and putting essentially the same sentence in - written better, but nothing editing couldn't have done. However, I live with my methods, but I've wanted to try a revision history when writing a short story as I need to ridiculously edit them compared to a longer piece.

I'll tell you what would really change the world: forcing our legislatures to use version control when writing our laws.

No more sneaking in cryptic revisions in the middle of the night without anyone noticing! No more anonymous additions from lobbyists!

http://radar.oreilly.com/2008/11/change-gov-revision-control...

Unfortunately if teaching our designers and managers to use git is hard, the prospects of teaching our elected representatives to use it are very slim.

I don't think teaching them, or their staffs, to use version control will be the major obstacle. Convincing representatives that they need to be more accountable when it doesn't really help them get reelected will be much harder.

(This is self-promotional, but hopefully relevant. Please downvote if you disagree.)

At Leanpub we fundamentally disagree with the point of this article [that writers don't want their old versions], and agree with what Etherpad was going for with the time slider: we use Git for all the book revisions (both with our in-browser editor or via a direct interface to GitHub), but we do this in an unobtrusive way. (We'll probably expose a similar time slider eventually, but for git revisions.) While this may or may not be needed for writing of essay length, we feel it is essential for books. Also, imagine the contribution to scholarship if, say, Ulysses had been written with version control...

Yea, I really don't see what the risk is here if the author has the expectation that only he/she will be able to access the version history (unless, of course, the work is so good that academics want to study it).

[warning: grumpy guy ahead.]

I'm sorry, but other than praising etherpad and pg, and sort of snidely lamenting the fact that nobody uses etherpad, was there a point to this article?

"change the face of writing" is a pretty big commitment to make in a title. I just spent 5 minutes reading something that told me absolutely nothing new, which had no conclusions aside from something like "writing and publishing are two different things" -- a statement that is glaringly obvious yet the author never seems to get around to actually saying.

I'm seeing a lot of Atlantic articles popping up on HN. Let's hope they stay of a high quality, and don't just rehash old ground and pander.

I was a big fan of wave, but one of the many criticisms it had was that nobody really wants everything they type published. You want to finish something, then you publish.

I mean really, think about this: watching the physical act of somebody editing themselves is going to make such a huge difference in writing? Oh really? We won't need editors, or conversations about strategies to use in our pieces, or the creative tension and problem-solving that goes on in a good publication? Just playing with a slider and watching characters appear and text be edited will change the face of publishing?

I gave pg a free pass when he wrote the first article. After all, it was a YC company and it was a cool feature. He had a right to overstate things. But simply repeating a canard doesn't make the underlying structure of the universe change, no matter how cool the feature is.</rant>

You're right, but, like a lot of old-but-new things on HN, I hadn't heard of Etherpad, or read pg's praise of it, before this article. There are probably a lot more people like me reading the Atlantic, than people like you, to whom it's all old news.

Fighting the digital dark ages isn't a big deal? I thought the author made it pretty darn clear how revision control would change the face of writing - it would change it back to how it used to be.

http://etherpad.com/ep/pad/slider/13sentences

Sad. Now that etherpad is no more this incredibly relevant link is gone forever.

(it was a replay of pg writing an essay http://news.ycombinator.com/item?id=495336)

Jon Udell's narrated animation of the collaborative process behind Wikipedia Heavy Metal Umlaut article is hilarious and captivating in a similar way:

http://jonudell.net/udell/gems/umlaut/umlaut.html

He makes the case that a million tiny edits by different people over time really can construct something magnificent and structured.

Oh Jon Udell, how did you fall off my feedreader?

My favorite, perhaps: http://jonudell.net/udell/gems/googleMaps2/gmap2_flash.html

/me goes off to catch up on jonudell.net

And I only half-agree that version control isn't important in writing. Certainly in (collaborative) non-fiction (e.g. science papers), most of the same merging, cutting, reinstating and backtracking that helps with programming would be of huge value to authors.

I can think of dozens of smart non-programmers who I think would relish a friendly Git for writers - it would, of course, have to be called Writ...

It's a big UX design problem to come up with something demonstrably better than a series of saved milestone copies plus Word's "Track Changes" feature.

The milestone copies are the equivalent of commits. If your doc is in Dropbox you probably get them for free, but there's definitely room to improve the interface to those old copies.

Track Changes is the prose version of "diff", plus it lets you put in editorial annotations which are kind of akin to commit messages.

I expect that the designer who attacks the problem will quickly realize that prose is just much less clearly structured than code. Code is built of very simple syntax, grouped into lines, which are grouped into blocks, which are often named as functions and then called from other bits of code via tightly defined interfaces. This is done to make the code easier to change, and it simultaneously makes the changes easier to track, easier to categorize and label and annotate. (That's not a coincidence; "making the diffs neater" and "making the code easier to change" are aspects of the same task.)

Prose does not look or work like code. It is not DRY, not at any level of abstraction from the grammar (change "he" to "she" and you may have to edit a lot of other words, not necessarily confined to one sentence or one paragraph) to the semantics (because English doesn't have a gender-neutral pronoun, a substitution of "he" for "she" threatens to alter the actual meaning of a sentence, and of adjacent sentences) to the theme (in 2010, a novel about a man who becomes President of the USA is necessarily different from a novel about a woman who becomes President of the USA). And so it is a lot of work -- work beyond the powers of a simple algorithm -- to detect which edits are logically grouped with which other edits. Composing sensibly organized "commits" and writing the "commit messages" for each revision of The Waste Land would probably be harder than writing the poem itself. Just ask one of the historians, whose job it is to do so.

If you are a historian with all the time in the world, an Etherpad session of T.S. Eliot composing The Waste Land would be awesome. But would it have been so awesome to Eliot? Or would it have been only marginally more useful than his typed and scribbled hard-copy revisions? Or maybe even less useful? Forgetting is a very useful skill. We've got fancy mental mechanisms for managing the process, and those mechanisms evolved for a reason.

> If you are a historian with all the time in the world, an Etherpad session of T.S. Eliot composing The Waste Land would be awesome. But would it have been so awesome to Eliot?

This is the key. We want to see that revision history and only in retrospect for a tiny proportion of writers in the world. I'm not convinced that the writers themselves have any motivation to do so, especially with the UX complexities that as you point out would be necessary.

If the people in the position to do it have no motivation to do it, it won't happen.

And then there's the other side of the coin. A writer may prefer not to show people drafts. Imagine a politician doing it. At the moment we vet things before we publish them. We do this for various reasons including for confidential information and to avoid upsetting people. How much more of a burden will it be to vet all revision history for this too?

Point them to scrivener. It's no git, but you can take snapshots of your work, rearrange by scene, keep notes, it's a very writer-friendly editor.

This is great; it's a pretty good feeling to know that people still use Etherpad.

jsomers: Do you frequently draft text using Etherpad? Assuming no one else will see your "breadcrumbs", do you find it useful just for yourself?

I do have a local instance of etherpad but the friction to get it running seems to be just enough to prevent me from firing it up. So mostly I use either textmate or something called writeroom. I like using textmate with markdown because you can highlight something, hit ⌘I, and it's italicized.

For a while I thought where I wrote mattered, but after producing good and bad just about everywhere I decided to just go with whatever jumps to mind first.

I'm amused when I look back at my "breadcrumbs," but I don't know if they're useful. They're just a record of a struggle you already went through. So it's not like you're going to learn a whole lot of new stuff.

Where they're useful is when you think to yourself, "maybe the way I phrased it was better." I did that a number of times, and probably wouldn't have been able to if I hadn't used etherpad.

It's really tough to say. I bet a lot of the benefits would accrue over the long run.

Word of the day: Maieutics

Wikipedia says "is a pedagogical method based on the idea that due to the reason innate within each of us the truth is latent in the mind of every human being, "

A link to the T.S. Eliot poem for the curious minded like myself: http://www.bartleby.com/201/1.html

I used Etherpad on http://piratepad.net all quarter for group write-ups. Any other workflow would have been cumbersome.

Google was foolish in killing Etherpad and would be smart to create a lite mode for Google Docs that emulates it closely.

Yeah, their list of versions is nowhere near as useful as the slider. Nor as fast.

I haven't used it for a bit, but isn't the multi-person editing a lot like etherpad? Chat and everything? Irritating that it's locked down, and doesn't have the slider, but it's a lot better than non-etherpad-clones have managed.

i would never have expected to read an article that mentions both ezra pound and paul graham. awesome

Reminds me of what I've heard about "Mt. Fuji" questions for technical job interviews: the interviewer doesn't really care that you get the question right, s/he just wants to see how you think.

Yet somehow, I think that the mere knowledge that each keystroke is being tracked will somehow skew the natural writing process. Some shameless self-promoters may turn writing into something like a performance.

(here's the timeslider: http://ietherpad.com/ep/pad/view/D8SOJGh37k/latest)

Maybe...or it will simply create a new branch of technical writing such that the performance is just as important as the final work.

It is really refreshing to read an article which isn't primarily about software development get some of the details about development so right.

When writing my PhD, I used CVS to track changes, and in addition revisioned personal milestones to a new file. All told there must have been ~100 versions, each with a revision history tracked in CVS.

Here the number of commits from one of my current papers (accepted, not yet published, names obfuscated).

  $ git shortlog -s
   193  First Author
   289  Second Author
    36  Third Author

If you write something together version control is just necessary.

Google docs has a revison history.

Google Docs doesn't save every keystroke, it is not as fast and elegant as Etherpad's.

Agreed. But its more then any other supported product I've seen.

I presented time-lapsed video of myself writing an essay and presented it at an event, with audio commentary.

I didn't know about Etherpad at the time, but there are some things captured here it wouldn't get, as they happen outside the browser.

It also might be more educational, as I explain in voice-over what's going on:

http://www.scottberkun.com/blog/2010/video-how-to-write-1000...

Scott Berkun (formerly of Microsoft and author of three bestsellers) did a talk at Ignite about this where he recorded himself writing an essay: http://www.scottberkun.com/blog/2010/video-how-to-write-1000...

Really interesting to see how much work even a 1000 essay takes!

In the words of Terry Pratchett:

"I save about twenty drafts -- that's ten meg of disc space -- and the last one contains all the final alterations. Once it has been printed out and received by the publishers, there's a cry here of 'Tough shit, literary researchers of the future, try getting a proper job!' and the rest are wiped."

What I found fascinating were the links to the reference material being used while writing the article that were later removed. I think people tend to forget that this is a hypertext medium and thus, you can always link to your source materials in context, in case anyone is curious enough to follow up on it.

Good catch. What mostly happened was that an old prof of mine, upon reading a draft, pointed me to a much better source than the ones I was using. So I ditched that material (and those citations) and just used the Rainey article, the one footnoted at the end of the final draft.

One solution is to eschew word processing and use a typesetting language instead. For example, I wrote the Ruby on Rails Tutorial book (http://railstutorial.org/book) in LaTeX and versioned it with Git. This proved useful on many occasions.

Mirror of the picture link in the last version (took a while to respond when I tried to view it): http://imagefrog.net/show.php/133491_tsewlms2.jpeg

vim has "persistent undo" (undo/redo changes for every file you edit, saved across quitting and restarting vim):

http://choorucode.wordpress.com/2010/11/27/vim-persistent-un...

vim also has "undo branches", which allow you to undo/redo not only to any point in a linear set of revisions, but to any node in a tree of revisions:

http://vim.wikia.com/wiki/Using_undo_branches#What_are_undo_...

Now if it only had a cutesy little slider bar at the top..

Just a note for anyone mourning the loss of etherpad and unhappy with the selection of clones, I've been really happy with http://sync.in.

Just wanted to point out that Simplenote saves old versions of notes, as well — not infinitely or as fine-grained as Etherpad did, but hey.

Doesn't Dropbox version stuff?

etherpad - now a part of google apps.

Not even version control saves every keystroke. Something I've always wanted was complete undo history, in every program I use.

I think that was part of the Google Wave vision. Also part of Jef Raskin's vision detailed in The Humane Interface, which so far I've only read summaries of.

I didn't make the direct connection between Google Wave and The Humane Interface until now. I wonder how much of the former was inspired by the latter. The first search brings up this blog post, where the author mentions that he's taking notes in Google Wave but doesn't compare Google Wave to The Humane Interface.

http://roberthanson.blogspot.com/2010/01/humane-interface.ht...

I'm amused.

Yes, Google wave incorporated that as part of it's platform which would be cool for building Google Wave apps since it's already baked in. For other programs though, it seems like it would just be a matter of persisting the undo data. This data could also be source controlled allowing you to peer into the process of developers before you and gain a better understanding of the code. Keeping all your history in Photoshop would also be awesome. I wonder why apps don't do that more often. It seems like such an obvious and kick-ass feature.

It's a complexity problem, though: not just for developers but for users. That's why it isn't about "other programs".

The way out of apps dealing with complexity individually is for the platform to do more. This is what The Human Interface argues, going so far as to say there shouldn't be a plethora of separate apps for different tasks.

When Google Wave first came out, Google touted its ability to be used all over the web, and to support custom-coded plugins that would get its versioning support for free.

There are tradeoffs to adopting a front-end platform that gives useful stuff that are consistent across apps or sites for free. It's a very interesting concept to me nonetheless.

Blasphemy, don't you commit every keystroke? :D

Textmate has char-by-char undo and it is deeply annoying, one of the most oft-cited criticisms.

I assume this is because it's too slow for normal undoing? The obvious answer is to make the undo interface a bit more sophisticated, such as having an option in the edit menu to control the size of the undo "steps".

Mathematica desperately needs this. I lost about a day's worth of work once (although I was able to replicate it in a couple of painful hours) and I was flabbergasted to read that Mathematica has no automatic backup by default whatsoever. Further, the backing up system is horrendous if enabled.

Etherpad may actually hinder more than help writer archeology.

Would T.S. Eliot's widow have had access to his drafts and correspondence if they were in Google Docs and Gmail, respectively? Or worse, what if they were in the now defunct Etherpad?

Reply on news.ycombinator.com