As a lawyer, I was both fascinated and horrified by the replay.
Imagine a company routinely using EtherPad (really cool, BTW) to create documents -- in the process saving thousands if not millions of interim drafts.
Now imagine the company getting into a lawsuit. Some subset of N documents -- and of all interim-draft snapshots of those N documents that are still in existence -- will have to be screened for possible disclosure to the other side. (There are tools for partially automating this, but lawyers and paralegals will still have to individually look at many documents / drafts.)
In PG's case, there were 2,886 such snapshot drafts for just one document.
Makes me shudder just to think about the legal expense.
Imagine that in a lawsuit, PG were to have his deposition taken by the other side's lawyers. The lawyer representing PG or his company would almost certainly insist on prepping him beforehand by reviewing with him, page by page, each of the documents the other side might ask him about. (Failure to prep a witness for deposition can result in the other side's playing a video for the jury, consisting of damaging sound bites harvested from the witness's testimony.)
The lawyer's prep of PG will likely include at least a glance at, and perhaps a discussion of, each of the EtherPad snapshot drafts for each of the documents in 'the PG collection.'
As a human, I'm both fascinate and horrified by your reply. Why don't you go through very half thought he has to see what might be a legal problem - oh wait, you are talking about that...
They could easily release it for everyone, and they probably will at some point, but right now they're busy working on the version that lives inside companies' firewalls:
It would be cool if we could see PG (and other good programmers) code with Etherpad. Watching a sophisticate peice of code "evolve" from the first line/function would be fascinating.
I liked the evolution of persuasion: it would be easy, just use Erdos numbers ... detailed explanation ... to finally, just hiding the details entirely.
The accelerated replay is a good default. My attention span would require some kind of commentary track if it was less than 200% sped up.
I liked the evolution of persuasion: it would be easy, just use Erdos numbers ... detailed explanation ... to finally, just hiding the details entirely.
The best writing advice I ever received was this: When you are writing for a general audience, any time you find yourself agonizing over the correct way to word a certain detail, delete it entirely.
Extra details scare off casual readers and anyone truly interested in your idea can approach you for a further explanation. On top of that, it's nearly impossible to anticipate how details need to be described to get through to any one individual. If you let that individual come to you first, then you can tailor your response to the way they form their question.
I saw a lot of thoughts getting deleted and never re-appearing again, but I was too slow to grok the significance of his choice to remove them. It occurs to me a person could do some sort of git style display. Anything that's over 5 seconds old is revision 1, and if you edit that later, you see the old crossed out, and the new highlighted. (this would be a mode that could be turned on/off as desired, as it could be distracting) Such a mode would allow a window into the thought processes of the writer, both for the voyeurs, and for the writer himself.
Fascinating. However, don't think you are seeing the complete creation of an essay from its spark to completion. The first sentence appears (to me at least) considered and calculated. I'm sure much more thought has gone into the essay than the animation would suggest.
I've worked as a writer, and am currently in the process of writing a novel (the first I actually think someone would pay a penny to read) so I was actually about to comment that it does look like the complete process.
I'm sure a video of me writing would appear very similar. It can take me a long while to think of that first sentence, sometimes even a few days, but once I've got it in my head it usually comes out onto the page near-completed. Frequently I hit the first key and don't stop until 200 words later. However, it's once you get past what's sitting in the buffer in your brain that things drastically slow down.
As a slight aside, not knowing your personal history or the grades you got in high school, it's quite rare to see someone with a clear and (what appears to be) a natural writing talent coming from a programming background. From what I've seen, programming has a propensity to decrease literacy skills. It's especially impressive that it wasn't your standard five-paragraph essay, some people manage to succeed at those and fail at anything a hint more complex.
What this needs is a chunk model - to be able to see what was written in one chunk, from the first insert up to the next delete, then to the next insert, etc. You'd be able to see more clearly the purpose of every keystroke and wouldn't miss the small word changes, while also not having to press "next" 60 times to see a sentence unfold.
Imagine a company routinely using EtherPad (really cool, BTW) to create documents -- in the process saving thousands if not millions of interim drafts.
Now imagine the company getting into a lawsuit. Some subset of N documents -- and of all interim-draft snapshots of those N documents that are still in existence -- will have to be screened for possible disclosure to the other side. (There are tools for partially automating this, but lawyers and paralegals will still have to individually look at many documents / drafts.)
In PG's case, there were 2,886 such snapshot drafts for just one document.
Makes me shudder just to think about the legal expense.