News for hackers. Stuff that matters.
Re:Ask HN: How many of you founders are girls ? by jacquesm on Tuesday March 16, 2010 @07:05PM (#1197076)
> written with a one time account

Smart move, but I'll bet you I can still identify you.

One really big eye opener for me how subtle this problem really is was this article, I've just posted it as a separate submission:

http://news.ycombinator.com/item?id=1197072

Re:Ask HN: How many of you founders are girls ? by onetimetoken on Tuesday March 16, 2010 @07:43PM (#1197168)
I was just trying to empasize my point. Just out of curiosity, how would you go about identifying me?

Thanks for the linked article. Interesting indeed.

Re:Ask HN: How many of you founders are girls ? by jacquesm on Tuesday March 16, 2010 @07:55PM (#1197205)
By taking the vocabulary (with frequency) of your comment and matching it against the vocabulary of all HN users.

It's a pretty large sample of text and I'm fairly sure that it is large enough that the match will be a close one.

Then there is 'stylistic' analysis, another angle of attack, mostly to do with the way you write, not just the words themselves.

Re:Ask HN: How many of you founders are girls ? by onetimetoken on Tuesday March 16, 2010 @08:34PM (#1197285)
Thanks, that would be my first angle of attack as well. The assumption that I have at least one other hn account is reasonable enough. Even when I wrote my first post, I was considering faking a different style/vocabulary (which is harder than it seems), but obviously, I was just using a throwaway account as a rhetorical device for emphasis.

Thanks for the exchange. :)

Re:Ask HN: How many of you founders are girls ? by jacquesm on Tuesday March 16, 2010 @08:52PM (#1197322)
Ok, so I've run my analysis, my first guess was wrong, we'll see how many it takes to get it, if ever.
Re:Ask HN: How many of you founders are girls ? by seasoup on Wednesday March 17, 2010 @04:55PM (#1199625)
Quirks I noticed in the text to help with identification:

"fully agree"

comma followed by etc with a period in middle of sentence , etc. can

use of e.g. with a period after each letter.

well-known with a hyphen high-end with a hyphen likes to use colons correctly.

no misspellings. odd. proper use of punctuation. strange. I'd say the writer fails the Turing test and is a robot.

Re:Ask HN: How many of you founders are girls ? by Splines on Wednesday March 17, 2010 @05:22PM (#1199691)
Also, odds are that someone wouldn't bother logging out and in between two accounts. I'd weight your search against people who commented in between the time that the throwaway account posted.

People who commented shortly before the throwaway account are also good candidates, as are people who seem to comment around the same time on other days.

Re:Ask HN: How many of you founders are girls ? by follower on Wednesday March 17, 2010 @11:59PM (#1200412)
> odds are that someone wouldn't bother logging out and in between two accounts.

Unless they use multiple browsers.

Re:Ask HN: How many of you founders are girls ? by dauwkuk on Wednesday March 17, 2010 @05:22PM (#1199693)
Well I fully agree with this comment i.e. I recognize the "quirks" too. It looks to me like the results of a U.K. education. As a Brit-born male, in my past I have had some non-Brit ladies comment about the robot...
Re:Ask HN: How many of you founders are girls ? by pbhjpbhj on Wednesday March 17, 2010 @07:15PM (#1199937)
I disagree on the UK edu unlees they're also a non-native - http://news.ycombinator.com/item?id=1199898
Re:Ask HN: How many of you founders are girls ? by Terretta on Wednesday March 17, 2010 @06:04PM (#1199815)
> no misspellings. odd. proper use of punctuation. strange. I'd say the writer fails the Turing test

That's depressing.

Meanwhile, "[space]-[space]" rather than "--" for "—" is idiosyncratic, and since HN preserves original spaces in comments, ".[space]" rather than ".[space][space]" rules a few more of us out.

Re:Ask HN: How many of you founders are girls ? by robin_reala on Thursday March 18, 2010 @07:23AM (#1200984)
Meanwhile, "[space]-[space]" rather than "--" for "—" is idiosyncratic

It’s not idiosyncratic, it’s British.

Re:Ask HN: How many of you founders are girls ? by HeyLaughingBoy on Wednesday March 17, 2010 @05:37PM (#1199729)
Except for "fully agree" and the hyphen bit, that would generally characterize my writing, if only because bad grammar & spelling bug me. But it sure ain't me :-)
Re:Ask HN: How many of you founders are girls ? by petercooper on Wednesday March 17, 2010 @06:37PM (#1199861)
if only because bad grammar & spelling bug me. But it sure ain't me

Heh. I see what you did there.

(Updated: "&" usage, "ain't", starting a sentence with a conjunction.)

Re:Ask HN: How many of you founders are girls ? by petercooper on Wednesday March 17, 2010 @06:48PM (#1199879)
I had an interesting HN corpus a while back but is there something newer/cooler you're using? Or did you just put it together yourself?

BTW, I strongly suspect stylistic analysis would work a lot better given the small corpus available for each user, though that idea goes out the window when someone deliberately tries to mask themselves.