I wonder what's a reasonable threshold for "probably the same person". I've never had an alt on HN, and when I searched myself, it found 3 other users above 0.6, none of whom I've ever heard of before.
0.6 is not high enough to indicate an alt
Show HN: Using stylometry to find HN users with alternate accounts
321–330 of 562 posts
Re: Show HN: Using stylometry to find HN users with alternate accounts
#322How much should we fear de-anonymisation ? A lot of discussion on the thread are over "how can we prevent this". I would like to know why should we not embrace this and similar technologies? The benefits in my view are large - online behaviour tracks back to real life - and epidemiology speaking the value of millions of test subjects across every question are invaluable - from traditional medicine to "mass psychology…
>online behaviour tracks back to real life This is good to you? Okay, let's just make it like China or SK where your login is your citizen ID and if you write bad things the bad word police will take you away. Also, no, I have no alts.
So I am asking because my views are only challenged inside my own head, hence the need for external thoughts.
But firstly the "governments will come and do bad things" argument - yes this is clearly and obviously a major problem - but not one solvable by technology in anyway. Fixing violent dictatorships is a IRL problem - one that requires enormous effort and sacrifices (see Ukraine for obvious example). We cannot pretend that a browser extension or a ground up rewrite of Twitter will defeat Putin or would have stopped Hitler.
As for "free" countries (something like 120+ have open free elections), we still have online abuse for voicing opinions that some people don't like (anything from pro/anti Trump to LGBT and bitcoin etc). Those are real consequences but rarely government inspired and honestly I suspect we need better support for police in prosecuting such things - I mean a death threat is a death threat.
In general my view seems to be we should have the same protections online as we do offline - and if those protections are "in theory only" that requires us to use our voting and other political power to chnage it - not to obfuscate IP addresses or so on.
The upside of tech is so great it is worth spending IRL to defend agains the downsides
Re: Show HN: Using stylometry to find HN users with alternate accounts
#323Earlier quoted context omitted.
Maybe to avoid attracting any extra attention to this user? Also, as someone who’s read HN for a few years, it only took me 2 guesses to find an account that the above comment describes (and not necessarily the same person).
It was a classy move by jsnell, too. Thank you. (I don’t know who the comment is talking about, which is how it should be. There’s no need to blow someone’s cover in a highly visible way. Even if they were satan, they’d still be welcome on HN as long as they’re writing substantive, interesting comments that follow the guidelines.)
Such quality comments would track with most thorough Satan representations.
Re: Show HN: Using stylometry to find HN users with alternate accounts
#324Funny thing would be to find most unique user account stylistically.
Which user has lowest best match?
Mine is 0.58 so I'm really not that unique.
Re: Show HN: Using stylometry to find HN users with alternate accounts
#325Very interesting, .59 is my lowest, .64 is my highest match, none of these accounts are one of my alts. Though to be fair the handful of times I've used a throwaway I used it for a single comment so I didn't give it much to go off.
Re: Show HN: Using stylometry to find HN users with alternate accounts
#326> ... This site works primarily by analyzing for each user the frequencies of the most common words and phrases in the English language. Accordingly, the easiest way to avoid being identified is to simply use different words than you ordinarily would when writing. More sophisticated models than the one I made can use punctuation, comma usage, and capitalization to identify you so try alternating those as well. Services like Quillbot can help with you this but depending on your circmstances you may not want to send your writings to a third party service.
HN offers many other threads which could be tied together, including:
- time of posting
- ratio of replies to top-level comments
- comments being mainly upvoted or downvoted
- sentiment (mostly angry, dismissive, questioning, etc.)
- most common topics (keyword analysis of post being replied to)
- ratio of new posting to post replies
- first-to-comment on a post
- lone comment on a post
- etc...
It seems very likely that sooner or later every pseudonym for posting content will get discovered and linked. The lesson here is don't post anything that would cause you undue shame or harm if linked directly to your legal name.
Re: Show HN: Using stylometry to find HN users with alternate accounts
#327After a few tries on boring accounts, I thought to try the account of somebody who was notorious for an incident outside of HN, and had a (deservedly) bad time at HN for a couple of years before the account went dark. And yeah, there's a bunch of high confidence (.6-.8) hits for that account, and from a quick browse of the comments of the recently active ones, they look really likely to be alts. Like, all three that…
MD5 of the username is 9abc27e93b7e3c04b7c599017c1cfe5f ? The top one seems an odd one out in that case?
Usernames aren't random enough to be safe as a simple MD5. Perhaps with a strong bcrypt, but similar to PIN codes, it might be better to give partial information like "is the second character an ...", assuming nobody else made similar statements. Or give the first ~two hex characters of the hash, so that it would match 1/(16²)rd of the usernames. I'm sure there's also a clever way for a zero-knowledge proof here, probably something with diffie-hellman using the name as your random integer or something, but I'm too sick to think about this stuff right now. Privately sharing data publicly is hard.
Re: Show HN: Using stylometry to find HN users with alternate accounts
#328I’d like to request the author takes this offline please until the implications can be thought through. This is breaking anonymity that people incorrectly thought would not be revealed. For some it might be awkward, others it might be quite problematic.
This is nothing new, e.g:
Analyzing stylistic similarity amongst authors
https://news.ycombinator.com/item?id=10050603
http://markallenthornton.com/blog/stylistic-similarity/
37 points by lingben on Aug 12, 2015
Re: Show HN: Using stylometry to find HN users with alternate accounts
#329Earlier quoted context omitted.
> Can you clarify what this means Imagine that for every new comment you want to post you would create a brand new account which you would use precisely once and never again. Then the stylometry would have just a few words and wouldn’t have enough corpus to get a reliable signature. If a lot of people does this it would be hard to figure out which account belongs with which human. ( Of course if you alone do this, yo…
At the same time, HN doesn't let you delete comments. Maybe with some GDPR magic.
Not sure what is your point, or how does that connect with my comment. Care to elaborate?
Re: Show HN: Using stylometry to find HN users with alternate accounts
#330Earlier quoted context omitted.
> but this functionality turned out way creepier than I thought the moment I tried it Hopefully this raised awareness means that people who actually need anonymity will be more likely to know to take precautions.
Quoted post unavailable.
You know everyone going to put your username in that tool after a rant like that.
If ever there were a good use for a throwaway account I’m thinking this is it…