Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Using stylometry to find HN users with alternate accounts

stylometry.net

Post reply on HN
Print view View original

511–520 of 562 posts

Re: Show HN: Using stylometry to find HN users with alternate accounts

#511

Post by yyt554 »

stavros wrote: post #435
rogual wrote: post #391

Earlier quoted context omitted.

Funny thing is, it thinks I'm you, but it doesn't think you're me! https://stylometry.net/user?username=rogual I'd have thought this stylometry thing would be commutative.

I guess it's a multidimensional space, so you can have someone closer to you than me, but they aren't also closer to me than you. Basically, they're close to you, but on the "other side" of me, I guess?

Don't need multiple dimensions for that.

0.1, 0.2, 0.3, 1.0, 2.0

To 2.0, 1.0 is closest.

To 1.0, 0.3, 0.2 and 0.1 are closer.

Re: Show HN: Using stylometry to find HN users with alternate accounts

#512

Post by yyt554 »

costco wrote: post #469
interroboink wrote: post #314

This is one reason why I like legal doctrines such as "beyond a reasonable doubt." Even a 0.9 match in a tool like this could be a coincidence, if there are millions of users. But that won't stop people from casually believing "aha it must be an alt account", based on some anecdata. It's so easy for something like this to be turned into a tool for a witch hunt, targeting innocents.

But a 0.8 or 0.9 match and something like Tor usage could be enough to justify a warrant. That's why I'm not sure I want to open source the code because I don't want to normalize this.

Keep in mind the potential to create false accusations by fabricating similar looking accounts.

Re: Show HN: Using stylometry to find HN users with alternate accounts

#513

Post by b800h »

InGoodFaith wrote: post #215
sillysaurusx wrote: post #186

Earlier quoted context omitted.

It’s a little writing trick I leaned from (I think) Orwell. Any time you’re about to use a common metaphor, try to tweak it. You’ll catch readers off guard, which piques their curiosity. It’s a fun game, too. I wish I’d used “the cat’s out of the hat,” but I didn’t think of it till later.

What you are describing is also known as an eggcorn. https://en.wikipedia.org/wiki/Eggcorn

An eggcorn is a soundalike though, isn't it? Deliberately altering idioms to catch people's attention isn't an eggcorn IMO.

Re: Show HN: Using stylometry to find HN users with alternate accounts

#514

Post by implements »

jll29 wrote: post #216

The method used, i.e. to calculate the cosine of the two authors' word vectors, is poorly suited for stylometric analysis because it is based on a poster's lexicon and the word frequencies of each word, but ignoring stylistically relevant factors like word order. Also, the cosine of the vectors of word frequencies conflates author-specific vocabulary and topics; in other words, my account is grouped (with >51% simila…

There’s https://en.wikipedia.org/wiki/Idiolect :

”Language consists of sentence constructs, choice of words, and expression of style. Accordingly, an idiolect is an individual's personal use of these facets. Every person has a unique idiolect influenced by their language, socioeconomic status, and geographical location.”

Re: Show HN: Using stylometry to find HN users with alternate accounts

#517

Post by chucksmash »

Alone out here in the 0.30s. the three times I've used a throwaway account, they've been for a single post on a single topic, so no surprise they did not get picked up by this analysis I guess.

Does a low correlation with other users imply higher susceptibility to de-anonymization if I were using alts regularly?

Re: Show HN: Using stylometry to find HN users with alternate accounts

#518

Post by b800h »

Well, interesting. This is one of the reasons we have the GDPR. @costco, if I were to make a GDPR erasure request, would you service it?

And I'm no lawyer, but it seems like there's also an outside chance of a breach of section 171 here as well, which is a criminal offence committed by a person who reidentifies de-identified data.

Plus - the laws have extraterritoriality. Vanishingly unlikely that you'd actually be pursued for it, but it's worth bearing in mind when you munge people's personal data.

Re: Show HN: Using stylometry to find HN users with alternate accounts

#519

Post by eps »

b800h wrote: post #518

Well, interesting. This is one of the reasons we have the GDPR. @costco, if I were to make a GDPR erasure request, would you service it? And I'm no lawyer, but it seems like there's also an outside chance of a breach of section 171 here as well, which is a criminal offence committed by a person who reidentifies de-identified data. Plus - the laws have extraterritoriality. Vanishingly unlikely that you'd actually be p…

It's an EU law.

Re: Show HN: Using stylometry to find HN users with alternate accounts

#520

Post by b800h »

eps wrote: post #519
b800h wrote: post #518

Well, interesting. This is one of the reasons we have the GDPR. @costco, if I were to make a GDPR erasure request, would you service it? And I'm no lawyer, but it seems like there's also an outside chance of a breach of section 171 here as well, which is a criminal offence committed by a person who reidentifies de-identified data. Plus - the laws have extraterritoriality. Vanishingly unlikely that you'd actually be p…

It's an EU law.

With extraterritoriality. And if identifying people in this way is covered (I'm not a lawyer, I'm not claiming it definitely is), then it's also possible that EU citizens using the tool are committing a criminal offence.

The law seems to only apply where the deidentification has been made by the data controller, but HN admins changing someone's username, for example, if they ever do, would count. A person then using the tool to match another non-anonymous username to that account would seem to be caught.

Important to stress how much of a technicality this is, but that sort of thing can be interesting sometimes.

Post reply on HN