It makes me wonder if we could use non-instruct LLMs to slightly alter the wording of text while keeping the meaning the same meaning. Perhaps by using perplexity or some other metric. I don't know, maybe you compare the distance of the "meaning" vectors.
You might also want to have some "style" vector associated with each pseudonym. For example, I might want it to produce british english under a certain pseudonym, and simulate an ESL speaker under another.
Essentially, you would want some way of re-styling text. The basic way to do this would be to run the same sylometry tools the hunter uses and manually make synonym word/phrase subsitutions to lower your similarity.
It's a cat and mouse game, but I think the mouse eventually wins. Consider a program that translates your english writing programmaticially into a low-entropy symbolic form and then translates them back to english in a procedural manner. Basicially you design an intermediary language that cannot contain style. It would be boring to read but it would remove all the style.
I still think people underestimate the power of even minor inconvenience. While you can't just click a button to reveal all pseudonyms of someone (e.g. you need to download some obscure tool or even perform statistical analysis yourself), I think this provides significant (and surprising?) protection for pseudonymous individuals, I'd say even (although to a reduced extent) for sophisticated threats like state-level actors. I hope LLMs continue to be unable to do so for the near future (I just tested and LLMs can't do it with a simple prompt).
Which is why I think privacy safeguards still work quite well even while being technically mostly bypassable.
So probably tools like this should be kept private if possible.
It is though.
> Do they, incorrectly, position their adverbial clauses? Underrated line.
Then during a presentation I was making today, a coworker wrote in chat "If <scubbo> says 'You're absolutely right' one more time, I'm checking his house to see if we're actually talking to an agent".
They got me.
As an aside, it's always surprising to see how English speakers split Greek words like "Apocalypse". That is to say, they always split them in the middle of Greek syllables, or just drop letters like "pseud[o][a]pocalypse" and often in a way that ends up sounding clunky and weird even in English.
Can't think of other examples now. Brain going to sleepzzzz....
Edit: oh wow there's actually a word for that:
https://en.wikipedia.org/wiki/Libfix
OK, now I go to sleep.
It's not trying to say "something like an apocalypse" it's saying "an apocalypse of pseuds"
- Claude knows who you are: https://www.lesswrong.com/posts/Jkb4CBB7rf4XYP5eb/claude-kno...
- ~ Opus 4.7 is the first model to correctly guess who I am based on unpublished articles: https://x.com/KelseyTuoc/status/2044962428547695007
Does anyone have some tools to share?
Try ChatGPT.com
No just use a text-mixer: in goes your text, set parameters to have output match <scapegoat> (or just pick "standard_Neanderthal_3"), out comes text conveying the message you wrote, in the style of your choosing.
Of course that would also strip the attributes that made it your creation.
"boom", pseudoanonymity (spell?) restored?
Lots of criminals have been identified by reanalyzing old DNA samples using data and genealogical techniques that weren't possible at the time the samples were left.
By his brother, apparently. His brother read the manifesto, and said, “hey, sounds eerily familiar!”
But wouldn’t it be deliciously ironic if A.I. had caught Kaczynski‽
Similar to ultra-accurate multimedia geolocation models.
Also, I wonder about this analysis in the age of AI slop. I wonder how much that removes the identifying bits, vs how much carries through of the original prompt (e.g. topic and guidance). It's interesting that a pseudonymous blogs might take on very generic Claude-voice, which could be worthwhile if the topics were interesting, but could also just be a completely humanless bot.
The problem, as the article correctly identifies, is that this is just the tip of the iceberg. There are tons of other cases (e.g. identifying a car without a license plate by the scratches) that AI is going to enable ... and in those cases we can't just get anonymity back by using an LLM.
Forensic research, NSA, Palantir…
Btw 42. Sleep, eat, have sex, have fun, be useful.
See, I used "whom", messing up my fingerprint.
Sometimes I see people that consume and take only. (I am not judging, I just observe.)
Which quotes Tao on using deliberate disinformation to preserve anonymity.
> …one additional way to gain more anonymity is through deliberate disinformation. For instance, suppose that one reveals 100 independent bits of information about oneself. Ordinarily, this would cost 100 bits of anonymity (assuming that each bit was a priori equally likely to be true or false), by cutting the number of possibilities down by a factor of 2100; but if 5 of these 100 bits (chosen randomly and not revealed in advance) are deliberately falsified, then the number of possibilities increases again by a factor of (100 choose 5) ~ 226, recovering about 26 bits of anonymity.
Intentionally adding writing "tics", scheduling posts to appear between 2am and 6am in your timezone, or pretending to have a different gender/location/age should help a lot in staying pseudonymous for a while longer.
This is one of the best things I've read on this site.
I have a side project/experiment that's tangential to this (wafertown.com), so my interest is 2x the usual.
The proposal that seems to infringe on free speech, but mostly I take issue because there just isn't really a precedent for what such a thing would look like. What constitutes doxxing? It is easy to accidentally doxx someone by mentioning some offhand fact in conversation, because you may not know what series of facts can be connected to form a doxx. Typicially, people doxx themselves by accident, and someone else is merely pointing it out. I know I have many times. You live and learn.
There isn't really a right to privacy in the way that there's a right to free speech. Free speech is important to processes of public transparency and justice which I think this would interfere with. But it's also true that justice is blind and doxxing can interfere with a judicial process.
Also, if everyone is going to have access to super-stylometry tools in the future, which is the premise of the article that GP reacts to, it will be fruitless to uphold such a right because anyone can just run doxxyou.exe themselves. There's no need to spread doxx because it can be reproduced individually.
The term doxxing came from a certain hacker culture where it was implied that you connected some real-life identity to a criminal pseudonym. It essentially meant "snitching". So the idea that doxxing itself would be a crime is interesting. It's a reversal of the original meaning.
But it's also true that the public has discovered sybil-suseptible techniques (like swatting) where you can screw with ordinary people's lives by knowing their identity. Which is interesting because we live in a new culture of "share everything online" vs old the hacker ethos of "don't use your real name online". The attackers have become stronger and the defenders weaker.
Let's say I offhandedly mention your email in this comment, for example, like we had an out of band conversation somewhere else. Then, someone takes that email and links it to another forum where they find your phone number. Then another person looks up that phone number in a phone book. Then another person finds your github account from the email associated with your PGP and finds where you work etc. Let's say a comment you wrote 7 months ago vaguely mentions something about where you live, which in combination with the previous information narrows it down to a single place.
At what point does that become doxxing and who is responsible? Pseudonymous people nessisarially, slowly leak small amounts of information about themselves when on the internet in order to engage in communication.
It is only when you collect "enough" bits of information that it becomes threatening doxx.
You could say publishing this collection of facts is the doxx. But the point of the article is that stylometry and AI tools can do this investigation for you. Anyone can trivially assemble the collection themselves.
I think ultimately, it's a matter of personal responsibility. It has to be. If you have perfect opsec, I can't touch you AI tools or not. And if I have perfect opsec and I'm doxxing you, then no laws will be able to catch up to me.