Now imagine that account was linked to a SIM. It's trivial for a nefarious actor to get it re-activated, infact there was a video by Veritasium just today where they didn't even need your SIM.

But even if they are not that hi-tech, it's not that hard to get a SIM issued in your name, or other hacks of a similar nature, we have all heard of stories.

Worse, you lost that SIM a decade back, the number gets back into the queue, and is eventually re-issued to someone new... and they try to create a facebook account, and are presented with yours.

They can then re-activate your old facebook account, and post a video/audio/text of "godelski" saying they like pineapple on pizza. and before you can defend yourself, the pizzarias have lynched you.

(I dare not use a real example even as a jest, I live here)

Are you 100% sure of all your old social media accounts, all the SIM you have ever used to log-in to accounts?

We leave a long trail.

While it is impossible to take back a disclosed secret, we can create a legal framework which issues immediate business-stopping corporate death sentences for spreading that data without consent (or after you revoke consent, yes it needs to be revokable). Your data is so valuable, because now that we have universal function approximators you can be simulated (to a prediction horizon) and that simulation interrogated.

Of course, this could be misused to post something with plausible deniability, but if you want to say something controversial, why wouldn't you make another account for that anyway?

I know that one could theoretically sign posts with GPG, but it would be much nicer and less noisy if sites would have UI to show something like: Signed by <fingerprint>, key used for N years.

One issues is that most social media want your identity to be the account on their service and not some identity (i.e. key) that you control.

  > There should be a way to cryptographically sign posts (everywhere).

Key signing will not be the solution because it isn't addressing the problem. It exacerbates it.

Ah, think I see your point. Your worry is that language use, etc. could be use to deanonymize you, by correlating with text that was not written anonymously. But that's a separate issue from voice or writing style cloning to pretend it's you that said it. In the latter case you could use a pseudonymous signing key?

I agree that deanonymization is an issue that is hard to tackle. I wonder if someone studied how unique writing style is. E.g. browser fingerprints are fairly unique, but I wonder to what extend you can filter a person from, say a pool of 100 million, using writing style alone (grammar, vocabulary use, etc.). I guess it becomes quite easy if you engage in a lot of domain-specific discussions and use their vocabularies.

E.g. if I'd talk about Marlin-kernels here, you could probably narrow me down to a few hundred people. Throw in another comment about the Glove80. Maybe ten people at most?

  > I wonder if someone studied how unique writing style is.

Your language is in fact a fingerprint. And like you suggest, topics too. Much of our freedom of anonymity comes from the fact that it is hard or not worth it to dox people.

I do agree that verification is a different issue though. I'm not sure keys will solve it because you're not going to sign anything that is scandalous, so it might even give evidence for those that want to falsely claim foul play. And how do you sign a leak?

The problem with signing is that it seems to work for the cases we don't care about and do nothing for the ones we do. That is unless we sign literally everything, including our voice, but then you kill anonymity (why I connected it) and you could then probably clone that too.

I suppose even a throwaway could be linked to my identity if a comment was long enough, but probably only with some limited certainty.

We're human, we all have stuff to "hide". I might want to vent about my boss or work and if they heard it that would certainly take it different than it was intended. We even complain about our friends, especially our close friends. Because no relationship is without conflict. But being human we often have to actualize our feelings with words, because we're social creatures. I'll at times complain about my partner, but that doesn't mean they also aren't my favorite person in the world.

To be human requires some anonymity. And no one should feel like they're going to be scrutinized for every little thing they say.

Not a great outlook though, if everybody does this...

Besides, topics and shared details are also pii. It's more that they aren't useful in that way until you have scale.

Plus, on HN you can't edit posts after some time.

at this stage and current behavior, I wouldn't mind some poison. That's the only way companies will learn at this point.

Yes this violates any EU citizen's right to be forgotten under GDPR. Welcome to silicon valley.

Sort of moot point considering the multiple HN archives that would still have the original username attached

Under which legal theory does EU law apply here?

I could be hallucinating, but I swore job boards also had peopel hire from the EU.

  > You can't nuke your account. You can close it but your comments remain on the site. 

If stylometric analysis runs on all comments on the internet then yeah.

Bad things will happen, very very bad.

I honestly think it should be at least illegal to do this kind of analysis because it'll be a treasure trove for the commercial sector to mine this data correlated to real people not to think of the destruction in millions of people with personal anonymous blogs etc.

Actually thinking about it further you could also easily group people political affiliations, and all kinds of other thoughts, dark, dark stuff!

Meanwhile gangsters or nation states are probably already working on automated deanonymization and blackmail at scale, but will target the dark web before the public one. Not sure any of this stuff is that changed by the advent of deep fakes and llms tho, probably just boring classical statistics does ok?

But I'm pretty sure this would make life very bland. Slow down innovation as we become less creative, less willing to deviate from the norm, and the Overton window of what's socially acceptable narrows. After all, we do love having enemies in some way or another, even if it means making them up.

So far I get a lot of pushback and statements like "it's just linear algebra" and "quality doesn't matter, it's quantity. Just scale. That's the bitter lesson". The former tells me they know very little linear algebra and the latter tells me they know very little about probability distribution and the concept of coverage... I routinely see papers get awards and tens of thousands of citations when they're making errors I would expect undergrads in my classes to do better at... I think a big reason I'm frustrated is that supposedly we share the same goals -- building AGI -- but I'm told I'm stopping them from doing so while they reject my papers and theirs gets accepted... (Fwiw I'm happy to accept scale papers. There should be a wide range of ideas explored. But I'm upset that we're on a railroad and any deviation from that is considered heresy)

Already happening at Google and Meta for a decade or more.