Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.
- 139
- 1
Jump in the discussion.
No email address required.
Notes -
But ‘identifies 2rafa’s writing as 2rafa’s writing’ is a big difference from reading your comments on instragram or company PR releases and linking you to known dissident 2rafa. I’m not saying it’s impossible, but doxxing someone is an order of magnitude harder as long as you don’t have a very particular stylistic tell.
That’s true. The latter case is more complicated. Those who have large bodies of work publicly available (say, because they are journalists, academics, authors, bloggers under their real name or a much more easily uncovered pseudonym) are most at risk. But it’s not just them.
Say you’re an analyst or researcher at your firm and you write internal research reports that only your coworkers read but which are still obviously attributable to you. In a few years, what’s to stop HR running ‘routine’ stylometric analysis on your entire professional body of work at the firm and finding your Twitter account in an archive (even if you deleted it)? You don’t even need to write for a living, they could run it on your emails! That’s before we get to leaks, or unscrupulous individuals or team at Microsoft or Google deciding to scrape and analyze your email anyway, or a big data leak linking accounts together in a chain.
I do think this is different from ‘the end’ of online privacy. Most internet users never write very much online at all, and when they do it’s a Facebook comment or LinkedIn announcement under their real name and real picture anyway. Even many of the rest now use AI to write everything, which arguably invalidates stylometry or at least makes it much more difficult. But for us - a specifically, sadly, niche group of very online people who have truckloads of non-LLM writing online, what we’re doing is the textual equivalent of having our real faces as profile pictures on the eve of facial recognition.
I am not hopeful.
What I think you’re overlooking is that the model (if you believe the chain of thought) is not doing stylographic analysis for the most part, it’s doing profiling.
Broadly:
British-presenting / ethno-nationalist / writes at length in a cultured register / HBD believer / argues for the ideal of the gentleman-scholar and a leisured aristocracy / argues from utilitarian logic therefore likely rat-adjacent + some other stuff-> 2rafa.Assuming you don’t put these convictions in your hypothetical internal research reports, I would expect it to be orders of magnitude harder to identify you.
It’s interesting that it calls me an ethno-nationalist. I can’t recall every comment, but I’ve always said that while I’m sympathetic to the earnest motivations of many given the mainstay intertribal violence has been throughout human history, I’m not an ethnonationalist and I do think functioning multiracial countries are quite viable when properly governed and when care is taken with their composition.
Ah, no, sorry, it calls me an ethno-nationalist. The quotes were mine, you’re just the closest person in broad profiling terms whose name it knows.
It thinks you’re a nostalgist with a “somewhat rosy view of pre-war gentleman-scholar society” or words to that effect :P
You can take the AI out of San Francisco, but…
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link