I feel like we need an "is this newly created account a chatbot" test.

Like, a turing test maybe.

Yeah, true, but the thing is, I already fail for most captchas. I don't know about @Lensflare bots, but my bot only did strong spam control on new members and of hundreds of spam messages there were just four false positives or so. So yeah, it's possible to make a 99% sure checker on that. But the thing is, if it's a bastard, what to do with him?

@tamagotchi I'm not just meaning spam though, like how hard would it be to just fine tune an LLM to rant about tech and insert it into this community. Are *you* a bot?

@tamagotchi no false positives yet, as far as I know.
If a user has more than 10 upvotes, they are not considered a spammer.
each keyword and each regex match has an associated spam score. Total score of 100 and more means spammer.
Old users which are recurrent spammers are being added to a blacklist.

Vetting would take hours of back and forth until the context filled up and it started repeating itself. Or maybe you have to copy paste a hundred thousand word block of text which would be trivial to a human but almost impossible for an llm.