Comment by rogerrogerr
2 days ago
> But people who accuse any comment longer than a paragraph of being LLM content are asses.
Many humans write good, lengthy comments on HN. That's not a heuristic. A few of the tells, none of them length:
- Unreasonable confidence (okay, fine, this is not rare on HN, but this is exuding a certain... omniscient vibe... that HN human commenters usually don't)
- Rule of three (sleep, meditation, relationship to thoughts).
- "X - not <...>, but <...>"
- "They still come - they just lose their grip"
- Emdashes are low signal but they contribute
There are many, many long comments on this page that do not trip my AI detection network. Why would you assume it's "anything long is AI" instead of "anything that smells like the shit that is all over Facebook these days is AI"?
FWIW, I scrolled through https://news.ycombinator.com/threads?id=porkloin and none of your comments strike me as AI, despite being long and well-written.
Do you think the root comment is _not_ AI? If it is AI, how is it providing any value at all in this conversation? Any of us can get the same info for free from $AI_PROVIDER; I come to HN to see smart _humans_ discuss things.
> omniscient vibe... that HN human commenters usually don't
au contraire!
HN commenters are notoriously overconfident of their (our?) knowledge, especially in areas outside our expertise.
> Do you think the root comment is _not_ AI?
I don't think it is AI, and more importantly I think the random accusation of "you are an AI comment" is a shitty thing to say. It's impossible to prove a negative.