Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?
This is your opportunity to ask questions. No question too simple or too silly.
Culture war topics are accepted, and proposals for a better intro post are appreciated.

Jump in the discussion.
No email address required.
Notes -
https://www.themotte.org/post/3671/culture-war-roundup-for-the-week/430737?context=8#context
Claude managed to peg me again, without web search, and with material that is 100% past the knowledge cutoff. I shared the comment above (it's very recent), and asked it:
It's COT trace immediately identified that it was someone in the Ratsphere, then that this was likely an excerpt from the Motte (!) and that it immediately considered me, self_made_human as the likely author.
The relevant parts of its reply:
This was n=1, a hit on the first go. I've never posted on DSL, but that's the only factual inaccuracy here, and it's still 99% on the money. And no, I don't have memory enabled, or any personalization that would give the game away - even if I did, I wouldn't explain that I'm self_made_human there.
AHA. Turns out I was not wrong a year ago when I said people would freak out a bit once they realized that LLMs could do this, just early.
https://www.themotte.org/post/1321/smallscale-question-sunday-for-december-29/282154?context=8#context
Hah. I was talking about truesight before it was cool. As proof, here's me talking about when o1 came out, and I definitely spoke about the deanonymization risks first, and I'm pretty sure it came to my attention in the GPT-4 days (no, you can't see that one, that comment goes to another forum in Canada). You can mail the internet points to my door (Opus knows the address).
More options
Context Copy link
Being early is indistinguishable from being wrong. For a while. Then it turns into being technically right, the best kind of right.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link