Thanks, this is a fascinating note.

You’re right that really it’s collaborative production that’s the heart, and there’s no real benefit to LLMs.

It does feel like some classical explore/exploit algorithms could be interesting, but I imagine the challenge to any effort is really just getting enough high quality contributors