Feels almost incredibly hard to swift through everything, even just to get embeddings of everything without classification.

Even Twitter / grok can only do work on your own social graph afaik. It would either need an insane amount of money / compute or a few more years but maybe I'm wrong and you can get away with just crawling through some popular npubs and you're good.

Reply to this note

Please Login to reply.

Discussion

It's a different problem altogether. I would start with the easier one which is filtering what I already have.

Yeah. I’m sure you can do some really simple heuristics (has profile, has posted in the last week, etc) to chop that number down a lot.