As a somewhat neurodivergent person who can’t stand twitter-style interactions because of the difficulty of pulling gems from out of the noise.… I would love it if someone built a bot that harvested the “best of rationality and science adjacent twitter/bluesky/mastodon” and put it into a simple website. Maybe using LessWrong’s codebase?
I feel like I’m missing out by being unable to comfortably read Twitter.
Some thoughts on this:
it could have a filter such that multiple semantically equivalent comments would always get merged into a single comment with an author-count
rude or non-sequitor comments would get automatically deleted
comments which have no info, just “applause” or “disapproval” would get turned into agreement/disagreement votes
All of this filtering seems like it could be adequately handled by a small-cheap LLM like Haiku or Gemini Flash in combination with some simple scripts.
As a somewhat neurodivergent person who can’t stand twitter-style interactions because of the difficulty of pulling gems from out of the noise.… I would love it if someone built a bot that harvested the “best of rationality and science adjacent twitter/bluesky/mastodon” and put it into a simple website. Maybe using LessWrong’s codebase?
I feel like I’m missing out by being unable to comfortably read Twitter.
Some thoughts on this:
it could have a filter such that multiple semantically equivalent comments would always get merged into a single comment with an author-count
rude or non-sequitor comments would get automatically deleted
comments which have no info, just “applause” or “disapproval” would get turned into agreement/disagreement votes
All of this filtering seems like it could be adequately handled by a small-cheap LLM like Haiku or Gemini Flash in combination with some simple scripts.