rabble on Nostr: I think this is one of many great kinds of custom algorithms which would be great to ...

I think this is one of many great kinds of custom algorithms which would be great to have in Nostr. Bluesky has a feed called “quite posters” which are posts from people who don’t post too often, very similar to most recent post per person… The Meta algorithms have the opposite optimization, they’re making creators publish multiple times a day in order to get an audience. I love the idea of making it easy to see the quiet person who’s being lost in the flow right now of wonderful but enthusiastic posters like Derek Ross (npub18am…p424) and holokat (npub1r0r…q9ac). These and others are people I REALLY want to follow, because they’ve got good things to say, but they say a lot, and we need to be able to see the other folks too.

I’d also love a feed which was AI summaries of a person’s posts or a thread. Take for example this thread:

quoting note1qcw…wvnr

I just gave a friend an update on what Nostr is and showed a bunch of apps in the ecosystem. Showing how fast it is and how all of these apps use the same identity and data but are different in their focus. He has been around the space of social media protocols for years but he was blown away.

We’ve really got something amazing. We need to do a better job of show and telling the story. It’s not about just zaps or censorship resistance or the way we’re doing web of trust moderation. We’ve built something which works really well. It fulfills the dream of the early days of the internet, web, and Web 2.0. An open interoperable internet, by and for the users.

As an aside, yes bitcoin is part of it, but it’s so politicalized that some people dismiss nostr as that bitcoin social media protocol. We see people join nostr and through it discover bitcoin, that’s the path to wide adoption.

It’s got tons of replies and discussion. Rather than just giving me the chronological or most ‘important’ replies or replies by people with the highest web of trust score for each user, we could feed that whole thread in to an LLM and create a summary with links directly in to the comments / discussions.