I don't dislike the idea of algorithms, I just never found one that really worked. Not even the YouTube algo works for me.

Follow-based feeds work. But it gets old quick. The secret is finding new communities to belong. Not necessarily in a NIP-29/Discord way. But communities nevertheless.

When you build a follow list, you are essentially assembling a community for yourself. You mark the people you want to engage with. Nostr is still a single community.

Maybe Clients need more of that "Follow list building magic". Or maybe communities will be formed around apps, like Olas is doing right now. Maybe we need different nsecs for different communities...

Maybe the whole point of Nostr identities is to build, discard it and start again.

Maybe multiple identities is the natural state of life. And Nostr is just the vehicle to make them digital...

Reply to this note

Please Login to reply.

Discussion

👍

Communities aren’t static either. They shape each other.

True.

I think if there were 'interest categories' and you could select things that interest you, that would give a good feed.

AI could make suggestions for you to accept or reject..

I dunno TikToks algo is pretty good. You can still view just people you follow it’s not an either or thing imo.

Global follow lists suck. Simple as.

How Nostr solves for this, is entirely the point of such a protocol.

We need to be able to:

1) Learn where users can be found

2) Remember where users can be found

3) Assign categories to both friends and lists of friends (as well as locations- i.e. relays)

Global follow lists don't aim to achieve these goals.

I also don't believe NIP-29 solves for this.

Relay.tools, nip-42, and nip-70 could very well solve for this, but I won't waste my breath trying to explain it unless someone wants to hear it. BlastrNet sucks in my opinion.

What about algorithms that allow you to see interesting posts from people you follow that happened when you were busy doing offline things?

Without an algorithm, those notes are effectively lost to the sands of time.

Facebook before 2012ish worked well for me.

Different follow lists for different purposes is like having different keypairs but without the friction of having to change between them. On Amethyst it would be 2 presses instead of 5 possibly. Instead of Follows and global you'd have your custom feed you have named for yourself. For me it would be "Nostr devs only" and "Nostr devs + bitcoin people". I've tried to solve the information flood problem by having two keypairs with little success.

I realise now I can do exactly this with kind `30000` on listr.lol

I can't believe it took this long for me to figure this out

My initial hope was that something like listr.lol would become the standard and I wouldn't need to code on Amethyst. But I think I still need to do it :(

Yeah I'd really like to add people to my kind 30000 follow sets directly from Amethyst

list management has to be integrated into the main client so it's effortless

youtubes algorithm is absolutely horrible! I think it tries to make me use youtube less! I like nostrudels blind spot feature to help find other people to follow and lurking on other people's lists.

in "the network state" balaji introduces the mimetic code - the idea that our commitment to social groups can be quantified.

as we live in multiple circles (example: programming, bitcoin, hiking), we create our unique, multi-layer code.

zaps w/ their corresponding events could be used to calculate one's code, and from there, maybe some zaps-based algos are the way to improve things.

thoughts?

what clients need to do is deprecate kind 3 and support named follow lists, and use algorithms to filter out lower quality notes

I am building this at https://nostr.thecaptain.dev

Yep, we to that as well. But building lists is annoying. I wish a client would build them automatically.

I think that's where the grapevine idea can come into play. after a few people bootstrap some lists, clients could that automate some of the list generation process

With nostr, anything is possible. We could have algorhythms weighted for more sustainable, pleasant moods and builders people that build better communities, rather than dopamine addiction weights.