reply on: Making nostr P2P \ stacker news ~nostr

183 sats \ 8 replies \ @Catcher 1 Sep \ on: Making nostr P2P nostr

The main question will be - why?

Yep - for mostly feed-based social media, what does p2p provide better than client-server?

It’s more decentralized obviously but is it worth the trade offs given the application?

Could it be used to fill in/check for relay gaps? E.g. when someone you're following is online at the same time as you, you tell it what notes you have and ask if any are missing, and those can be sent directly.

50 sats \ 0 replies \ @k00b 1 Sep

Sure, I think for augmenting relays a dht be cool or for discovering relays/nostriches in the first place or as a place for relays to crawl and check if a peer is available to be read from.

Otherwise, I'm not sure it'd provide a twitter-like experience with online/offline/roaming/behind-nat peers.

Nostr works pretty well today because it emulates the way most applications work - notes are concentrated and duplicated to a few places and clients read from them. It's a lot like ActivityPub but with keypair ids. Moving beyond that model will require some pretty radical and practical insight that no one has managed to produce and scale yet.

0 sats \ 4 replies \ @BallLightning OP 1 Sep

Because it's cool. Besides you can use it simultaneously with the standard server based relays. Feed based social media features can be achieved by some kind of decentralized indexing, which I said in the original post don't know if they are really possible or fast enough.

Another cool thing would be a p2p global git repository btw :)

0 sats \ 3 replies \ @k00b 1 Sep

Cool to say and cool to use more often at odds than not, but I agree it'd be cool.

0 sats \ 2 replies \ @BallLightning OP 1 Sep

Well... would saying that relays being easier to block than a true p2p connection count as a valid argument :)

0 sats \ 1 reply \ @k00b 1 Sep

It's a valid argument, but a social protocol that isn't very easy to use won't be very valuable.

0 sats \ 0 replies \ @BallLightning OP 1 Sep

I think end users won't see this so it will be as easy as it is now (unless something that is possible now is impossible in pure p2p architecture)