Gossip and DVMs
DVMs are a great idea, but have some problems in practice.
* DVMs don't advertise where you can contact them (which relays).
* DVMs that produce the popular 5300/6300 feeds don't tell you where to find the events they recommend, they just give "e" tags without recommended relays or any other relay indications. At least this is true for the most popular DVMs ("Current Popular Notes DVM", "Top Zapped Notes").
* DVM discovery isn't quite right. NIP-89 is about clients that handle events, not DVMs that handle events remotely (ashodlbod (npub1jlr…ynqn) recognized). Are clients discovering DVMs by listening to recommendations from their followers? Are people making such recommendations to their followers? Or are clients hardcoding stuff? Or are they just load 31990 events from some popular DVM-centric relays and "trusting" them? I don't know, but I'm not convinced that we have solved the DVM discovery problem correctly.
For gossip to plug into the DVM ecosystem I have to compromise my values and work towards centralizing nostr upon some well-known relays where DVM activity happens, with hardcoded URLs and maybe even hardcoded DVMs. I don't intend to do that.
Instead I'll work with hodlbod on PR 1728 in the NIPs repo, while I work on the implementation on a development branch that probably won't merge for some time.
* DVMs don't advertise where you can contact them (which relays).
* DVMs that produce the popular 5300/6300 feeds don't tell you where to find the events they recommend, they just give "e" tags without recommended relays or any other relay indications. At least this is true for the most popular DVMs ("Current Popular Notes DVM", "Top Zapped Notes").
* DVM discovery isn't quite right. NIP-89 is about clients that handle events, not DVMs that handle events remotely (ashodlbod (npub1jlr…ynqn) recognized). Are clients discovering DVMs by listening to recommendations from their followers? Are people making such recommendations to their followers? Or are clients hardcoding stuff? Or are they just load 31990 events from some popular DVM-centric relays and "trusting" them? I don't know, but I'm not convinced that we have solved the DVM discovery problem correctly.
For gossip to plug into the DVM ecosystem I have to compromise my values and work towards centralizing nostr upon some well-known relays where DVM activity happens, with hardcoded URLs and maybe even hardcoded DVMs. I don't intend to do that.
Instead I'll work with hodlbod on PR 1728 in the NIPs repo, while I work on the implementation on a development branch that probably won't merge for some time.