Actually, the more I think about it. The classic gossip protocol (as the wiki article lays it out) is a very bad outcome for Nostr. This is basically what Blastr is doing. Any event it sees, it tries to send it to all other known relays. We don't want data consistency across all nodes because we don't need (or even want) consensus. We want users to have control over their data. We want clients to do two things: 1. Be aware that many relays exist and know that they need to use several strategies to understand how to find a user's data. 2. Intelligently calculate how to fetch the right data from the right relays in the most efficient way possible. The other important thing we want clients to do is to have an idea of how to find content/people when it doesn't have any hint on where that data exists. This mostly isn't an issue right now, because we have a small-ish number of large relays where you're very likely to find someone. Basically, it's about having a good fallback strategy.

1. Now several large relays have provided information retrieval services. 2. It doesn't seem to be a problem right now. I have been watching everyone discuss the problem of big and small relay these days. Confused, actually. I don't think that's a problem. Nostr users do not need to control their data, users only need to control their accounts and keep their private keys. The account belongs to him. Because all the data has to be signed by the private key before it belongs to this account. While the private key is unique, the data can be copied and pasted. The user's post data can be used and stored by anyone, in order to decentralize the storage of data, the more relay stores these data, the more secure. Whereas censorship is all about blocking users' accounts, the great thing about Nostr is that it can't block anyone's account. You can even set up your own relay service to broadcast outside, as long as someone is willing to follow you, they can receive your private key signed events.