User Relays should be "sharded", meaning we should all be put in a pool of relays together automatically by the Nostr clients.

Instead of each popular Nostr client hosting their own single large relays for everyone, there should be a pool to select from. Those should have some metrics API available to allow Nostr clients to consider migration away from certain relays if the become slow and unresponsive.

Nostria will host many relays and distribute user's across them, ensuring scaling will never be an issue.

I wrote about this here: