enjoying the podcast so far, my thoughts on fixing nostr relay performance and content discovery i think we need to change to only writing to 1 to 3 relays and maintain a complete copy of all our events there, this means having local backups too to restore at the relays, we then need a dht which references likely just one of these relays to our npub to keep the dht small as the dht will hold all npubs known, this would open the door to p2p even more aswell where now clients with full local copies could fulfill content requests with the dht changed to facilitate p2p connections