Can they be built for scale or is it an inherent limitation of the protocol?
If it's a limitation of the protocol wouldn't client caching solve the issue?
I see Primal getting heat for caching but it just feels sensible to me. I wonder what I am missing.
Primal forces you to use it's Cache relay or it don't work, overriding what you want to see from relays. And if you have your own relay,
Most Nostr relays just were designed poorly. ditto is probably the best because it has admin controls and uses Postgresql which can use a memory cache.
Most decentralized tech in this space uses postgresql+Redis to scale to many users.
This is not the case for Nostr.
Sounds like we've got to keep building then.
