But yeah, I'm with you mostly on the relays and also the good work Jumble has done to help visualise what's possible there. I still think though that the whole thing is mainly a use case issue. Change the use case to B2B and many nostr problems just go away, you no longer need the things nostr doesn't have (like how after linux moved away from consumer desktop OS and towards servers it no longer needed a sexy GUI which it couldn't figure out how to make anyway), and with lots of money to be made as SIs, and no grants needed. There's no other use case where the potential income goes up by 10 and the crazy-difficult problems go down by 10.

It would already be bad if they only used a "global" server for content discovery, but no, they also use that for the normal feed building anyway, instead of the outbox model.