You are effectively saying that Nostr can be anything, but it can't, not unless the clients with most market share options in, but if that is to be taken for granted, then Nostr can have Pkarr IDs and Peergos data model etc. I wish that was true, but you can't have both decentralised development and agility at the same time, the more independent clients the more ossified the development becomes, this is an old story since Email failing to get any end to end encryption, while Signal doing it in few years.
Nostr could do some of the things Alex says Bluesky does better:
1) We can create a new kind that replaces kind 1 which has spans (or facets) indicating bold, or url, or anything else. I wanted to do it that way since years ago when we had the '[0]' replacements (I thought it should just be a span in a tag with no placeholder in the content)
2) If we think a chain of events is needed to prove that none of your events were censored (so we can properly claim censorship resistance), we could add another kind where authors publish their chain of events as a list of IDs. If it needs some kind of mergeability to fix race conditions, I've talked about how to do that with a last-write-wins (LWW) conflict free replicated data type (CRDT) in https://github.com/nostr-protocol/nips/pull/1630.
nostr:naddr1qqgxymr4v4ekk7fdweej6mn0wd68yqgwwaehxw309ahx7uewd3hkctcpz4mhxue69uhhyetvv9ujuerpd46hxtnfduhsxpqqqp65wq3qq3sle0kvfsehgsuexttt3ugjd8xdklxfwwkh559wxckmzddywnws2fkr4c
Discussion
Some changes are easier than others
GPG can be used for e2e encrypted email but almost nobody (including me) bothers
thank you for your fair take of pointing out that the world has constraints and not all things can be done at once, it was very insightful and smart