What's wrong with nostr:

1) Reliance on a single key for all signing, without any ability to delegate or rotate keys

2) JSON serialization format (already resulting in bugs between different implementations due to unicode differences) should have been binary

3) No real anti-spam mechanisms

4) Economic model for paid relays doesn't make much sense – one time payment for on-going service isn't feasible long term

5) Economic model for free relays is awful: you want to determine the value of an individual event, eg with a proven ₿ sacrifice, to make spam expensive. But Lightning alone can't do that because:

6) Related to #5: Zaps can easily be faked because you can't prove a LN payment represented real funds.

7) No clear scalability solution: if nostr actually scaled, it'd support images...

I could go on...