Nostr is an amazingly versatile protocol, and has huge potential to solve a number of real world problems. But it isn't a silver bullet, and I think it'll be interesting to watch as the Nostr ecosystem develops and discovers problems that Nostr can't solve, and which complementary technologies fill those gaps.
For the record, I have no problem with people building crazy things on Nostr - hackers are going to hack, and it's more important to stretch the protocol than to nitpick. But I'm going to nitpick anyway.
You may already be familiar with the CAP theorem. If not, it's the idea that any distributed data store can only provide two of the three following guarantees:
The Nostr protocol describes a distributed system that stores data, and this distributed database is no exception to these rules. I'll give you three guesses which guarantee Nostr doesn't provide, and the first two don't count.
In case you need a hint though, it's consistency. Nostr optimizes for censorship resistance (availability) and redundancy (partition tolerance) at the expense of consistency.