There is a protocol for syncing distributed databases called negentropy that does something similar to what you're describing. And although your solution works on paper, there are trade offs. 1. Relays become bloated as every relay needs a copy of every event making it harder for anyone to run a relay. 2. This would certainly make queries for events slower as a result (since now there are so many events to go through to find the right ones). 3. Relays can no longer serve small isolated communities that know nothing about the rest of the nostr network and instead serve one giant public square removing some flexibility in the network.