Yep, I coded a similar solution before I tried NIP-94. It's not as robust, the performance improvements are minimal to non-existent, and it mainly doesn't allow the creation of all the new Nostr micro apps NIP-94 enables.

The only feature it has going for it is backward compatibility at the cost of being stuck in time.

This situation is similar to DMs. The current implementation is safe, but not the safest. We could make a better one, but it's going to break everything because it will have to use a new encryption protocol. So no one is even trying. We have been stuck with it for years. And the more we wait, the more clients implement it and the harder it gets to migrate to a better one because of "backward compatibility"