Throughput and consistency, especially at scale. As well there are limitations with relays that will absolutely be a detriment at scale without some recourse.

I can pick beef with different NIPs of course, but the idea of relays that only allow read/write access needs some framework above it for managing events more consistently.

We don't even have that. We have relaytools which is the only architecture I really see stepping up to the plate. It did its best but it cannot fix lower-level problems because it is designing up from where we are.

So I follow mleku because I always wondered how the protocol would operate if relays worked over http instead. I have high hopes that this will allow me to continue developing in this space. If my projects seem unlikely to succeed under that paradigm, then I will keep building in other directions.

The idea of Nostr is great. The result is kinda good. But it has to scale to meet me demands, and what I see doesn't scale.