Can ordinary users blacklist sites/clients at the protocol level...? To stop their messages appearing on porn or hacker or hate or corporate sites that try to drag everything in..? But having to do that for every site would get boring very quickly, so what about some kind of generic protocol level "no porn/crime/spam/bad thing" opt-in..?
Discussion
There is a NIP-51 in progress that is about mute lists, but are clients that have to implement it.
At protocol / relays level the content is free to move.
Probably there will be (paid) relays with content filtering features.