A lot of the stuff on there seems way beyond what can be achieved in something incrementally added to the protocol. Fair enough in an article about "v4" of the protocol. But I don't think we are - nor should we be - talking about a full new protocol version here. Instead we are talking about extending the protocol, where the extensions are opt-in.
You are correct we are not talking about a whole new protocol, but why not ?
Seems to me we would have a lot more latitude to get it right if we didn't have this limitation.
