Isn't content persistence a relay issue rather than a protocol issue? Someone could run a relay that promises to preserve all your content. Doing so is a responsibility that not every relay wants to take, especially given that people are running relays for free.