You can't enforce that, but you can enforce that they open their own protocol, and also that the API is open to be used by anyone else, i.e. anyone can develop an own client.
How does that work with top-down moderation? E.g. some user X is harassing other users, so the company running the platform wants to ban them. If the protocol is open, isn't banning basically impossible, since they could just change to a client not controlled by said company?
edit: hmm, I guess this depends on what it means for the protocol to be open vs the platform itself...
