Thank you for your response, I appreciate your insights!
I think if there were really serious problems with a future version of ActivityPub, we could feasibly do one of two things:
Maintain a fork of the protocol - this has actually already happened once, with an implementation standard called Litepub.
Move over to a different protocol, such as Zot.
The second route is probably much harder, but there’s no real reason why a zombified Meta version of the protocol would do much of anything to anybody running vanilla ActivityPub at this time. You’d probably have some feature incompatibilities and breakage, but…if you’re not going to federate with them anyway, what can they actually do?