Meta's virtual world projects always give me the sense that they're aimed at set of personas/demographics that has almost zero overlap with the people actually running or building those projects.

That's why they fail. Virtual worlds - almost like real ones - are super-platforms, not applications. The interactions are too varied and complex to reduce to a set of simple primary use cases or personas.

So, if you're building a virtual world that you yourself wouldn't want to use, you're almost certainly going to fail.