1. Social vs goal driven is a false dichotomy: multiplayer games are always both. In any player population you're going to get a mix of players driven by social and creative elements, driven by competition, driven by exploration, etc (cf. Bartle types or Yee player motivation taxonomy).
You can try to encourage some play types over others, but it's really hard to do a game that's social first, just because social by itself doesn't drive retention and acquisition enough: you need something else for players to do to have fun while they're also having fun socializing.
History of gamedev is full of social-first environments, from legions of old MUDs, to Second Life or There, to IMVU, etc. They never reached the same heights as MMO games because it's hard to make a game "sticky" without having the game there.
2. Building a platform first, and then building a game on it, is a quick way to lose all your money. Building a game first and then extending it to a platform is the correct direction, because if you don't have a game, you won't have users, see what works, etc. And until you have users, you don't know what players want from a platform.
3. "Still, it’s clear that we want to come together, not just in Instagram DMs and email threads, but as avatars navigating shared spaces. Somehow." That's just plainly false. Players are completely happy with social interactions without avatars, without 3d space, etc. There are many ways to make games more social, as gamedevs know well.