The low execution quality of Meta's metaverse effort surprised me, too.
But they wanted it to run on their relatively weak headgear. A good metaverse needs a decent gamer PC, a serious GPU, and a few hundred megabits per second of Internet bandwidth. (I've written a Second Life client in Rust, so I'm very aware of the system requirements.) Facebook needs to serve a user base which is mostly phones and people with weak PCs. Not Steam users.
If you have to squeeze it onto underpowered hardware, you get something like Decentraland or R2 or Horizon - low rez, very limited detail, small contained areas. Roblox has made some progress on this problem, but it took them two decades, even with a lot of money.
The real problem with metaverses is that a big, realistic virtual world is a technical achievement, but not particularly fun. It's a world in which you can spend time and meet people, but the world is not a game. It has no plot or agenda. This throws many new Second Life users. They find themselves in a virtual world the size of Los Angeles, with thousands of options, and are totally lost. It's not passive entertainment. As Ted Turner (CNN, TBS, etc.) used to say, "the great thing about television is that it's so passive."
I think the problem goes beyond that. Meta never had a particularly coherent story for what "Horizon Worlds" was supposed to be to users - it was variously pitched as an online conference room, a social hangout, a way to explore 3D models, a video game... it felt as if they were throwing ideas at the wall to see what stuck, and nothing really did.