You're just describing multiplayer video games. We've already had those for more than 20 years.

Well, yeah. Honestly, a massively multiplayer GTA V or Red Dead 2 would basically be the whole Metaverse concept. The only differences would be: (1) You aren't there to complete missions; there's no storylines and it's totally open-ended. There are no writers, and (2) The spaces and content are created by anyone, not a central company, and (3) You can go from GTA world to Red Dead world to tons of other worlds without changing missions or skins or avatars or whatever; they're part of the same huge space you can traverse without leaving VR. (4) You can code your own physical spaces and code your own avatar's abilities.

So, like, yes Stephenson's metaverse is similar to a lot of video games we have now, but it's critically different because his characters are "heroes" in that universe specifically because they coded their own avatars and their own environments.

Which is another way of saying that it rewards thought, code, and creativity -- exactly the things that video games steal people away from.

We already have games almost entirely like this, and the market is clear: no one plays them, to a good approximation. People play games to compete or cooperate on specific goals. This idea of sandbox multi-player interaction is only a gimmick that's interesting a few weeks tops, then the vast majority leave to play Candy Crush or Fortnite where they have clear goals.

If they want to meet with friends and hang around, real life contact just can't be beat. That's the reason why virtual persistent worlds are ultimately bullshit (as opposed to game worlds where you have an actual game to play, be that chess or matching colored gems or killing other players).

I don't know, that just sounds like pitching a TV show without writers. Even a reality TV show or a cooking show needs writers to keep things entertaining. Otherwise, you sort of just have nothing to look at.

> You can code your own physical spaces and code your own avatar's abilities

I ask again: why can't I code my avatar to be the most OP avatar that ever exists? I will create a weapon that kills people if they look at me. What's the incentive to purchase new things if everyone can just code the most OP version and undercut the person next door? I will have the best cars, the funniest T-shirts, and the strongest potions, and you cannot stop me. Because I write Lua scripts.

There's some element of cooperation that is usually assumed when people talk about the metaverse but I don't see any incentive to maintain it that way. Garry's Mod and Roblox are platforms that do what you're talking about, today. The way they maintain this cooperation is that they prevent users from having strong autonomy in games that others make. In Roblox, your avatar is limited to a few cosmetic choices for customization, and in Garry's Mod, you don't really have any customization at all in most game modes.

Real world is like that, you can “code” manipulate the world. However there is a limit on some extent and in such worlds there will be other limits like a consensual physics rule on servers entry, like “local laws”. Rather than LUA think on declarative contracts that limit your crafting based on agreed parameters.

> The only differences would be: (1) You aren't there to complete missions; there's no storylines and it's totally open-ended. There are no writers, and (2) The spaces and content are created by anyone, not a central company, and (3) You can go from GTA world to Red Dead world to tons of other worlds without changing missions or skins or avatars or whatever; they're part of the same huge space you can traverse without leaving VR. (4) You can code your own physical spaces and code your own avatar's abilities.

So Second Life ...

http://wiki.secondlife.com/wiki/Content_Creation

>how would that not be a great way to sell a car?

Because there's no guarantee that the experience is in any way comparable to the actual driving. It's in the manufacturer's interest to do the opposite. If the controls weren't standardized the simulation might have still been useful to learn the controls, but they are.

Besides, when AI cars come along the criterias will change completely. From driving experience to cost and comfort. That could be virtualized... but there wouldn't be much point in it.

Following up on the racing of infinite-speed cars, there was also an emphasis of optimizing your own user interface. That when every car could go at whatever speeds you want, the most important thing is how to interact with the car, and to control it. So it isn't just somebody else's code, but the skill and challenge of making your own code to suit your particular needs in an interface.