The situation with online persistent 3D/VR worlds ("metaverses") is basically that we currently have AOL, Compuserve, Prodigy and GEnie and we are waiting for someone to invent the web browser.
(For "Compuserve" and "AOL" substitute "VRChat" and "Horizon Worlds", and maybe you'll see what I mean.)
This isn't how "the Metaverse" was supposed to work. The book the word "Metaverse" comes from was written in 1991, when "online" actually did mean AOL and Compuserve but there was not yet such a thing as the web (WorldWideWeb existed, but not NCSA Mosaic). The book "Snow Crash" is trying to imagine something larger and freer than that. Snow Crash is not really about VR. It's predicting the Web. Snow Crash imagines online spaces that you can move between fluidly, where anyone can create whatever system or service they want without partnering with (or being) the firm that owns the dialup interface. That is what the Web delivered. When Hiro walks into the Black Sun the exciting thing, the thing that makes the book's world work, isn't that Hiro is literally embodied while using a computer program. The exciting thing is that the Black Sun is run by, and its internal rules are defined by, a different organization than the one that is responsible for the street outside. That's a hyperlink¹.
The Web of course since moved backward in this way. Many people spend all their online time in single web apps, or in phone apps those web apps have spun off. Facebook once founded a thing called "Internet.org" which would give away free cellphone internet which was gated to only run Facebook². They were trying to get us to a point "The Internet" meant "Facebook" and that felt natural. They almost got away with this. Then last year Facebook launched their 3D chat app (Horizon Worlds) and called it "The Metaverse".
Tech business reporters seem to be embracing this framing and talk about apps like Horizon Worlds or Roblox as "Metaverses". Plural. Snow Crash had "The metaverse", singular. "Metaverses" or "A metaverse" would be a contradiction, interconnectivity was the point, it would be like talking about "Internets". But these days we're used to everything being siloed, Facebook might as well have its own Internet and a siloed "metaverse" seems not only natural but inevitable. After all, it's how the web works.
I think the thing that broke our thinking here, the thing that made siloing seem inevitable³, is the switch from sites to applications. The early internet had some CGI but it was made of static files, not programs. Sites like Facebook, Instagram, Twitter, Cohost etc are computer programs with a web frontend. Hyperlinks still work for the basic act of switching your display from one site to another, but the interesting thing that has people using social apps isn't just the base "display content" verb, it's richer verbs like sharing or rich features like metric tracking and persistent identity. If you use Cohost you've probably also at some point evaluated Mastodon ("The Fediverse") so you've got some sense that it's possible to make a social app which is distributed in such a way those rich social verbs work across sites. But if this is you then you've also experienced (even if you like the Fediverse, which I do) that making those verbs distributed comes with usability tradeoffs, costs, friction. Applications have a gravity that leads toward siloing.
I think the reason we struggle to make a Metaverse that is open⁴ in the sense of Stephenson's imagined "Metaverse"⁵ is that a 3D persistent world is more like an app than it is like a website. Part of what made the web work (to the extent there was an era it worked) was that the content was flat and relatively simple. You didn't need app features. It's hard to imagine 3D shared worlds (to the extent I can imagine anyone wanting such a product) that work at all without app features (persistent verifiable identities, some degree of moderation). If we want open, non-siloed online VR spaces to exist, I think either we have to find some way to make it less like an "application", or we have to get good at writing applications that are distributed.
¹ My wife insists that even a reading of Snow Crash like the one I'm doing here fails, because in her read Snow Crash is supposed to be a dystopian novel, all of it, even the Metaverse. In this read, any attempt to cast the Metaverse as aspirational is a mistake. I'm not sure if I agree with this or not— is Stephenson's Metaverse supposed to be dystopian, or is it supposed to be the one bright spot in a dystopian future?— but it doesn't really matter. Even taking the pessimistic read just tells us that the dystopia we built is not even as good as the dystopia Stephenson imagined.
² And like one or two other figleaf sites, Wikipedia or something, so they could sell this plan to the Indian government with a smidge of plausible deniability. Nobody fell for it.
³ Lots of people point to the death of RSS as the thing that brought about the web's siloed age, and in this interpretation the problem is not that distributed apps are hard to make (as after all RSS, and descendants like ActivityPub/Fediverse/"Mastodon", are steps to making groups of disconnected websites act like apps) but rather capitalism and the instinctive need corporations like Google or Facebook have to capture users and deny oxygen to potential competitors. This is a solid interpretation of events but destroying capitalism is outside the scope of this post.
⁴ Mozilla Hubs does exist, but I am unaware of anyone using it and at its current level of development I don't think this is the 3D-persistent-world "Web". I think if we are using the Compuserve-AOL-Prodigy analogy then Mozilla Hubs are BBSes, and if we're using the Facebook-Twitter-TikTok analogy, then Mozilla Hubs is Cohost.⁶
⁵ Stephenson himself is working with a company called "Lamina1" that recently put out a whitepaper I have not read on what they believe an "open Metaverse" would look like, but Lamina1 describes themselves as using "blockchain", so I assume the people who wrote the whitepaper do not understand distributed applications and the technology does not work.
⁶ Update 2024-April-03: Mozilla Hubs has been discontinued.
