The Negative Appeal of the Metaverse

An era of brute-forcing innovation (or why AI art is different than Web3)

Legs won't save us (Photo via Meta)

Perhaps, one day, we’ll all be trapped like legless rats in the “infinite office” of Meta’s Horizon Worlds and toiling in a parade of conference calls under stultifying virtual halogen lights. But for now, while I can still drag my flesh prison through the analog hallways of our meatspace, I am basking in the schadenfreude of Facebook’s metaverse stumbles: Reports in The Verge, The New York Times, and The Wall Street Journal show that Meta is having difficulty attracting and retaining Horizon Worlds users, even from inside the company.

According to a leaked memo sent to Meta staffers last month, many Facebook employees “don’t spend that much time in Horizon and our dogfooding dashboards show this pretty clearly.” The memo asks why this is the case, then pledges to “hold managers accountable” and make sure they’re having their teams use Horizon at a minimum of once a week.

The memo goes on to argue, essentially, that beatings will continue until morale improves:

Everyone in this organization should make it their mission to fall in love with Horizon Worlds. You can’t do that without using it. Get in there. Organize times to do it with your colleagues or friends, in both internal builds but also the public build so you can interact with our community.

The subtext of all these stories is obvious: It is generally a bad sign for a piece of technology if employees of the company that makes it won’t use it. This is actually the opposite of the problem that Facebook had throughout the mid-2010s. When I interviewed employees of the social network at that time, many would talk about the old-school Facebook product and its spinoff tools as being truly useful. Staffers sometimes expressed genuine enthusiasm about quirky new services, like Facebook “Rooms,” that would quickly wither on the vine. This is because Facebookers were using these services to get work done internally in ways that didn’t quite mimic how regular users interacted with the platform. While the rest of us were observing our relatives in our feeds get increasingly hostile with strangers about politics, Facebook employees had a skewed, rose-colored perspective informed by their own day-to-day use of the platform’s tools.

But what’s even more grim than today’s employees failing to use the tools is Meta putting its head down and making Horizon use compulsory inside the company. It’s an effort to recreate the mid-2010s dynamic and trap employees in its own garden by making the walls so high they can’t see everyone on the other side snickering at them. And, hey, it might actually work. It’s at least worth a try.

About those people on the other side of the wall laughing: Meta has—naturally—adopted a scolding posture when it comes to criticism of their investment of more than $10 billion. “Being a cynic about new and innovative technology is easy,” Facebook spokesperson Andy Stone told the Times. “Actually building it is a lot harder—but that’s what we’re doing because we believe the metaverse is the future of computing.” It’s an extension of the tech culture war between maintainers and builders that I’ve been harping on about for the last year. It’s easy to criticize; it takes bravery to put yourself out there and try, risking failure.

I personally don’t think that Meta’s metaverse expansion will be much of a consumer success, for a few reasons I’ll get into in a bit. But my reasoning for criticizing the company’s Horizon initiative isn’t very complex—it’s not even rooted in concerns about Meta building the future of computing (which, for the record, I’d rather not have happen). I think Horizon’s ambitions, as they stand now, will likely fail because the platform violates a primary, if obvious law of new technology: It usually has to have near-instant utility and instant appeal.

My favorite tired, but still helpful, example of instant utility and appeal is the iPod. Your portable CD player can hold like 12 songs, and if you want more, you need to lug around a cumbersome leather carrying case that you will inexplicably feel compelled to reorganize every two months. But the iPod, a little white brick, can hold 20,000 songs. The value proposition is instant. You didn’t have to invite somebody to try out an iPod for them to realize they wanted it. And, if they did want to try it first, they understood what they were getting right away.

We’ve had failed technologies as long as we’ve had technologies, but right now we seem to be in a golden age of brute-forcing innovation. Web3, countless blockchain products, and the metaverse all have the near opposite of instant appeal and utility.

Nearly every Web3 project I’ve encountered is wildly difficult to access, and its usefulness is suspect—often, it seems that these technologies aren’t as decentralized as they claim. And, rather than solve a problem, they either replicate the problems of the old technologies, only with different intermediaries, or they add new problems (fraud, complexity, irreversibility). When critics comment on this inaccessibility or lack of immediate appeal to regular people without .eth in their Twitter display name, a standard response is that much of the blockchain ecosystem is in its early days, and rendering judgment on them is foolish and hasty.

It’s been helpful for me to compare these decentralized projects and metaverse ambitions to another piece of new technology: AI-powered text generators and image generators like GPT-3, DALL-E 2, and others. One can make a convincing argument that these types of AI tools, like diffusion engines, are in their very early days. Diffusion technology is still quite new, and the engines that power AI art less than two years old (the Bitcoin whitepaper, by contrast, was published in 2008). And yet, one doesn’t need help in understanding the power of these tools. This summer, I showed an AI art generator to a 5-year-old and an 8-year-old and asked them to come up with weird prompts for it. Their roving imaginations cooked up ideas like “a flea surfing on a tsunami wave with a pineapple,” and the machine spit out renderings in seconds. They were not confused; they were entranced and deeply entertained.

The power and utility and appeal of these tools is so strong that many commercial artists are very worried about what it could mean for their jobs. There’s all kinds of fear and hype (much of it either overblown or not grounded in reality) around AI tools, because they seem so useful and efficient. This perception is part of why people seem so eager to assign an almost-spiritual quality to AI. Of course people will use this technology, the logic goes. Why wouldn’t they? It’s simple and pretty good at what it’s supposed to do. This doesn’t mean the technology is de facto good, but it most definitely works.

A lot of crypto stuff is billed by true believers as extremely internally complex, but with an immediate utility: a trustless reformatting and organizing of many of the world’s systems, thanks to the distributed ledger. But those words, however concrete-sounding, are actually extremely vague! Meanwhile, AI actually is quite internally complex (which isn’t necessarily a good thing!), with people building systems that process and synthesize information in ways even some of the builders don’t really, fully understand. But the utility is immediate and simple: These tools make connections among objects to replicate human processes like image and text creation. It’s not vague! The networks make these decisions and you see the result, even if you cannot see exactly how it reasoned itself to the outcome.

I’m almost certain that my rule of immediate utility breaks down some upon closer inspection. I’m sure there are a lot of examples of things that were built and went looking for a use case, only to find them later. And I’m sure there are way more examples of somebody creating a technology off the backs of an earlier, less immediately useful or appealing technology that then becomes extremely useful (this is often just how innovation happens). I realize that the AI art engines are the product of a long period of trial-and-error innovation in this space.

But many of the brute-forcers aren’t really arguing that their technology is but a stepping stone for something great to come. They are not behaving like DARPA or some research lab—they’re behaving like venture-backed, consumer-facing companies that are looking for returns on the products they’re building right now. Meta isn’t asking you to endure a bunch of garbage for a decade until this stuff gets good. When Meta says that “legs are coming soon”—the emphasis is on soon.

But I’m not sure if any amount of lower-appendaging can save Meta from the current purgatory of Horizon. Via The Wall Street Journal:

Horizon is designed to be a sprawling collection of interactive virtual spaces, or worlds, in which users appearing as avatars can shop, party and work. Yet there are rarely any girls in the Hot Girl Summer Rooftop Pool Party, and in Murder Village there is often no one to kill … According to internal statistics, only 9% of worlds built by creators are ever visited by at least 50 people. Most are never visited at all.

If a digital avatar is murdered in the Murder Village but nobody sees it, were they even murdered at all?

If you talk to people like Wagner James Au, who has been chronicling Second Life and adventures in the metaverse since 2003, you get the sense that Meta seems intent on replicating many of the mistakes that virtual-reality developers made in the aughts—mostly by designing tools without much consideration of how real people will use the technology. Sure, you can build the Walmart Brand Activation experience, but what the people want is this:

But, as Au points out, Meta isn’t optimizing its product for normal people. Au’s reporting suggests that people aren’t buying many games from the company for its Quest VR headset and, instead, are spending more time in free, third-party apps like VRChat. This is a problem for Meta, and so it seems the company is raising the price of its new Quest headset to try to make up some of the revenue difference. It seems very plausible that Meta will continue to build bland consumer products and focus more on building hardware for other companies that have real adoption.

Meta’s other focus, as New York’s John Herrman and the writer Rob Horning both point out, is capturing the interest of clueless C-suiters who are both confused and anxious about missing out on the metaverse. If you can’t build a product that humans want to use for pleasure, you can build some bossware that traps them in an “immersive meeting experience.” As Horning notes in his excellent newsletter:

Any efforts to promote the consumer-facing metaverse, I think, should be understood as a way to try to distract us from noticing that trajectory: that it is a means of trapping and containing labor.

Horning provides a helpful frame that applies to other technologies that are being brute forced into our news cycles and culture. In the case of many crypto projects, the hype and consumer-facing nature is a way of distracting from Ponzi dynamics or the notion that many of these investments are a way for venture capitalists to park money from LPs that they don’t know what to do with while collecting their fees. To borrow his phrasing, it is a means of trapping and containing capital. There doesn’t need to be any immediate (or even eventual) utility if there’s money to be made by creating a fear of missing out.

The brute-forcers embody this cynical vision of innovation by trying to foist adoption onto users, and, when it doesn’t work, they switch to more coercive tactics. In this light, Meta’s mandatory weekly metaverse time is an early template that we could see in workplaces in the future.

Last week on Twitter I read a good back-and-forth about the “impact” of the techlash and the argument that, despite lots of negative media coverage, there hasn’t been that much real change in U.S. tech policy. It’s an interesting argument that got me thinking about Facebook’s metaverse pivot and rebranding, which one could argue is directly related to being the central figure of the techlash.

While it’s true that U.S. tech policy hasn’t kept pace with the public distaste toward Big Tech, I’d argue that Meta’s struggles and low adoption toward its consumer-facing products are probably a direct result of the techlash. The company’s inability to produce any successful new product in recent years is a result of a brand that’s tainted and synonymous with political discord, decaying social graphs, and your sweet little old aunt posting aggressive anti-vax Minions memes. From a cultural perspective, the techlash has eroded trust in the company, but more than anything else, it has helped accelerate Facebook/Meta’s deep uncoolness.

This, I’d argue, is generally a good thing. But the specter of Meta’s enterprise metaverse ambitions looms large. It’s entirely possible that one byproduct of the techlash will include making bad tech unpalatable to regular people, instead driving it into the arms of big companies that use it to grind down the bones of their employees. What sucks about the influence of cynical brute-forcers in our current tech landscape is that instant utility and appeal aren’t necessary to capture broad attention and achieve wide adoption. One month you’re cracking jokes about “legs coming soon” and the Horizon’s Murder Village, and the next month your bipedal avatar is shuffling over there for an all-hands meeting.

Charlie Warzel is a staff writer at The Atlantic and the author of its newsletter Galaxy Brain, about technology, media, and big ideas. He can be reached via email.