Sleeping on MS-DOS
Our look towards VR's future is fixated on physical specs.
What will Apple's new headset look like? When will eye tracking, varifocal lenses, and XR-specific chips converge to create a pair of magic glasses? 2022? The last quarter of 2029? We are biased towards innovations in hardware.
But if we look at the history of the personal computer, hardware almost didn't matter to the market. It was all about an innovation in the operating system. The graphic user interface was the OS phase-change that put Windows 95 under the Christmas tree.
A company called RP1 has emerged in the last month, focusing on a standalone operating system that can be deployed across multiple VR devices. The name draws a parallel to the movie "Ready Player One," a book that shaped our culture's vision on what it could be like to live in a machine.
RP1's technology could enable something like it: A seamless network of worlds that can hold up to 200 million concurrent users. But how can a start-up launch a VR OS without their own hardware?
This is exactly how Microsoft started.
Microsoft built a prototype, sold it to AT&T in 1980, and then attracted the attention of IBM. IBM gave them the resources to build out MS-DOS, but their hardware bias prevented them grabbing any stake in the thing that would evolve and take over the world.
Embedded in RP1's "shardless architecture" could be an innovation on par with the GUI (graphic user interface). It could fundamentally change, not only how VR software and hardware gets developed, but the shape of the Metaverse itself.
By looking at the evolution of the PC operating system, we might understand why a standalone VR-OS startup is in a position to beat out some of the tech-giants pouring billions into this space.
Sleeping on MS-DOS
On a 1974 fall day in Harvard Square, a wide-eyed Paul Allen found a magazine showing the Altair 8880.
You can fit in your house!
You can leave it on a table!
This is an Altair 8800. It was the world's first micro-computer. At the time, no one conceived a computer could be this small. While other companies would iterate on form for over a decade, Allen and Gates saw a different opportunity:
To build a software architecture for a new paradigm.
The name Microsoft came from the combination of two words that were relatively unknown at the time: "microcomputer" and "software." They registered the company in New Mexico in 1976.
Their first operating system was called Xenix, and it was acquired by AT&T in August of 1980. It was a Unix-based system, and it was deployed to several pieces of hardware. IBM took notice, and after a failed deal with Digital Research, hired Microsoft to build an OS for them in 1981. Allen and Gates called it MS-DOS, and their deal with IBM went through in July of 81. But there was one critical clause that was overlooked:
Microsoft retained the right to sell their OS to other hardware companies.
IBM was sleeping on MS-DOS. They thought their own hardware was their competitive advantage. To them, MS-DOS was just a feature they could tack onto an IBM machine. They assumed their cutting edge table-top plastic number machine would keep them afloat forever. They were quickly cloned.
Microsoft was founded on the vision that by owning the OS, the hardware would become irrelevant. Within 4 years, MS-DOS evolved into Windows. In 1985, Windows was just a visual extension built over MS-DOS. But over the next 8 years, the graphical user interface was re-built from scratch. Working from tabula rasa let Windows evolve and become the default OS on almost every personal computer by 1993.
There's a lesson to be learned in IBM's perspective on hardware vs. software.
IBM's rise in the 1980s was because of their innovation in hardware. They shot from 5% to 60% of the personal computer market share in a 5 year period. Their success framed their world view on what was important.
But IBM's radical loss in market-share was due to their hardware bias. Their 75% share of the entire technology sector fell to less than 10% by 1995.
Bias in Developing
Almost the identical thing is happening in the VR hardware space.
Companies are racing to research, iterate on, and develop headsets, while in some cases, completely neglecting the OS. Magic Leap, an AR headset, raised billions of dollars to launch a high-spec device that runs on an Android Watch OS.
Android Watch OS!?
The hardware challenges are so novel and distracting, that OS-design is reduced to choosing from off the shelf parts. Hardware challenges are bleeding edge, involving physics and computer vision. But it's not an excuse to neglect design and UX.
Don't get me wrong, hardware is critical, especially in a medium where you put the thing on your face. My argument is that VR headsets are already near their "final form," while our software is still in it's own MS-DOS like era.
1993 DECpcAXP 150
Over 10 years, the shape of computers from the perspective of the consumer didn't change that much. Sure, it would be a crime to overlook all the hardware-level advances (the floppy disk was a break-through). But the real revolution that changed consumer attitudes was in the OS. It was in the ability to use a mouse, and watch images form and melt based on your commands.
These images below show a series of headset prototypes from Facebook Reality Labs. It gives us a glimpse in how headset form might evolve over the next decade. Headsets of the future will be lighter, sharper, more powerful, and better calibrated to our bodies and surroundings.
Similar to the image of the two PC's above, we're looking at 10 years of hardware refinement. Improvements will be felt, but we won't see a phase change (from PC to smartphone) in the next decade. Subtle upgrades in PC hardware didn't bring the masses, and neither will cooler VR headsets.
The headsets we already have today are ripe for someone to build a VR-specific OS from scratch that will open up an entirely new vista of functionality. Innovation in the OS is what will enable both spatial computing and a society-scale Metaverse.
This time around, Facebook is the IBM of our immersive future. The parallels between the two are many. Facebook gained their scale from a social network, while IBM gained their scale from manufacturing. Neither were electronic hardware companies, but each is expending an insane amount of energy to conquer the "next" phase of computing. Is Facebook the clumsy Goliath doomed to fail, or is there a critical nuance?
Facebook is well aware of IBM's blunder.
Facebook is throwing billions into VR, but it's not all focused on hardware and consumer releases.
They're building a huge research division, and have over 800 VR-related job openings right now. In addition to aggressive product launches, their research team is consistently breaking grounds in human-computer interface design. They're the first to build hand-tracking into a consumer device, and an upcoming launch integrating Logitech keyboards will create an "input-agnostic" OS. That's a big step, and will potentially solve lots of the UI confusion we see in the space today.
Then there is "Horizon," their social VR app, which is actually the prototype of their future OS in disguise. Horizon has the potential to fuse their app store, app library, social graph, avatar system, and even the content itself, into a single three-dimensional fabric. When you put on the Quest 3, I wouldn't be surprised if you boot directly into Horizon.
Similar to how Microsoft dabbled with software products (word), Facebook is making a play to own the entire content ecosystem. They're acquiring any VR studio that proves they can make a hit (ie: Beat Saber). Facebook is even developing in-app creation tools, so that you can easily make your own applications from within their OS (without needing a game engine).
Facebook has their eyes set on a vertically-integrated solution. Meaning, they want to own every conceivable dimension of the ecosystem. Their imperative is to shepherd the entire Facebook social network through a phase change, and get there before anyone else is even close.
Immersive medium add a new dimension of data capture for advertisers to drool over: your eyes!
Some are calling Facebook's vision a "monoverse." It's a monopolized metaverse. It looks a bit like the IOI, the dystopian corporation from Ready Player One.
Despite their vision and resources, they are still extremely vulnerable to their own biases. No matter how much money and talent they throw at building their own weird future, perspective mattes. An incorrect world view could render their aggressive charge obsolete.
Android & the smartphone paradigm
The Oculus Quest is built on Android. The back-end architecture is similar to Fruit Ninja, except now you can use your real-life arms to flail virtual knifes around. It can take minutes to download and move between apps. It's a real thumb-twiddler. We live in an on-demand culture, but the UX is on par with AOL and the dial-up age.
The assumption is that XR-chips will eventually get stronger over time, enabling faster load times and parallel-processing. Software developers are more focused on "What makes a VR application good?" instead of, "How can we link all of these VR applications together?"
Concurrency & server architecture
If you go to a concert in Facebook Venues with 8,000 people, your experience will be a private instance with 8 strangers. I think the best we've seen in VR is ~100 people in one instance. The Metaverse is here when one million people can gather in a virtual stadium to watch a real football game. This is called the "concurrency" problem, which Matthew Ball talks about in a few of his essays.
It's not a pressing problem to solve right now, since the market of people looking for multi-player VR experience is still relatively small.
The Facebook social layer
The Quest 2 requires a Facebook log-in to use the device. They're getting sued over it. They've recently replaced their chat feature with Facebook Messenger. While they have honest engineers working for them, so many of their moves will revolve around fusing VR with their Facebook ecosystem. People are forced into it now because there is no real competition in the standalone headset space. This could completely change in the next few years.
Match-making and connections currently happen at the app-level anyway. There isn't the need yet to create an OS-level social graph that connects people across all devices.
While Facebook is building a comprehensive walled garden, it's very possible that external parties will come together in the spirit of collaboration (or even, revolt). Companies can focus on one specific horizontal element (the OS, the app store, the social graph), make it the best that they can, and make it so that it works across a huge range of devices.
Rapid app switching, infinite capacity, and open social graphs, will completely change the nature of what time spent in VR can be. This is Facebook's potential blind spot.
Facebook's commitment to VR gives them a lead now, but the market is still so young. In an emerging market with so many unknowns, your perspective matters more than a 12-month head start. What the VR OS layer needs to be is still an unanswered question. It is both the most important AND the most under-prioritized piece of the puzzle.
The VR OS is the attack vector for surprise disruption.
RP1 and the "Shardless" Metaverse
It helps to think of the current network of VR applications as a web of disconnected "shards." Each application is a self-contained .exe that you cache onto your device. It's slow to download these apps, and slow to load and move between them. Even though we're seeing multi-player social networks arising, each one is a self-contained universe. Each shard also has a max capacity, making it hard to pull off festivals that require 1,000s of people to come together.
RP-1 is developing a "shardless" operating system.
Their prototype is going live in 2021 Q2, where they will demonstrate how they can run a world with 200 million people, in a single instance, over only 33 servers.
It's like where going straight from dial-up to FIOS.
From 4-player Mario Kart to the Daytona 500 million.
I heard their founder speaking in Clubhouse, and there is even the possibility of apps having near-zero load times, meaning you'll be able to seamlessly move between scattered worlds.
My guess is that the goal of their upcoming prototype is to build partnerships with VR app developers and hardware manufacturers. Partners would have to rework bits of their codebase to reap the benefits from RP1. But if they do, we could see RP1 emerge as an independent OS used across multiple headsets.
It's likely that manufacturers who are less-invested in building a full VR ecosystem would embrace RP1 (HP, Lenovo, Dell, Acer, ASUS, Samsung). What's interesting is that all of these headsets have already adopted a shared OS called "Windows Mixed Reality OS." It's one of the first standalone operating systems, but it's lacking in several dimensions. Given these companies aren't investing in OS development, they are the ones that might embrace cutting-edge software elements from start-ups.
On the other hand, companies like Facebook, Apple, and Vive are dedicating serious effort to building their own OS. When it comes to interoperability, they might not be interested. When it comes to concurrency, they will be attempting to clone.
Concurrency and interoperability are just two of the many design problems that a mature VR OS will need. But I think RP1's potential success would align with Microsoft's MS-DOS story to prove that a pattern exists: young companies can out-innovate technology giants when the latter's blind spots are fueled by their own ambition.