One day around Thanksgiving 2011, Philip Rosedale had a vision of a virtual world. He was sitting in the San Francisco offices of Coffee and Power, the job-listings startup he'd co-founded in 2010, playing with a circuit board attached to a gyroscope. Such sensors are an integral part of virtual reality (VR) headsets: by following head movements, they help give a wearer the impression of being inside a computer-generated environment.
Rosedale, 48, has been dreaming of VR his entire life. He knew that, to be effective, gyroscopes must have extremely low latency for the lag between movement and response to feel natural.
Holding the tiny circuit board, Rosedale tilted it back and forth, watching the output on his MacBook screen. The latency was almost non-existent. When VR had last attracted attention, back in the late 80s, the headsets contained trackers, but they were jittery and expensive. This modern, $15 sensor traced Rosedale's movement faster and more accurately than the naked eye.
All at once, Rosedale's VR dream came flooding back: a virtual world made up of millions of connected spaces, like a three-dimensional version of the internet. A world which he would build.
For Rosedale, this was no idle consideration. In 2003, his company Linden Lab launched Second Life, an online world which its residents could alter in any way they saw fit. He believed he was creating the next internet - and, as Second Life thronged with a million avatars, building everything from jazz lounges to full-scale cities, it seemed he might be right. Ford, Sony and Dell invested in real estate. Reuters appointed a Second Life correspondent. At its peak, the site's GDP - residents traded in Linden Dollars, perhaps the first true virtual currencies - was bigger than that of many small countries.
Then, in 2006, Second Life stopped growing. No matter how many people registered, the number of users remained stuck at around a million. In 2008, Rosedale stepped down. The following year he left the company altogether.
Philip Rosedale in the real world
Credit: Jason Madara
Now, as Rosedale examined the output from the gyroscope, the hopes he'd had for Second Life resurfaced with sudden urgency. He called over the seven employees of Coffee and Power, the "stretching exercise" he'd embarked on to cleanse himself from the disappointment of Linden Lab. Together, the team stared at the output fluctuating on the screen. "Look at this," Rosedale told them. "We've got to shut the company down and go back into VR."
Nine months later, on August 1, 2012, a 19-year-old named Palmer Luckey posted a VR headset on Kickstarter using the same gyroscope Rosedale had been testing. Oculus VR raised $2.4 million; in March 2014, it was acquired by Facebook for $2 billion. The VR revolution was ready to roll.
Today, however, almost everything about VR remains unclear. What is its best use? What makes the most engaging content? What, in short, is VR for? “It’s impossible to have a demo of a modern high-end VR system and not come away believing this is part of the future,” says Benedict Evans, a partner at VC firm Andreessen Horowitz (which invested $75m in Oculus VR in 2013). “The question is how much: whether it will be deep but relatively narrow – like games consoles, for example – or much broader.”
Rosedale believes VR should be as broad as humanly possible. His new world, High Fidelity, which launched its open beta on April 27, 2016, is like Second Life, only more realistic and much, much bigger: "As big as Earth and beyond," he predicts. "We are about to leave the real world behind."
High Fidelity has its VR rivals. Chief among them: Linden Lab, the company Rosedale himself founded back in 1999 (and in which he retains a large stake), and which to this day owns and maintains Second Life.
Under new CEO Ebbe Altberg, Linden Lab is building a VR world which goes by the name of Sansar. Whereas High Fidelity is radically open, not only in its code but also in its player, Sansar is closed-source, built for stand-alone experiences.
Open vs closed. Creation vs control. This is how the future of VR turns on a decade-old debate about a world called Second Life.
This cinema is one of Sansar's many small-scale destinations
The demo room at Rosedale's office is a large, open space on the less salubrious side of San Francisco's trendy South of Market district. The room is set up for HTC's Vive headset, with motion sensors in each corner and pistol-shaped controllers on the desk. Rosedale boots up High Fidelity, which opens in a VR games room. Inside, Chris Collins, High Fidelity's director of product development, is showing off his basketball skills. "Look," he says. "I'm passing the ball behind my back."
As Collins speaks, his voice comes from his location, bouncing off surfaces as it would in the real world. His avatar - a handsome young man named Matthew - moves its lips in time with the words. He waves. Our eyes meet. Contact.
In VR, this feeling of being "there" is called presence. Presence is made up of a thousand tiny details - the way a basketball bounces, the way leaves rustle in the wind, the way skin glows as light passes through it - and each one comes with a specific technical fix. Skin, for example, is simulated by a "subsurface scattering effect". The way a face moves is recreated by a machine learning algorithm trained on thousands of videos of actors speaking to camera.
One day, headsets will be packed with sensors able to detect facial expressions directly. For now, these approximations have to do the trick. But the thing about presence is that you know it when you see it. High Fidelity has presence. It isn't lifelike exactly - but it's not so far away.
When Second Life stopped growing, Rosedale could see from the user analytics what the people who stayed had in common. "There was something about them," he says. "The one thing they all had was a huge amount of time to invest in it." Second Life was a retreat for escapists, an outlet for pent-up creativity - a place, as Rosedale once put it, for "smart people in rural areas, the disabled, people looking for companionship". But for less motivated visitors with limited time, it was hard, confusing and alienating.
Inside High Fidelity's clubhouse-style games room
Credit: High Fidelity
Inside High Fidelity, Rosedale enters, wearing the same "Matthew" avatar as Collins. My avatar appears as a ghost-like alien. We head outside to a garden to have a go at Tetherball. The movements for the game come easily, if a little clunkily. "One of the things that's cool about our system is that the physics of it allow us to interact naturally," says Rosedale.
VR's ease of use, says Rosedale, is perhaps the most important thing about it: "A big part of our focus is to make it possible for people to build anything." He had the same goal for Second Life - but there, it took users 40 or 50 hours to become familiar with complex keyboard-mouse commands. In VR proper, you do more or less what you'd do in real life.
For Rosedale, the difference is fundamental. "That is the only reason why Second Life has a million people using it today and not a billion."
Still, the experience is far from perfect. High Fidelity may have hands, but it isn't able to simulate the movement or feeling of fingers. To pick up an object, you shoot a laser-beam at it, then manoeuvre it clumsily into range, where it floats weightlessly on your palm as if it were made of gas.
For Rosedale, this is a problem. Without true dexterity, how would people build things? High Fidelity has 26 full-timers and $15 million in funding, but it can't do everything, so it is forced to leave many crucial details to third parties.
Still, Rosedale is certain a solution will arrive: "Somebody's going to figure out how to build rapidly in here. Once they do that, the sky's the limit." His belief in virtual worlds is absolute. He saw one, in a vision, and he has believed in it ever since.
"A big part of our focus is to make it possible for people to build anything"
As a boy, Rosedale dreamed of virtual worlds. Born in 1968 in San Diego, California, the son of an English teacher and a Navy pilot, his parents moved six times before he was 13. When they divorced in 1981, he returned to San Diego, where, as a shy, technology-obsessed teenager, he built custom, pre-internet networks and sold them to car dealerships. He imagined himself floating in virtual space, his toolkit on his belt, calling structure into being with a wave of his hand.
Aged 16, his fantasy took a more concrete form. Sitting with a friend at his aunt's Windows computer, he pulled up an image of a Mandelbrot set, a never-ending intricate pattern created by a few mathematical rules. He began to zoom in, following starfish and curlicues until he ran out of resolution. He had gone so far, he calculated, that the screen he started from was now the size of the surface of Earth. A thought struck him: if a computer could contain a world, it could also make a world. "That was the thing that drove me," he says. "This obsession. I've got to see that place."
Rosedale realised how he would make his world in 1994. After studying physics and computer science at the University of California, San Diego, he moved to San Francisco and discovered the internet. He instantly saw its potential, not only for communication, but also for a single vast space, created by the processing power of millions of connected computers. But this place would have to be in 3D, which computers couldn't do yet, so instead he founded a video compression startup, which he sold in 1996 to streaming firm RealNetworks (then known as Progressive Networks). "It was cash plus stock, and the stock ended up being quite valuable," says Rob Glaser, CEO of RealNetworks.
Rosedale was a millionaire by the age of 28; in 1998 he was made RealNetworks' CTO. But the lure of virtual reality was ever-present. "Philip talked about the idea the first time I met him," says Glaser. In 1999, Rosedale went with a group of friends to see The Matrix. Afterwards, while his companions gushed over the film's brilliance, he was slumped in a corner, depressed. "That was my dream," he told them. "I wanted to build that." A few months later, he took $1 million of his own money and started Second Life.
In Sansar's immersive world, players don dragon and robot avatars
Rosedale's obsession animates every aspect of High Fidelity - above all its ambition. Whereas Second Life ran on servers owned and run by Linden Lab, High Fidelity is peer-to-peer. By downloading its Sandbox software, anyone with a computer can host their own VR domain. The company plans to make money by charging a fee for domain registration, making it what Rosedale calls "a GoDaddy for VR".
More importantly, this software allows High Fidelity to tap into a vast pool of processing power - enough, Rosedale claims, to have all the world's internet users in-world at the same time. High Fidelity is not simply a cool way of interacting in VR; it is a connected server system, built for huge virtual worlds. As Rosedale puts it: "High Fidelity is the internet. High Fidelity is not a company or a thing. High Fidelity is the net."
High Fidelity's structure is intended to resolve two of Second Life's most persistent problems: scale and latency. Because Second Life was built on Linden Lab's servers, there was often a lag between action and response, and no more than 40 or so avatars could be in one place at any time. But Second Life was built before cloud computing. Why doesn't High Fidelity use Google Cloud or Amazon Web Services? Because these giant server farms are too small for what Rosedale has in mind. "In the next ten years we'll see the internet become VR," he says. "That's not a hosted service. The internet is not a hosted service."
High Fidelity's investors like its ambition. "Every now and then there is a longer-term, riskier but bigger-vision bet that has a hundred-billion potential, and this is clearly one of them," says Steve Hall, managing director of Vulcan Capital, the lead investor in its $11 million Series B round.
So do its hundred or so beta testers. At a virtual public meeting in June, they discuss their hopes for the platform. "The problem I had with Second Life is if you want to build big, it costs a fortune," says a man whose avatar has large, dark rabbit ears protruding from the side of his head. Michelle, a former Second Life player with a brown-haired avatar, sums up the mood. "We saw what we could do before, and we want to do more."
But openness comes at a price. High Fidelity's meeting is being held on the Playa, a Burning Man homage littered with shipping containers and neon signs, including one reading "Rosedale". It is very artful, but amid the intentional mess are signs of genuine disarray. One user is running a script that summons ghosts to float across the sand. Another has added a herd of cows to the plains. Like Second Life before it, High Fidelity is distinctly strange.
"High Fidelity is the internet. High Fidelity is not a company or a thing. High Fidelity is the net"
The physical home of Second Life is a large, wood-framed office near the old dock area of San Francisco. Altberg, the company's third CEO since Rosedale stepped down in 2008, meets WIRED outside the VR demo room. Inside, the walls are covered with framed press reports from the glory days of Second Life: Philip Rosedale in The Washington Post, Philip Rosedale in Newsweek, Philip Rosedale on the cover of Inc. Altberg is tired: he's been in New York on a Linden Lab awayday. Still, he perks up quickly. Earlier, he promised: "Without trying to sound harsh or anything like that, but you haven't seen the best yet."
When Altberg, 52, arrived at Linden Lab in early 2014, Sansar consisted of "three or four people futzing with some low-level stuff out of frustration with Second Life". The game still has a sizeable community and a GDP of "half a billion". Using the proceeds from this "money-generating machine", Altberg has invested heavily, building the team up to 75, more than a third of Linden Lab's staff. The moment he committed completely to VR was when he heard that Facebook had bought Oculus. "As soon as that sold, we were just like, Sansar is going to be fricking awesome for VR. We knew that people were going to want to create content in massive quantities - right now it's too damned difficult."
Linden Lab CEO Ebbe Altberg: "High Fidelity is incomprehensible to normal human beings"
Credits: Jason Madara
WIRED dons an Oculus Rift headset and Sansar starts up in a landscape borrowed from the film The Martian. Bjorn Laurin, Linden Lab's VP of product, is waiting there, wearing an avatar of a cuddly green dinosaur with a woman's face. "You'll notice the lips moving in time with our voices," he says with pride. WIRED looks down. Instead of hands, Oculus controllers float in mid air. "We're getting round to hands," says Altberg. Realism, it seems, is less of a priority here than at High Fidelity.
We cycle through a series of small "scenes" created by Linden Lab's team - an Egyptian tomb; a 3D surfing video; a flat filled with IKEA mock-ups. Each is beautiful, impeccably produced and very empty. With the Second Life players apt to "get twisted around in knots about what it means", as Altberg puts it, Linden Lab has kept Sansar under strict wraps before its 2017 release.
A burly, forthright Swede who was a champion slalom skier in his youth, Altberg started his career at Microsoft, where he remained in the 90s. He worked on Word: Word 2.0, Word 6.0, Word 95, Word 97. "When I left Microsoft, we were dominating Earth," he says. In 2000, he went to a startup offering a marketplace for telephone services, before moving to Yahoo! in 2008, where he ended up running media engineering. He makes products that consumers like to use, he says, in contrast to Rosedale's "geeky" style: "Philip is taking a tech approach, not a consumer-facing approach. That was true with Second Life as well."
Like High Fidelity, Sansar is intended as a place for players to create their own VR content. Yet its scenes will be replicated, not connected. One school group visiting the Egyptian tomb won't bump into another - they will be in separate, identical spaces. These spaces, Altberg expects, will be largely owned by big corporations, just as they are online. "There will be Facebooks and Amazons and all these different things," he says.
For Rosedale, this segmentation makes Sansar something less than a virtual world. "Linden is still providing curated hosted experiences," he says. "So Sansar will still be kind of a game world." He means that it will be artificial, not in its interaction, but in the scope it provides for unpredictability. "It will even be more of a game world unless it's editable like Second Life," he adds.
Despite falling out of the public eye in recent years, Second Life still attracts 900,000 monthly players
Credit: High Fidelity
Unlike in High Fidelity, editing in Sansar won't be in-world: instead, players will make changes in a separate mode. According to Altberg, by keeping editing separate, Sansarcan optimise content to reduce memory usage, scaling up quickly and easily without the need for complicated server structures. "You install a server," he says incredulously of High Fidelity. "Who the fuck installs servers?" Besides, in his view, only professionals want to edit. "Most people are just consumers of experiences as opposed to creators," he says. "It's the same in VR as it is in any other medium, especially when you come to creating quality content."
The biggest difference concerns openness. High Fidelity is radically open, from its code base to its server structure. Sansar is not, and Altberg doesn't care who knows it. "You will have the freedom people, the anarchists, whoever, who will say I want 100 per cent control and it should be open," he says. "Then you will have the vast majority of users that obviously don't give a shit - because how many billions of them are on Facebook every day?
"For years, Second Life ran what I would call organisational anarchy," he continues. "Philip's like, 'Open source, involve everybody'. The idea was that everybody could work from anywhere and participate in this thing. Yet I've seen that you can, with a small team, impact billions - if you do it right."
Rosedale and Altberg meet regularly to discuss their projects. "We're frenemies," says Altberg. "We agreed to kind of have a friendly competition," says Rosedale, who still holds "lots of stock" in Linden Lab, which in turn is one of the "small investors" in High Fidelity's $2.4 million seed round, along with Google Ventures, Kapor Capital and lead investor True Ventures.
The two companies can coexist. "I don't believe there will be a single VR platform, although there will certainly be market leaders," says Jason Jerald, author of The VR Book. "Apple iOS versus Google Android is very much the model here," adds Jamais Cascio, a San Francisco-based futurist who has written extensively on virtual worlds. "Android is more open, flexible and extensible - but also less consistent, less reliable and less secure because of the mish-mash of conflicting open and controlled aspects."
Perhaps none of these differences will matter. After all, it is far from certain that VR will ever reach a mass audience. (To date, around a million headsets have been sold, according to Forrester Research, a sign of low interest or budding enthusiasm, depending on your perspective.)
"Both Sansar and High Fidelity are operating under a very shaky premise," says Wagner James Au, author of The Making of Second Life. "They're gambling their whole companies' futures on the premise that there's going to be a large market consisting of tens of millions of VR owners."
"The vast majority [of Sansar players] won’t give a shit if it’s open – because how many billions of them are on Facebook every day?"
Ebbe Altberg, Linden Lab CEO
At High Fidelity's office, it's easy to forget the ambition and urgency of the project. The grey-carpeted room, its curtains drawn to keep out the summer Sun, has an air of stillness. Developers work quietly at wire-tangled desks, slipping on headsets to gesture unintelligibly in VR. During WIRED's week-long visit, a group of four or five play Overwatch during their lunch break.
To the outside observer, it is sometimes hard to tell where the whole project is heading. There's a sense of looseness, of enquiry for its own sake. One developer, a pale man who comes in every day wearing a "Slytherin Quidditch Team Captain" T-shirt, is building a scale model of the Tardis in High Fidelity. Another has constructed a realistic representation of his flat. "All of us within the VR community are fumbling around right now," says High Fidelity co-founder Ryan Downe Karpf. It's analogous to the early days of the web. It's hard to even know what we're… what we will become."
Towards the end of the visit, WIRED asks Rosedale what he thinks could go wrong with High Fidelity. He pauses, considering the question. "One of the points I've always made about Second Life is that it was well named in a way, in that it was fundamentally competitive with real life. If the real world for your human life, for your love and play, if the real world is there for you - if you live in London, or New York, or San Francisco - why wouldn't you enjoy all this?"
It seems a good question, but for Rosedale it is merely a passing consideration. In the darkened office, there are tasks needing his involvement. And out there, in a land beyond the horizon, there are virtual worlds, waiting to be built.