Microsoft’s Leader in Web Apps: Meet Dr. Flakenstein

Strange things are going on in the lab of Gary Flake. Can this Web geek make Microsoft come alive?

"Can I get you a Coke?" Bill Gates asked.

It was 2005, and Gary William Flake, the head of research for Yahoo, was visiting Microsoft. Finally. Senior executives had been calling for a year, trying to coax him to come meet a few people, hear them out. But Flake thought he already had the best job in the industry.

He hadn't agreed to fly from his home in San Jose, California, to Redmond, Washington, and trade up from his usual T-shirt, until he heard that Gates wanted to meet. That was an offer no self-respecting geek could turn down.

Now here was Gates, greeting him outside his office. The world's best-known and wealthiest technologist, taking his drink order.

Why the royal wooing? Flake had been the chief science officer at Overture, where he had helped build the advertising-search technology that became the dominant Internet business model. Google exploited it better than anyone else. Yahoo had wanted it so badly that it acquired Overture.

Now Microsoft wanted to acquire Flake. But it wasn't his only suitor. Days before meeting Gates, he'd had breakfast with Google cofounder Larry Page. "Are you serious about going to Microsoft?" Page asked. Flake said he wasn't.

Then he sat down with Gates, who sipped a Diet Coke and shooed away his handlers when the scheduled hour was up. They talked for 45 more minutes. Flake had intended to pick Gates's brain, but it turned out the Microsoft founder wanted to pick his. About Internet search and online advertising. How Microsoft, the ultra-disciplined and occasionally plodding maker of desktop software, could turn its advanced research into actual products more quickly -- generating the sort of dazzling Web breakthroughs Microsoft isn't known for. How the company needed someone to inject a dose of inventiveness that could transform the Microsoft culture into something more agile and adaptable. "The humility and candidness stunned me," Flake recalls.

After the meeting, he called his wife to share an unexpected realization: "I think I want to work for Microsoft."

Microsoft's desktop applications and platforms still generate the vast majority of the company's $17.7 billion annual profit. Its products run more than 90% of the world's personal computers. That's what the company is built around, what most of its 92,000 employees focus on. So it's not surprising that Microsoft has been slow to pursue other technologies, especially ones that could potentially disrupt its money machine or that lack a clear business model.

The flip side of this focus, of course, is that Microsoft has fallen far behind on some of the biggest tech growth industries. CEO Steve Ballmer has said that Microsoft's future lies in ad sales, not software sales. He's laid out a vision of "software plus services," desktop applications combined with Internet features. Meanwhile, the likes of Google, Facebook, and Apple have gotten big head starts, running away with the dominant models in Web search, social networking, and online music distribution. Those companies, not Microsoft, are most often praised as innovators.

To Flake, 41, a techie steeped in experimentation and risk taking who calls himself "Dr. Flakenstein" on his blog, that paradox represents a "historic" opportunity -- not just to bring Microsoft up to speed but to advance radically the Internet experience. He's not interested in making incremental improvements, he says, but rather the kind of great technological leaps that Microsoft is going to need to create visually arresting applications that offer new ways of organizing and displaying information on the Web. In Microsoft's unparalleled reach -- more than a billion computer users worldwide -- Flake sees an unparalleled collective power; the more people contribute data to a site, he says, the richer it becomes for each user. This powerful network effect is as thrilling to him as an elegant mathematical solution. It represents Microsoft's advantage.

Microsoft gave Flake the rare and influential role of technical fellow -- a kind of free-ranging visionary with the clout of a corporate vice president. He was the first outsider to be given that distinction from day one. In January 2006, Flake started Live Labs, a rapid-development Web team that exists outside of any specific product group. As he told Microsoft executives last year, paraphrasing the new team's manifesto, "Despite all of the hype that has been somewhat omnipresent with the Internet over the past couple of years, we are still fundamentally undervaluing the total proposition that it represents to society, and all the various industries that connect to it."

Flake reports to Ray Ozzie, a kindred Internet spirit who was hired in early 2005 as chief technical officer. Under Ozzie, Microsoft has launched several innovation labs to create new products for groups such as adCenter and Microsoft Office. Flake's Live Labs is the largest, with 170 researchers, developers, engineers, and designers, about half of whom come from various product groups. It's also the most independent, less bound by the usual rules. "We were the first that could take raw research to a product release, soup to nuts," says Flake.

In less than three years, Live Labs has launched dozens of new technologies (and Flake has filed for more than 100 patents). Some launches take mere weeks, lightning fast for a company with multiyear development cycles; those tend to be new features for existing software. Several stand-alone products have also rolled out, or are about to. Volta is a Web-based service that facilitates complex software programming. Political Streams charts political blog activity with an interface similar to a stock chart; it's part of an eventual blog-analysis product that will identify and track the most-popular Internet memes. And in August, to Google-like buzz, Live Labs released Photosynth, a Web application that offers an entirely new way to view photos online by turning everyday snapshots into 360-degree virtual worlds. (The current version allows you to create a 3-D "synth" using only your own photos; later versions will allow you to crowd-source everyone else's images of the same place.)

The Photosynth launch illustrates Flake's guiding principles of the Web. Thanks to the network effect, ultimately there will be more photos of more places, more ingredients for collaboration, more experiences for people to share. Flake believes that the Internet is becoming a mirror of the physical world. He calls Photosynth "one-and-a-half life," a semi-synthetic world between Second Life and real life.

But Flake thinks his group's most substantial impact will be as a "perpetual startup" that spawns other startups. Since Live Labs isn't tied to a specific product group, it's positioned to reverberate throughout Microsoft, releasing its new technologies and teams to other divisions and, in the process, prodding Microsoft's culture.

"Success," Flake says, "is getting kicked out of Live Labs."

As his nickname suggests, Flake is a bit of a mad scientist, with perpetually tousled hair, a penchant for obscure pop-culture references, and major geek cred: He taught himself how to code at 11, and in his twenties, after penning his PhD dissertation on machine learning, he spent three years writing a computer-science textbook. The Computational Beauty of Nature explores the work of thinkers from Sir Isaac Newton to the fifth-century Greek philosopher Zeno, as well as how neurons process information in the brain and the behavior of ants. When word of Flake's hiring at Microsoft hit the nerd grapevine, it was as if he had betrayed some unwritten compact among hard-core coders. As someone posted on the tech-news site Slashdot, "Fallen to the dark side, young Flake has."

"A younger version of myself might have also looked at the world in those black-and-white terms -- Apple versus Microsoft," Flake says. "But I've been around the block, and I've come to realize that what I thought was good and evil is not. There are many more subtleties, and it's a more interesting world because of that." He sees the challenge of creating new technologies and businesses at Microsoft not as a function of inability or misguidedness, but as a failure of process. As director of Live Labs, he's not only trying to innovate. He's trying to create the best methodology for innovation, which Microsoft's other labs can adopt.

Flake is focused on repairing what he considers a critical shortcoming at the company, the gap between researchers and product engineers. Typically, the former explore technology long term as if they're at a university; the priority is publishing papers. Meanwhile, engineers concentrate on customer needs, reliability, and long development cycles, leaving little time to experiment and little incentive to take chances. Live Labs acts as a bridge between the two, he says. Flake's staffers are, as he poetically puts it, "human Rosetta stones" -- meaning they speak the languages of scientists as well as pragmatists. "By using rapid prototyping, by uniquely bridging research and engineering," says Ozzie, Live Labs "has proven to be a novel and effective method for new idea incubation."

Live Labs' "startup factory" consists of small teams of two or three researchers and engineers who apply for early short-term funding to develop an idea, like a startup seeking seed money. To avoid perpetuating subpar projects (a common flaw in internal incubators), the system is set up so that early funding doesn't guarantee further rounds. Last year, Flake and his senior staff killed a project called Listas, for instance, a list-management application that had been in the works for months. The rejection was considered proof that people were taking chances and that the bar was appropriately high.

Flake demonstrated to his team early on that it was okay to take risks. He wanted Live Labs to have an office in downtown Seattle. Instead of going through channels, he found a loftlike space on his own -- defying the warnings of other execs that locating off campus was "suicide." Now it's the envy of Microsoft, with a living room/screening room and an orange-wallpapered meeting space known as the "bordello." Similarly, when MSN, the company's Web portal, wouldn't promote Photosynth on its home page unless it was rebranded using an MSN name, Web address, and logo, Live Labs launched the service solo.

"We're like a tightrope walker, between acting like a startup and managing relationships within Microsoft," says Alex Daley, 29, Live Labs' group product manager. Daley has solid geek cred of his own: As a junior at Rutgers, he taught computer-science classes, and he was managing the university's IT-services department by graduation. "I make sure we get what we need from the rest of the company without it getting in our way."

It's as if the alpha geeks are slyly trying to reprogram the company. And indeed, on the whiteboard in Flake's office, a colleague has taped the cover of a recent issue of Fast Company, with a picture of Flake's face superimposed on it. The story in that issue was about an ad agency hired to rebrand the company, but the cover line could apply to Flake just as well: "Can This Dude Make Microsoft Cool?"

To understand the challenges facing Flake and Live Labs, there's no better example than Photosynth. The project began after Flake acquired Seadragon, a Web startup whose founder, Blaise Agüera y Arcas, 33, had spent years figuring out how to easily display and navigate huge amounts of visual information. Seadragon can show an aerial view of every page of a book on one screen, then zoom in as close as an individual letter in the text -- all in high resolution.

The acquisition, made two months after Live Labs began, gave the little-known team a "shiny object" to tantalize product groups, says Brett Brewer, Flake's director of incubations and a 10-year veteran of Microsoft. It also demonstrated how Flake could use Live Labs as a recruiting tool, attracting Internet coders who think outside Microsoft's business model. (Agüera y Arcas's geek cred: He started Seadragon at 27 and was recently named one of Technology Review's top innovators under 35.)

Photosynth came together after Agüera y Arcas attended Microsoft Research's annual showcase and saw a 3-D model made from snapshots. A trio of researchers at Microsoft and the University of Washington had written software to recognize the same objects in multiple photos, calculate the depth of field, and stitch the images together.

To turn that research into a seamless consumer experience that happens in seconds on any given laptop -- instead of days on a supercomputer -- Agüera y Arcas formed a team of 15 that raced to create a bare-bones prototype in 2007 to get early feedback from the tech community. Photosynth epitomizes Ballmer's "software plus services" strategy. The photos are stored online, in "the cloud," at Microsoft, while the software performs the data-crunching necessary to synthesize the 3-D rendering on a user's computer.

Photosynth conveys a noticeable flair for design that's unusual for Microsoft. The reason: The ratio of engineers to designers at Live Labs is 10 to 1, not the usual 100 or so to 1. And thanks to Flake, product managers don't automatically have the final say on design. "I'm not saying working with Gary is like working with Steve Jobs," says Don Lindsay, an Apple veteran who had decided to leave Microsoft until he encountered Flake and joined his team as design director. "But Gary is pretty close."

Still, the question remains as to Live Labs' -- and by extension Flake's -- broader impact. Photosynth, like most new Web services, is free, its business model a work in progress. "How does it make money?" Flake says. "There are so many ways. You could link to stores that appear in photos, or individual products, and display an ad or turn it into a commerce event. Imagine what this does for real estate and hotels and tourist attractions."

Even if Photosynth does take off, it'll still deliver a pittance compared with Windows and Office. But Flake is unconcerned. Live Labs, he insists, is about generating multiple projects and leaders that have a cumulative impact throughout the company, financially, technologically, and culturally. "Whenever you make an effort like this, of course you're going to encounter skepticism," Flake says. "There's nothing I can do to convince people to have a change of heart, absent good evidence."

Evidence such as the Photosynth team tripling in size following the launch, when it relocated to MSN. (Plans are under way to incorporate it into various channels, such as Virtual Earth.) And evidence that Live Labs' innovation model is spreading to other parts of Microsoft. Agüera y Arcas, now a leading architect at MSN, is launching an applied research lab, and Ozzie, Flake's boss, has created a new startup lab in Boston. Both are modeled on the Live Labs methodology.

It's early, but Dr. Flakenstein's creation is up and walking.

Add New Comment


  • Nate Lawrence

    Fantastic article on Flake.

    I'm wondering, though, where I could read more about Blaise Aguera y Arcas' new applied research lab that you mention in your last proper paragraph. Are they completely internal or do they have some sort of public face?

    Thanks for any information you can offer.