The Chief Of Microsoft Research On Big Ideas, Failure, And Its New Skunkworks Group

As the company searches for its next big thing, Microsoft’s scientists are striking a tough balance between short-term R&D and disruptive, blue-sky ideas.

The Chief Of Microsoft Research On Big Ideas, Failure, And Its New Skunkworks Group
[Image: Abstract via Shutterstock]

Google’s secretive R&D lab Google X gets lots of attention for testing and developing moonshot ideas, even though its work has touched few consumers. Fast Company took readers for a detailed behind-the-scenes look inside the Google X operations in April–a world of driverless cars, high-flying Wi-Fi balloons, and even space elevators.


The search giant’s rival Microsoft–a company that could use some disruptive ideas as it struggles to gain major new revenue streams in the shifting computing market–is now taking a cue from its competitor and launching a “Special Projects” group, headed by former deputy director at DARPA Norman A. Whitaker and under the umbrella of Microsoft Research, the company’s sprawling university-like research division.

At a lab office ribbon cutting in New York City, Co.Exist spoke with Microsoft Research chief Peter Lee, who last summer stepped up to oversee his division’s 1,150 scientists and engineers. His comments provide a glimpse into how the $78 billion company is changing the way it thinks about innovation.

Lee isn’t willing to go into much detail about the new Special Projects division, but he does say it will focus on disruptive ideas.

“Roughly speaking, it is a part of Microsoft Research whose primary focus is to expand what people imagine is possible,” he says. The comparison to Google X–a division headed by Astro Teller, who took a class under Lee at Carnegie Mellon University–isn’t unfair, he says. But Microsoft’s Special Projects may have more of a focus on delivering. “We would really be looking to not only inspire, awe, and surprise, but also ship. So I think we’ll have to, as we do throughout all of Microsoft Research, do some pretty fundamental things–but also have real impact.”

Lee thinks of the idea of disruption in a more specific way than most. He draws an “investment map” on a whiteboard to show what he means. At MSR, he asks each lab director to make sure his or her research portfolio covers all quadrants on his map: mission-focused research, sustaining research, disruptive research, and blue-sky research. “We invest in research that is reactive–the product team comes to us with a hard problem it doesn’t know how to solve, and we’ll work with them to solve those problems. Then, as you go up on the axis, you have the more classical open-ended search for truth and beauty.” Special Projects, he says, will focus on the “upper-left quadrant” (i.e., disruption).

Of course, it’s the specifics that matter when executing these artificial distinctions. Where does one draw the line between big ideas that are just close enough–like self-driving cars–and big ideas that are just crazy?


“I think that nothing is too far-out, roughly speaking. When I was at DARPA, one of the lessons I learned there . . . was how fragile and fleeting disruptive ideas are. On Monday, you could come up with this great idea–just the coolest thing and you could be just excited about it. You go home, go to bed, wake up in the morning, and think, What was i thinking? You could go and have a meeting with a bunch of smart people, and they could just punch holes in it left and right. So now you are left in this confusing state: Was it a good idea or not?”

What’s left, he says, is the thin line behind a visionary technical concept and science fiction. “For the most important disruptive ideas, it’s almost impossible to know in the heat of the moment which side of that line you’re on,” Lee says. “In an operation like ours, it is important to err on the side of going forward–with protecting the idea and pushing it through. With Special Projects, or anything else that we do at Microsoft Research, we need to have mechanisms to balance in that direction.”

Every week, however, Lee struggles with the question in meeting with his leadership team. There are success stories of out-there ideas that keep everyone going, like the time in 2009 when two Microsoft researchers decided to try an entirely new analysis method–called deep neural nets–that ended up leading to the biggest improvements in speech recognition technology in a decade.

“If I were king of the world at that time and had to cut projects, I would have killed that project. And I think any world-class speech researcher would have agreed with me,” he says.

“If I had just looked at the passion and the dedication of those two researchers, then the right thing would have happened. Luckily, the right thing did happen,” he says. “Now the question is how long do you sustain that idea before you give it up for dead? That’s the thing I struggle with every day. At what point do you conclude an idea is not valuable?”

Today at Microsoft, that balance is a struggle. Scientists are being asked to help product teams like never before, especially after a reorganization last year that eliminated the company’s traditional product groups and created more “horizontal” divisions that span different products. Lee’s job is to help balance his team’s desire to help the short-term business with MSR’s ultimate mission.


“The opportunities to collaborate with all of these teams has just increased dramatically,” says Lee. “Everyone’s just really motivated to make an impact on Microsoft’s business. And the product teams are just welcoming researchers with wide-open arms. There’s been this kind of big rush by researchers to really do a lot for the company,” he says.

The challenge, however, is not losing sight of the longer term. “Ultimately Microsoft Research exists to look further ahead–to look beyond the headlights, to understand what will be the next big thing.”

About the author

Jessica Leber is a staff editor and writer for Fast Company's Co.Exist. Previously, she was a business reporter for MIT’s Technology Review and an environmental reporter at ClimateWire.