“The Best Place in the World” for Interdisciplinary Research: A Talk with Microsoft’s Jennifer Chayes
After publishing my story yesterday about the opening of Microsoft’s newest research lab in Cambridge, MA—where social scientists and computer scientists will work side by side to understand technology-mediated phenomena such as social networking—I attended a Microsoft-sponsored launch symposium at MIT and had the opportunity to meet with the lab’s director, Jennifer Chayes. Though Bob has spent a bit of time with Chayes and her husband and deputy director, Christian Borgs, this was my first time meeting her, and I have to say that Microsoft could hardly have picked a more dynamic and outgoing person to lead its first formal research facility in the Hub.
Chayes labels herself as “loud, crazy, and intense,” and jokes that she’s probably a headache for her boss Rick Rashid, senior vice president of Microsoft Research. If so, I’m sure it’s one of those worthwhile headaches: Chayes is clearly one of those exhilarating, exhausting types who just can’t stop making connections between ideas and people.
She’s no stranger to the Boston area, having done her postdoctoral work in mathematics and physics at Harvard in the 1980s. But she’s been away for a while, serving as a professor in the math departments at UCLA and the University of Washington and, for the last 11 years, as part of Microsoft Research, where she gained recognition for her work on game theory, the modeling of random graphs (of which the Internet and the Web are examples), and “phase transitions,” or sudden changes in the properties of a graph as it grows.
Chayes says it’s good to be back—and that Cambridge is the perfect place for a Microsoft lab. “Being located right next to MIT makes a huge difference,” she says. “We have faculty and students coming over all the time. We have major relationships going with Harvard, BU, and MIT—and I’m going out to speak at Northeastern and Tufts. Things are taking off so fast, it’s unbelievable.”
The Microsoft lab’s newest hire, announced yesterday, is also a Boston returnee: the deliberately uncapitalized danah boyd, a Berkeley-trained ethnographer who is famous in the blogosphere for her ethnographic studies of teen behavior on social networks like MySpace and Facebook. Currently a fellow at Harvard Law School’s Berkman Center for Internet and Society, boyd will start at the lab full-time in January.
“I just love the way she looks at the world,” Chayes says of boyd. “I’ve already started talking with her about how we can modify some of the standard models of online networking to account for the kinds of behaviors she observes”—for example, the fact a social networker’s universe of online friends tends to shrink as he or she ages. “If we had more accurate models, we would be able to much more effectively come up with things like recommendation systems that give you recommendations based on the strength of your ties and the number of ties you have to others,” says Chayes.
Over lunch, I asked Chayes about a range of issues, from the genesis of the interdisciplinary focus she and Borgs have outlined for the lab to how she plans to measure the success of the new group over time.
Xconomy: You’re just celebrating the official opening of Microsoft Research New England with this symposium today, but you’ve actually been in business since mid-summer, right?
Jennifer Chayes: We’ve been in operation for about two and a half months now, so we’ve already got a lot of research projects going, but we waited until now because we wanted the universities to be in session and we wanted to draw in people who didn’t already know that we were here or what we were doing. And the vast majority of the people here today are from the university community. It’s very interesting for me, because some of them I haven’t seen in 23 years, since I was a postdoc.
X: Considering that your own background is in game theory and cryptography, I’m wondering where the idea for the Cambridge lab’s interdisciplinary focus came from, where you’re bringing together people from economics, the social sciences, and traditional computer-science theory.
JC: Well, if you look at my resume, I’ve changed fields every few years. My undergraduate degree was in biology, my graduate work was in physics, I was a professor of mathematics, and then I ran a theoretical computer science group. And over the last three or four years, I have become very interested in networks. I see them everywhere. It has just become clear to me that mathematicians, physicists, and computer scientists have done about all that we are going to be able to do without connecting with the people from the disciplines whose phenomena we are trying to describe. We have to understand how people are truly interacting over social networks, what their motivations are, how you put incentives into a system, how you understand the incentives implicit in a system, and the unintended incentives in a system.
It’s already been fantastic. For example, we have an incredible economist visiting us, Susan Athey from Harvard. She is an expert on auctions and she advises governments on large-scale auctions like FCC spectrum auctions, and now she’s getting into the auctions used to sell ads on search engines. She’s working with our data now, and Microsoft is completely open to considering modifications of our own [online ad auction] algorithms based on what she sees.
X: It’s great to see you hiring people like danah boyd, but she’s really an ethnographer and an observer. I’m not sure how easily you can translate her findings about adolescents on MySpace into algorithms that might eventually turn up in products at Microsoft.
JC: I’m not talking about product development. I’m not a ‘let’s see what widget we can build tomorrow’ person. And that’s not the kind of people that Rick hires. If we have a really great idea, we are going to want to see it go into a product. But we are not hiring danah for that. I think that danah is coming here for the same reason I came here—this is a place where she will be able to expand the boundaries of her research. If she were going to an academic department, she would have to do much more mainstream work to make sure she got tenure. And that is not necessarily what is going to lead to the best work. She should be allowed to flow wherever her instincts take her, and she has very unconventional instincts. Microsoft Research is the kind of place where that will thrive.
X: I asked Rick Rashid this question earlier today—in an open-ended environment like the one you’re talking about, with no commitment to finish a product, how do you evaluate the success of your researchers?
JC: The same way I would evaluate the success of a faculty member if I were still in academia—by the impact a person has. Have they advanced the state of the art? Maybe in some cases they have had an impact on a Microsoft product, but that is not the only way to excel here by any means. I would rather my researchers have a tremendous intellectual impact than build a better widget. It would not be constructive to hire researchers to do product development, anyway—we hire amazing product developers to develop products.
We do make sure than when we find something that would improve our products, it gets deployed. That’s what separates us from a Bell Labs or a Xerox PARC. But our value is in the big ideas and the changes in perspective. What drives us is trying to see over the horizon and trying to help create that future, and making sure that Microsoft is part of it.
X: And one level up—how do you measure the success of a Microsoft research outpost, whether it’s in this Cambridge or the other Cambridge, or even Beijing?
JC: We’re looking for impact on the academic community, impact on Microsoft, and collaboration across laboratories. But to demand that on any short time scale is not the right way to do it. It’s not really different from academia—if you ask that a faculty member have impact on a short time scale, you are going to encourage second-rate work. In fact, danah talks in her blog about her decision to come to us, and she poses herself the questions that she thinks people are going to be asking, and she says she’s looking at her colleagues in academia who are jumping through hoops trying to get grants. And if you look at the time scales on which academics have to produce these days, I would say those are more constraining than the time scales on which we have to produce.
There is no question in my mind that when Microsoft looks at this lab 10 years from now, they are going to say, ‘Boy, that was a smart investment.’ But they may not be looking at it that way in two years, and that’s okay. These things have to be judged on very long time scales, because great science is done on long time scales.
X: You must feel lucky to be part of a company where you have the luxury of thinking on those time scales.
JC: I feel very lucky, but quite frankly that’s the only reason I am working for a company. I was a tenured academic, and I get offers from academia all the time—everybody here does. But there is no other company in the world I would consider working for. What I am trying to foster is a new kind of interdisciplinary research, and I feel that Microsoft is the best place in the world to make that happen. If it weren’t, I would leave.