By Rob Knies
September 26, 2011 6:31 PM PT
Peter Lee is bullish on the future of computer-science research.
Little surprise there. It’s been exactly a year since Lee, formerly with the Defense Advanced Research Projects Agency and, before that, head of the Computer Science Department at Carnegie Mellon University, became a Microsoft distinguished scientist and the managing director of Microsoft Research Redmond.
And as Microsoft Research marks its 20th anniversary with a series of symposiums—led by the managing directors of its research facilities in China, India, the United Kingdom, New England, Silicon Valley, and Redmond, and intended to provide a day when leading scientists, researchers, and students can discuss technological shifts that are dramatically changing the computing landscape—Lee takes pause for a few minutes to explain why he is so optimistic about the future of computer-science research.
It isn’t difficult to identify some of the current-day computing themes that could help define society in the 21st century. The development of natural user interfaces that enables the use of gestures, touch, voice recognition, environmental and context awareness, and adaptive behavior—stitched into an immersive whole—would, with only a small learning curve, enable users to interact with a computer as they would with a person.
Similarly, advances in machine learning and the manipulation of massive data sets figure to transform the things computers can do and what can be done with them.
Such developments, Lee says, will lead to a not-so-distant future in which the potential for computing to change the way we live and work will be unprecedentedly vibrant and fruitful.
“The number of potential and real breakthroughs waiting to have tremendous impact on people’s lives is huge,” he says. “It has just exploded.”
One way, he suggests, to consider the impact of computer-science research is to look back to 1991, the year Microsoft Research was created, to see what emerging technologies make use of research ideas from that year. The list, Lee observes, is dramatic, including:
“Zoom ahead 20 years to 2011,” Lee says. “The amount of stunning, deep research results has expanded greatly. You can ask the question, ‘What are the Bill Gates, the Linus Torvalds, the Tim Berners-Lee, the Jim Gosling of today trying to do with computer-science research that will have the same kind of impact 20 years from now?’ The opportunities are dramatically bigger than they were in 1991.
“The access to powerful technology, to massive computing and storage through the cloud; the access to amazing development tools; the access to literally every major research publication, available today on the web in a way that it wasn’t at all in 1991—the access to all of that creates a democratization in access to research results that is incredible. I would expect that in 2031, 20 years from now, when we reflect back on what got started in 2011, we won’t be looking at just five examples; we’ll be looking at 500.”
In fact, Lee notes, the evolution of industrial research facilities virtually dictates that Microsoft Research is positioned to accelerate its history of working in interdisciplinary fashion with academic, industrial, and governmental partners to deliver breakthroughs that extend the frontiers of computing.
“My view is that Microsoft Research, as illustrious and as successful as it’s been—and it’s been hyper-successful; the recruiting power is massive, the quality of the researchers and the engineers here is unparalleled, and their work finds its way into products such as Windows, Office, Bing, and Xbox—even with all of that, it has only now grown up. The best and most important ideas are still to come.
“That’s not so surprising. If you look at some of the great industrial research labs, like Bell Labs, as successful as they were from the start, it wasn’t until the third decade that really big things like the transistor or UNIX emerged.”
As he explains it, the first decade for such organizations is all about the ability to recruit great personnel, which leads to a reputation as a legitimate force in the research community. The second decade is concerned with growth, among individual researchers and as a lab. Once a certain critical mass is achieved, then, in the third decade, if the investment has been well-managed and sustained, people begin to turn their sights to a legacy that extends beyond individual fame and respect to a more meaningful impact with the potential to effect a significant societal difference.
“I think the visionary, sustained investment of Microsoft Research for two decades is likely to pay off in the same way,” he says. “Over the next five to 10 years, not only will we continue to see things that have tremendous impact on human knowledge and on Microsoft’s success in the marketplace, but we also will see other work that will be transformative in the world in the same ways that the invention of the transistor was transformative.
“The primary reason I’m here is that I see Microsoft Research as a place that it is at that point, that’s poised to do something like that.”
One of the hallmarks of Microsoft Research over its two decades has been its willingness to collaborate openly with the greater research community. By supplying knowledge, tools, and funding, Microsoft Research aids in scientific discovery that leads to breakthroughs for Microsoft and for society, all fueled by people and their passion to turn research ideas into reality. That passion helps to shape the future and gives Microsoft the agility to respond immediately when the world changes—or to change the world itself.
“What we, and the computer-science research field more generally, are motivated to do is to break out of a mindset of thinking inwardly about understanding computing and the mechanisms of computing,” Lee says, “and instead look outward to the role of computing in the world. The success of computer-science research is expanding rapidly, but for all of that dramatic progress and expansion, the gap that people see between the progress of computer science and what society needs from computing … People still see that gap not shrinking.
“It’s almost like a Catch-22, where the more we discover and understand about computing, the more we start to believe that computing can do in the world. Ten years ago, it would have been very uncommon for anyone to connect computer-science ideas with improving health care or addressing energy independence, but today, we’ve learned so much more about the possibilities in computing that everyone says that computing has a crucial role to play in making people healthier or improving energy efficiency. Suddenly, there are whole new sets of opportunities or problems—and gaps between computer-science research and societal needs have opened.”
How, then, can the value of research be measured?
“We don’t do it by dollars,” he says. “What I ask from managers is to think about impact: ‘What high-impact results have you produced? Give me something to brag about. Tell me how you’re affecting Microsoft’s product groups and businesses. Tell me how you’re contributing to the research community. What is your plan? How are you structuring your team to position itself to make progress?
“There is always a temptation to connect everything we do in research with financial impact in one of the Microsoft businesses, but that’s a pretty fruitless exercise. For one thing, researchers come up with ideas; there’s a huge gap from those ideas to making a product. Tracking through that whole, long chain of events is not possible. We like to know when our ideas have been incorporated into Microsoft products, but it stops there. We don’t try to tally how much of the profit of a business accrues to a research idea.”
Other ways to measure the value of research can include the success of its academic interactions and how it enhances the organizational ability to attract great talent.
“We support a lot of young researchers through fellowships and internships,” Lee notes. “It’s not hard to see the impact on our recruiting. A lot of these interns go away, write stellar Ph.D. dissertations, and want to come back and work for us full-time—a surprising number.”
Add it all up, and maybe computing research is starting to progress toward a future vision held by many.
“There is a dream I have that’s shared by a lot of computer-science researchers,” Lee says, “where computing power and communications becomes so embedded in everything around us that it recedes into the background.
“Today, we think about having a special machine on your lap or on your desk. You have to sit in front of it, you have to learn to operate it, and the computer is going to consume 100 percent of your time and attention while you’re doing it. The dream is to break free from that: There is no longer this primary mode of interaction of operating a special machine, it’s just all around you, and it knows what you want and understands you over the long term.”
Of course, there could be, as Lee observes, certain consequences.
“That’s a dream,” he smiles, “but one that might reduce, in the distant future, demand for scientists like me.
“Let’s face it: As long as computers are special, computer scientists and computer-science researchers will always be special, also. But maybe it’s not such a bad thing to have a dream of putting us all out of business.”