REDMOND, Wash., Sept. 5, 2001 — A lot can change in 10 years, especially in the computer industry. New technologies are born, grow and are replaced by another generation of new ideas. Similarly, companies can change directions in order to stay afloat in this dynamic market.
But ask the people at Microsoft Research (MSR) and they’ll tell you that a few things have remained constant since Microsoft announced the founding of its research arm 10 years ago this week. They point to Microsoft’s steadfast support and funding for MSR’s mission of pursuing basic research unfettered by industry fads or product-development cycles.
Since September 1991, MSR has grown from a longtime aspiration of Bill Gates, Microsoft chairman and chief software architect, into a respected research organization with nearly 600 scientists and staff at five locations on three continents. The research MSR does influence virtually every product that Microsoft releases. It also advances the debate at scientific gatherings worldwide with a constant supply of research papers.
Equally as important to Microsoft, MSR is helping shape the future of the company and computing by expanding the potential of the PC and other devices, as well as imagining new possibilities and solutions to potential barriers that may develop along the way.
“Microsoft Research is brimming with ideas and enthusiasm,” says Dan Ling, vice president of MSR. “And the fact that we are at Microsoft gives us a fast, direct path to turning those ideas into products that millions of people will use every day.”
First of its Kind
The success of MSR was far from certain in 1990 when Nathan Myhrvold, then the company’s chief technology officer, proposed the idea of creating a research organization within Microsoft. Although Gates had long considered the idea of a research lab, Myhrvold’s proposal was radical for several reasons. Most government and industry research budgets were shrinking at the time. And unlike most corporate research labs, the organization Myhrvold proposed would be located on the main Microsoft campus in Redmond to encourage transfer of MSR technologies to Microsoft products.
There was another radical element. “No one had ever created a basic research group in computer science in a software company before,” recalls Senior Vice President Rick Rashid, whose hiring is recognized at Microsoft as the founding point of MSR. Myhrvold and others at the company recruited the former computer-science professor from Carnegie Mellon University in Pittsburgh. His resume included development of the groundbreaking Mach operating system.
Gates and the rest of the Microsoft Board of Directors approved Myhrvold’s request for US$10 million to begin building MSR. Myhrvold and Rashid modeled the fledgling organization after university computer-science departments such as that at Carnegie Mellon. Researchers were free to pursue the work of their choice, with no formal requirements to develop technologies for specific Microsoft products. They also were expected to get their work published in academic journals and other publications — a foreign idea at most corporate labs but important at MSR because of Microsoft’s goal of advancing the overall field of computer science, not just its own products.
As Myhrvold predicted, the close proximity of researchers and product teams soon began to spark partnerships and other types of efforts that led to enhancements to many Microsoft products, including Office and Windows. These relationships continue to develop today. (See related story, Microsoft Research Contributions Keep Microsoft Products on the Cutting Edge .)
“There’s more than one way to make this happen,” Rashid says. “Basically, technology transfer is a full-contact sport. It depends heavily on people just talking to one another and interacting.”
Within a year, MSR had 28 researchers and staff in three groups: Natural Language Processing, Programming Languages and User Interface. By mid-1995, there were 176 researchers and staff in 12 groups, and a second lab, the Bay Area Research Center (BARC) in San Francisco. Two years later, MSR added its first lab outside the United States, in Cambridge, England. The following year, 1998, a second overseas lab opened in Beijing, China. These overseas labs now include almost 200 researchers and staff working in 17 different groups — which represents nearly a half of the research groups now within MSR. Each group works to solve problems in a different area of computer research.
“Today, we have a very diverse group of researchers working all over the world,” Rashid says. “We also have some of the world’s best computer scientists, statisticians, mathematicians, physicists, psychologists, sociologists — you name it.”
Apart from the constant flow of innovations Microsoft can add to its products, Microsoft’s annual investment in research — currently approaching US$5 billion — serves as a down payment on the future of computing. With MSR now doing research in so many areas — everything from computer programming languages to virtual reality — Microsoft is confident the company can keep shape or at least keep pace with the changes in the computer industry.
Rashid points to Moore’s Law, the time-proven axiom that computing power doubles every 18 months, as just one indicator of the need for technology companies to constantly look for new answers, as well as predict and understand upcoming problems. “If things are changing at an exponential rate, everything up to now doesn’t count,” he says. “You’re constantly throwing away your past. It’s no longer relevant. You have to always be willing to learn.”
The work of MSR senior researcher Gordon Bell is a prime example. The founding chairman of the MSR Technical Advisory Board, Bell has been dubbed the father of the VAX minicomputer. He also played a defining role in the growth of the Internet when he led the National Science Foundation, which planned and ran the development of the Internet at the U.S. Government agency.
Coming Soon: Virtual Immortality, Automatic Video Editing
Bell, who began working at MSR six years ago, is now testing the limits of the personal computer for storing and cataloging personal information and experience. As part of a research project he calls “Capture All,” Bell has begun scanning and electronically storing nearly every piece of information he comes in contact with — everything from airline tickets to newspaper articles — in his computer. He’s also electronically storing all of his books, CDs and photos. The next phase of the project, which he calls “mymainbrain,” is to capture conversations, interviews, meetings, presentations and personal videos on his computer.
Increased cataloging of information, Bell says, will allow people to more easily pass down accurate records to family members, creating a new form of immortality. It also will simplify tracking and accessing information in increasingly congested lives.
“With the vast amount of information flowing out of our fingers and eyes, we need help dealing with it,” he says. “Not having to remember everything is vital.”
Kentaro Toyama is among the MSR researchers tackling another challenge: making it easier to browse and edit videotape on computers or other devices. “People who own video cameras often find that after a week of shooting home movies, the video camera ends up in the closet,” Toyama says. “That’s because home videos are amazingly boring if you have to watch the entire video end to end to find a few select scenes. What’s needed is a tool that makes it easy to create short highlight videos from raw footage.”
He and other members of the Media Space Project at MSR are developing such a tool by applying basic research they have been doing for several years. The technologies will allow an editing device to stabilize jerky video, detect and track people, and segment video into short, easily browsed clips. In the long term, Toyama and his colleagues hope that increasingly less of the editing process will require user intervention. “The ultimate goal is a tool that can automatically extract meaningful highlights from raw video,” he says.
Future Visions of Computing
A number of MSR researchers are looking beyond software and applications to help Microsoft understand the future of computing, predict roadblocks and plot courses around these barriers.
Marc Smith, an MSR sociologist, spends his time imagining how the rapid evolution and mushrooming power of wireless and other technologies will eventually fit together and, ultimately, alter the way we all live our lives.
Among his predictions: “We will be eating a lot of computers in the future.”
Smith imagines a not-too-distant future where microscopic computer circuits and sensors will be smart and inexpensive enough that they will literally be everywhere. These “bots” will become so pervasive that they will even be in the cereal we eat, analyzing our insides and letting us know if our daily fiber is doing its job, he said.
Smith, a member of the Collaboration and Multimedia Systems Group at MSR, envisions a future where Microsoft’s .NET and related technologies change the way people do almost everything, including how they commute to work and even meet one another. He expects high-speed networks and servers will help workers find a ride home with a coworker within minutes of stepping away from their workstation. These networks will help link people who want to buy and sell items, maintaining privacy through double-blind electronic interactions until both parties are ready to communicate directly. Online reputations, maintained on computer networks, will store a person’s online history and verifiable personal information, providing other users an additional assurance, he said.
Similarly, Smith envisions people sharing information such as personal restaurant or product reviews via handheld devices. They will access this information by scanning their personal devices over barcodes, which he expects will eventually become as ubiquitous as brand labels and logos are today.
“The future of personal computing won’t be about virtual reality,” Smith says. “It will be about augmented reality.”
Senior researchers Jennifer Chayes and Christian Borgs, with others members of MSR’s Theory Group, are looking outside of computing to help solve some of the problems the industry faces now — and even some it may face in the future. One example is phase transitions, abrupt changes in the state of a physical system, such as when water boils and turns from liquid to gas. Chayes and Borgs hope to find parallels to abrupt changes in computer networks, such as system crashes, that might help researchers better understand and eliminate these problems.
“If I look at a problem in computer science and can find a parallel or analogy from physics, then I can bring all the years of experience in physics to bear on the issue,” Chayes explains. “We’re trying to mimic Mother Nature, in a sense.”
The Theory Group is also looking into the potential benefits and hurdles of future technologies such as nanotechnology and quantum computing, two potential replacements for the silicon chip when its power and storage potential are no longer great enough to meet the demands of the computer industry.
Nanotechnology is based on the idea that single molecules may one day serve as transistors within computers, as opposed to silicon, which requires hundreds or thousands of individual molecules. Quantum computing would dramatically increase the speed of computing because it takes advantage of the parallelism inherent in quantum mechanics. But this parallelism is extremely fragile, and therefore requires elaborate error correction, which may or may not be practical, Borgs explains.
“Whatever option prevails,” Chayes says, “we are going to be there, coming up with the algorithms that will help make it possible.”