From the issue dated April 24, 1998

Funds and Freedom Make Microsoft Nirvana for Some Researchers Top computer-science professors are lured away from academe; company wins praise for willingness to let them publish widely and participate in scholarly meetings By VINCENT KIERNAN REDMOND, WASH. Jennifer T. Chayes never planned to be a poster child for Microsoft Research. Two years ago, Ms. Chayes was a mathematics professor at the University of California at Los Angeles. She specialized in applying mathematics to problems in physics, and tinkering with ways in which that work could be extended to solve problems in computing. Then she encountered Nathan Myhrvold, a classmate from her graduate-school days at Princeton University. As the chief technology officer of Microsoft Corporation, Mr. Myhrvold was adding to the cadre of world-class A SPECIAL REPORT: Microsoft's reach in higher education Introduction Microsoft Marketing Brings New Business and New Skeptics Corporate Largesse: Philanthropy or Self-Interest? Microsoft's Campus Brain Trust: $10,000 a Year for Providing 'Input' Microsoft Pays $200 for Mentioning Its Tools Funds and Freedom Make Microsoft Nirvana for Some Researchers Colleges Wonder if Microsoft Is Their Next Competitor Microsoft Headquarters Evokes Ambience of Campus, but 'It's Just an Office Park' What you said: Is Microsoft's growing role on college campuses helping or hurting higher education? Does Microsoft have too much influence in academe? Read the responses we received in Colloquy. researchers in Microsoft's basic-research division, created in 1991. He insisted that Ms. Chayes join them. "I thought he was crazy," she recalls. When she finally agreed to submit her curriculum vitae, she didn't even bother including copies of her published articles. But when she visited the research division's offices on the Microsoft campus here, she was impressed by the talent that Microsoft had assembled, and by the company's commitment to allow its researchers to pursue a wide range of innovative research. She jumped ship and joined. So did her husband, Christian Borgs, a professor of mathematics at the University of Leipzig, in Germany. (If nothing else, the jobs eliminated the trans-Atlantic distance between them.) Today, they jointly head the theory group -- that is, the math department -- at Microsoft Research. "It's like being a kid in a candy shop," she says. In the seven years since the company created Microsoft Research, the division has swelled to include more than 200 researchers. It hires two new people every week, with a stated goal of having 600 researchers by 2000. Thanks to its academics, such as Ms. Chayes and Mr. Borgs, its stature has grown quickly as well. By all accounts, it already is one of the most prestigious computer-science research organizations in the world. The division is aggressively pursuing basic-research goals of interest to computer scientists everywhere, supplemented by the research of university faculty members with grants from the company. But Microsoft Research is also keeping one eye on the possibility of developing new products and improving existing ones. Its leaders say it has already paid for itself many times over. In a series of interviews at Microsoft headquarters here, Ms. Chayes and several other scientists describe Microsoft Research as an academic's nirvana: Funds seem unlimited. Scholars are free to decide what projects to undertake, as well as what to publish. There are no departmental committees to serve on, no bureaucratic hassles, no worries about tenure review. Richard F. Rashid, a former computer-science professor at Carnegie Mellon University, is the vice-president of Microsoft Research. He says the company, in setting up the research division, consciously emulated academe. Unlike many other industrial laboratories, Microsoft Research does not require its scientists to receive permission before submitting a paper to a journal or striking up a collaboration with a scientist elsewhere. 'It's a very academic-like environment, but without a lot of the pain and anguish associated with being a new faculty member," Mr. Rashid says. At many universities, young faculty members in computer science spend a lot of time seeking grants to pay for their research. But Microsoft Research employees, unlike those of some other corporate research arms, accept no outside funds. "You can come here and just do research," Mr. Rashid says. Microsoft Research employees say they are not even expected to focus on areas relevant to Microsoft products. Asked in a joint interview how much freedom they have in setting the direction of their research, neither Eric Horvitz nor David Heckerman hesitates for an instant. "Complete freedom," Mr. Horvitz replies. "Complete," echoes Mr. Heckerman. Their own research seems to bear this out. One recent interest of Mr. Heckerman's has been developing statistical methods that could identify causes and effects -- such as whether cigarette smoking causes cancer -- without resorting to experiments to prove them. The research is unlikely to have much commercial benefit for Microsoft, he says. Mr. Horvitz is co-chairing a Microsoft-financed workshop at the University of Washington this summer that will consider parallels between computers and the nervous systems of living creatures. That, too, is unlikely to lead to new Microsoft products, he says. The division does not extend a business equivalent of tenure to its researchers, but that doesn't seem to bother them. Mr. Horvitz says he feels liberated by the absence of tenure reviews. "We don't have to create an aura about who we are," he says. "Here, you can take more time, be more introspective, and do it right, without worrying about publishing two or three articles per year," Mr. Heckerman says. Working here does have its drawbacks. One is the absence of graduate students, says Anoop Gupta, who joined Microsoft Research in 1997 after 11 years on Stanford University's computer-science faculty. There he supervised about a dozen doctoral students simultaneously, which greatly extended his ability to explore a variety of ideas. And Ms. Chayes, the mathematician, says the seemingly endless supply of money and other resources can, ironically, be a source of pressure, because there are fewer reasons not to pursue a given project. "I can travel to any conference I want, and I can have any visitor here," she says. "One of the things that I have found most difficult here is that anything is possible. It's just limited by my energy." Microsoft's largesse is proving to be a smart move commercially, Mr. Rashid says. The financial benefits of the research division to the company already amount to more than $1-billion, he estimates. For instance, Microsoft's Word 97 word-processing program incorporates an improved grammar-checking tool that was inspired by Microsoft Research's work in natural languages. And artificial-intelligence techniques incorporated into user- help materials for Microsoft's software have cut costs for user support by 18 per cent, he says. "Virtually every product that we sell has had some impact" from Microsoft Research, he says. Mr. Rashid also says Microsoft stands to benefit substantially if it can help spur developments in computer technology that will attract more users -- even if competing companies build on Microsoft's basic research to produce some of those developments. "We need to enhance the value of computing to society," he says. Microsoft Research also has had a positive impact on the company's reputation with scholars. Before the division was created, Microsoft employees were notorious for their silence at conferences -- they would take assiduous notes on presentations and grab copies of whatever materials were available, all without breathing a word to others about what Microsoft itself was doing in research. But today, "the people who are involved in Microsoft Research are publishing as many papers as anyone," says Dan R. Olsen, director of the Human-Computer Interaction Institute at Carnegie Mellon. Mr. Rashid notes that Microsoft Research employees wrote one- fifth of the papers presented at last year's annual meeting of the Association for Computing Machinery's Special Interest Group on Computer Graphics, a prominent conference in the field. This spring, Microsoft Research plans to use its World-Wide Web site to sponsor a "global seminar series" on computer science, says Mr. Gupta, who studies teleconferencing. Live video of speakers at universities around the country will be presented, and the talks will be archived for later viewing. Given the technical nature of the seminars, Mr. Gupta expects fewer than 100 people to watch live, but he says the effort will give people at smaller universities access to cutting- edge research. The division also promotes collaboration with academe through grants that it provides to university scholars. Microsoft researchers identify promising projects by attending conferences and keeping tabs on university research in their fields. If some work seems particularly interesting, Microsoft Research may ask whether that researcher would like financial support, says Dennis Adler, whose title at Microsoft Research is -- really -- director of technology applications and evangelism. To qualify for funds, a campus-based research project generally has to have real-world applications within three to five years, Mr. Adler says. By contrast, Microsoft's product divisions generally support only those university projects that could pay off commercially in 18 months or less, he says. As a company, Microsoft has run into trouble with federal regulators over allegations that it has improperly forced computer manufacturers to promote usage of its Web browser, Internet Explorer, over that of Netscape Communications. Microsoft Research, however, does not dictate that university researchers use or develop Microsoft products. Randy F. Pausch, an associate professor of computer science and human-computer interaction at Carnegie Mellon, recalls fearing that he might upset Microsoft Research by pursuing his strategy for the development of Alice, a program that can be used with a Web browser to create animated graphics. He had received a grant from the division, but for technical reasons had chosen to develop Alice for use with Netscape's browser before extending the program to work with Internet Explorer as well. "That didn't even raise an eyebrow" at Microsoft, he says. Nevertheless, some observers in academe say Microsoft Research seems more interested in hiring the best researchers than in giving them grants. "Microsoft's strategy is to accumulate as many smart people as it can," says Michael A. Cusumano, a professor of management at the Massachusetts Institute of Technology, who is co-author of Microsoft Secrets. The book, published in 1995, explored the company's research programs. Mr. Cusumano has also done consulting work for the company. Carnegie Mellon's Mr. Olsen remembers when his university experienced Microsoft Research's appetite for scholarly talent firsthand. In 1991, the company hired away Mr. Rashid, who was in line to be dean of the school of computer science. Mr. Olsen says, "There is an attitude at times of, 'Why would I want to rent a Ph.D. when I can buy one?'" Some Scholars Microsoft Research Has Hired From Academe RICHARD F. RASHID, a professor of computer science at Carnegie Mellon University who was in line to become dean of the school of computer science when he was recruited in 1991 to become director of Microsoft Research. He recently was promoted to vice-president for advanced technology and research. ROGER M. NEEDHAM, a professor of computer systems at Cambridge University in England and its pro-vice-chancellor, who was hired last year as director of the Microsoft Research campus in Cambridge, England. JAMES T. KAJIYA, one of the world's foremost researchers in computer graphics, who was an associate professor of computer science at the California Institute of Technology when Microsoft hired him in 1994. Today, he is assistant director of Microsoft Research. JAMES BLINN, another highly regarded computer-graphics expert, was a researcher at Caltech's Jet Propulsion Laboratory. He is a graphics fellow at Microsoft Research. MICHAEL H. FREEDMAN, formerly a mathematics professor at the University of California at San Diego, and a recipient of the Fields Medal, mathematics' equivalent of the Nobel Prize. He is a researcher in Microsoft Research's Theory Group. http://chronicle.com Section: Information Technology Page: A31