Microsoft Breakup Could Endanger Top Research Lab
If Microsoft Corp. is split into two companies, its research unit--one of the world’s top laboratories for long-term study of computer science--would be thrown into disarray, several prominent technology academics fear.
“I think it would be a major blow to the research establishment nationally and globally,” said Andries van Dam, professor and co-founder of the computer sciences department at Brown University. “Will it kill off basic research in computer science? Of course not. But it would be a significant blow.”
The potential breakup of the unit, known as Microsoft Research, focuses light on what academic experts say is a deeper problem: inadequate federal funding of fundamental computer science research.
Spending on such research has been flat or declining for nearly a decade, while the importance of information technology to the economy is increasing dramatically, according to a recent report by the President’s Information Technology Advisory Committee.
“This trend threatens to interrupt the flow of ideas and threatens efforts to solve nationally important problems,” the report said. The committee calls for Congress to double the amount of research funding to $1 billion annually by 2004.
“There has been too much emphasis on short-term research, including in academia,” said Dan Reed, head of computer science at University of Illinois. “We’re very concerned. To provide the seeds for the next revolution, there have to be groups looking out over the horizon.”
In its breakup proposal, the Justice Department does not directly spell out how Microsoft Research would be disassembled. But in supporting court papers, the government’s financial advisory firm, Greenhill & Co., said it is possible to separate the research and development assets, patents and copyrights, and distribute them to the two new companies “in a way that preserves viability and promotes innovation.”
If it comes to that, company executives would have to grapple with the mechanics of dividing intellectual property between the applications and operating system companies. Among the thorniest issues: how to divide its research lab and split its 1,275 patents. Many of the company’s inventions are used in the software of both the applications business and the operating system.
Microsoft is one of a handful of technology companies that has been building a lab emphasizing long-term research, and spends an estimated $100 million annually on such work. The company has more than 500 scientists assembled at its Redmond headquarters complex. They are conducting research in 40 fields, including, artificial intelligence, speech recognition, graphics, computer networking and signal processing.
The company’s scientists have made important although not industry-transforming contributions. Researchers have placed software code in a wide range of Microsoft products and have developed ClearType display technology, which improves on-screen readability; natural language processing technology, which enables computers to understand human language and interact with users; and audio compression algorithms, which deliver twice the amount of audio data for use in streaming media.
While the research arm is one of the largest among major U.S. technology companies, experts say that Microsoft’s unit stands apart because of the way it is set up. The company’s researchers are free to decide what projects to pursue, as well as what to publish. They do not need permission before submitting a paper to a technical journal or to collaborate with scientists from other organizations.
“I think that one has to acknowledge that Microsoft has built one of the strongest computer research labs in the world,” said Raj Reddy, professor and former dean of computer science at Carnegie Mellon University. “They have come from nowhere to being among the very finest in a short amount of time. Should there be a breakup, my concern is that one of the first casualties would be Microsoft Research.”
When a company falls on hard times, one of the first casualties is its research and development budget. In the early to mid-1990s, virtually every major company trimmed research and development expenditures as a result of financial woes and a darkened economic environment, said Bob Buderi, author of “Engines of Tomorrow,” a book about corporate research labs.
Overall corporate spending on basic research fell to $5 billion in 1995, down from $6.3 billion in 1991, according to the National Science Foundation.
During that period, IBM cut its research division by 25%, slashing more than $120 million out of its $550-million annual research budget, Buderi said. IBM was experiencing huge losses, nearly $8 billion over three years, and reduced the work force by 100,000.
Since then, overall corporate R&D; funding has increased slightly every year, according to the NSF. But academic experts say that technology companies are emphasizing short-term, product-oriented research and are focused less on risky basic research.
“AT&T;, Xerox and IBM have been the strongest research labs, but they have all declined as their company’s fortunes have faltered,” said Ed Lazowska, chairman of computer science and engineering at the University of Washington. “Of all the emerging high-tech companies, only Microsoft has invested significantly in fundamental research as opposed to R&D; product research.”
Mellon’s Reddy, Brown’s van Dam and UofW’s Lazowska are members of Microsoft Research’s Technical Advisory Board, which means they are consulted by the company on research questions and receive some compensation, though they declined to say how much. Microsoft is one of several high-tech companies that contract for their services and expertise, a common practice in the engineering and computer science fields.
Microsoft formed its research unit in 1991, and has been attempting to build an operation that matches those of top universities and preeminent corporate research labs--one that balances long-term research with the flexibility to rapidly turn innovations into products.
While not household names, many digital legends work at Microsoft Research, including laser-printer inventor Gary Starkweather, personal computer pioneer Butler Lampson, multimedia visionary Linda Stone, and graphics gurus James Kajiya and Alvy Ray Smith.
Rick Rashid, vice president of Microsoft Research, said splitting Microsoft into two companies fails to recognize that research spawns diverse innovations that don’t necessarily fit into predetermined categories.
“Such line drawing would be contrary to the recognition that basic computer research may ultimately provide commercial benefits in several different product categories--or may prove to be of no commercial value whatsoever,” Rashid wrote in a brief submitted to the court. Splitting the unit would “cripple the organization.”
A federal government source said the proposed remedy is likely to stimulate innovation and the research that produces it, not stifle it.
