Cuts in grants for computer research assailed

Government funding for basic computer-science research in the U.S. is decreasing, and many academics believe agencies that award grants are reluctant to pay for longer-term, "audacious" projects.

That was the message Monday from a panel of computer-science experts at Microsoft's eighth annual Faculty Summit, a two-day gathering of about 400 professors and researchers from institutions around the world.

The "feedback loop" of research grant proposals and the government process for reviewing those proposals are "inherently very conservative," said Daniel Reed, a professor at the University of North Carolina and member of the President's Council of Advisors on Science and Technology.

"When funding is scarce, as we feel it has been the last few years, that drives us even more to conservatism and incrementalism," Reed said. "And there isn't a good way to make high-risk bets on ideas."

With that lament as the backdrop, Microsoft announced its own research grant-making priorities for the coming year, including improved power conservation in computing, better tools for analyzing and using data from the human genome, and developing ways to use the cellphone as a tool for improving health care.

The company plans to make $4.7 million in new grants to outside researchers.

Microsoft also announced funding for a partnership with the University of Washington on collaboration technologies used in distance learning.

Craig Mundie, Microsoft's chief research and strategy officer, said grant making should be more entrepreneurial, putting money behind outstanding individuals and letting them move the field forward in a model more akin to venture-capital funding.

Reed said the President's Council is looking at ways to "change the funding mechanism and the reward structures inside government to try to incent more of that creativity."

The current grant-making structure also tends to discourage multidisciplinary projects that combine the work of a computer scientist with a biologist, for example. This is another area panelists agreed the computer-science field needs to emphasize.

Too many people doing rigorous scientific research — the kind that increasingly requires a computational element — "aren't really contemporary in computing," said Mundie, one of the two executives set to assume Bill Gates' responsibilities at Microsoft when Gates reduces his involvement with the company next year.

Mundie cited an example of a physics project he reviewed recently in which three-quarters of the researchers writing software were not professionally trained to do so.

"There's some coupling, but not at the level that I think is going to be required to solve society's biggest problems," Mundie said.

The National Science Foundation (NSF) is launching a grant-making program meant to encourage this kind of multidisciplinary research with computer scientists, said Jeannette Wing, an assistant director for computer science at the NSF.

About $52 million will be allocated to the new program in the 2008 fiscal year, she said.

At last year's Faculty Summit, attendees worried about declining interest among students in the field and the implications for academic institutions and companies that rely on highly skilled computer scientists. (Microsoft Research has grown to nearly 800 people in 15 years.)

The "talent pipeline" was not the prominent topic it was a year ago, but still a top-of-mind issue.

"To the extent that we are able to attract the best and the brightest to the discipline, that will determine the success of the discipline long term," said Sailesh Chutani, director of external research and programs for Microsoft Research.

As part of a growing focus on early-career computer scientists, the company said it will host an event aimed at graduate students and newly minted faculty next year.

Benjamin J. Romano: 206-464-2149 or bromano@seattletimes.com

Information in this article, originally published July 17, 2007, was corrected August 19, 2007. Microsoft will be funding $4.7 million in new outside research grants in the coming fiscal year, not $3.7 million as reported in a Monday story about its research division. It is starting a $1 million program called the A. Richard Newton Breakthrough Research Award, which was not included in the reported total.