Industry urges feds to set milestones

SAN JOSE Calif. - he federal government has computing requirements that the information technology industry cannot satisfy and agencies need to set clear milestones for investing in research to meet these demands some of the nation's top computer scientists said at a trade show here last month.

At a "town hall" meeting held by a presidential advisory panel that is developing recommendations for a federal IT research agenda academic and corporate computer researchers said the public and Congress need to better understand why investments in more powerful computer systems and networks are important.

Even though desktop computers are becoming fast enough to support some high-end applications agencies will still need cutting-edge machines for such problems as medical research and defense scientists attending the SC '97 supercomputing show said. For example a scientist from the National Cancer Institute said biochemistry research such as analysis of one aspect of the HIV virus requires too many calculations to be done effectively with anything other than a supercomputer.

"You cannot do these problems on a microprocessor and you never will be able to " the scientist said. "Some areas will clearly be handled by Bill Gates and others " said Sid Karin director of the National Partnership for Advanced Computation Infrastructure a coalition of research groups that includes supercomputing centers funded by the National Science Foundation. The advisory committee should focus its recommendations on "things the federal government needs to have done and needs to worry about " Karin said.

Ken Kennedy co-chairman of the advisory committee and director of Rice University's Center for Research on Parallel Computation said "You've provided a skeletal framework on what our report should say. It would be very important for me to try to get some opinions from people" as to what those areas include. NGI: No Benefits Specified Panel members at the forum noted that a major reason Congress was reluctant to provide all the funding requested for the Clinton administration's Next Generation Internet project this year was because it felt agencies had not specified the benefits that the advanced system would provide.

"We can't go in and advise the government that they should make investments without saying what we will get out of it " said Susan Graham a computer science professor at the University of California at Berkeley and a member of the advisory group. "One of the things our community has done is intertwined goals and milestones " she said. "We need to think about the fact that we're always going to be pushing the envelope."

Others said however that the federal government's research budget should not be constrained by foreseeable benefits. "There's also real value in high concept " one scientist said. "One has to be more radical in trying to articulate a research agenda." Kay Howell the incoming director of the federal National Coordination Office for Computing Information and Communications which will manage the administration's response to the panel's recommendations said in an interview that the discussion "very much captured issues that have existed for some time."

She agreed with Graham that the research community needs to set milestones. "This will be forever I think a progressive effort " Howell said. "That was such a key point." Kennedy said the panel expects to produce a preliminary report in March so that its recommendations can be considered during congressional debate on the fiscal 2000 federal budget.