Shifting Missions

DARPA’s mission is to foster development of technologies that help the U.S. military. Few question that goal, but over the years, Democrats in Congress and the White House have also tended to view the agency as a good vehicle for jump-starting risky commercial technologies, from semiconductors to supercomputers and flat-panel displays. Republicans, on the other hand, decry such “industrial policy,” insisting DARPA stick to its guns, so to speak.

Vinton Cerf

Vinton CerfThere’s a lot in a name, and the ARPA/DARPA label has flip-flopped several times. In 1972, the Nixon administration added “Defense” to Advanced Research Projects Agency. In 1990, the Bush White House sacked DARPA’s director over initiatives that it said were more aimed at boosting U.S. economic competitiveness than at improving military capabilities. In 1993, DARPA swung back to ARPA, in accordance with a Clinton administration strategy paper on the role of technology in the economy. Finally, in 1996, a Republican-led Congress changed the name to DARPA once again.

Those swings have been accompanied by changing views on the relative importance of results-oriented, short-term research vs. basic, long-term research. Vinton Cerf, an Internet pioneer who co-developed TCP/IP while working for DARPA from 1976 to 1982, and now chief Internet evangelist at Google Inc., recently reflected on DARPA’s shifting priorities.

Critics complain that DARPA has cut back funding for basic research, especially long-term research. Where’s the grumbling coming from? The most critical community lodging these opinions is the computer science community, a group strongly funded by DARPA during the 1960s through at least 1990, when the Arpanet was retired. Between 1996 and 2001, DARPA research funding at major university computer science programs dropped by a factor of two, owing in part to the refocusing of DARPA’s efforts toward classified work and engineering-oriented initiatives with more immediate relevance to military applications.

Was that shift a good thing? The notion of a go/no-go test fairly early in a research program forces the effort to have measurable results, often within 18 months. It could be argued that this philosophy may prune nonproductive research quickly, but it may also inhibit major discoveries or developments, owing to the short window for a decision to extend the effort.

The computer science community feels that it has received less support ... than before. Senior DARPA officials see this as a drifting of the computer science community away from relevance to the military, or at least to DARPA.

How would you judge the effectiveness of DARPA programs? There are some very ambitious programs under way at DARPA, and I think it would be fair to argue that DARPA’s success metrics are still very, very high. But there is a certain impatience for results compared to some of the earlier network research. The results [of the earlier work] have been pretty spectacular, taking the Internet as an example.

Copyright © 2007 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon