As its weapons have grown more complex and more dependent on operating software, the U.S. military has cut spending on software research and development, counting instead on U.S. companies to develop the needed computer code.
It's a potentially dangerous trend, said a former high-level military software researcher.
Despite the proliferation of dazzling, high-tech consumer goods, U.S. technology companies are not producing the kind of extremely complex software that's needed to operate weapons such as the Joint Strike Fighter or the interwoven system of systems being designed for net-centric warfare, said Douglas Schmidt.
The cuts have occurred over the past decade, said Schmidt, a former deputy director for information technology at the Defense Advanced Research Projects Agency. DARPA's budget for research into embedded systems, for example, shrank from $300 million in 2000 to zero today, he said.
Schmidt, who is now a computer science and engineering professor at Vanderbilt University, spoke to an audience of congressional staffers two weeks before the Pentagon presents its 2011 budget request to Congress. He said he does not know how much money the military will seek for software research and development in 2011, but expectations are not high. Spending on that kind of research has declined steadily since 2005.
"There is a misconception that commercial IT is an oasis that the Defense Department can come and drink from," Schmidt said. Not so, he said. "Civilian software can't be dropped easily into DoD systems," he said.
Typically, commercial software is developed to fit specific products. Commercial firms do little of the research that leads to the long-term technology improvement that the Defense Department needs for future generations of weapons, Schmidt said.
Dwindling capability to develop the kind of software the military requires has led to years of delay in weapons such as the F-22 stealth fighter and the JSF – and fees of $1 million a day paid to companies to fix the software mistakes they made, he said.
The situation is likely to get worse. The military is moving toward "ultra-large-scale systems" such as the Army's Future Combat Systems, which attempt to integrate many different types of high-tech hardware with "inherently conflicting, unknowable and diverse requirements," Schmidt said.
Compared to today's complex weapons, such as the JSF with 4.5 million lines of software, ultra-large-scale systems are much more complicated. Quoting former Assistant Army Secretary Claude Bolton, Schmidt asked, "How can future systems, which are likely to require a billion lines of code, be built reliably if we can't even get today's systems right?" Schmidt offers two possible answers. One is an Internet portal called SPRUCE, which he said attempts to introduce software scientists with the appropriate military technology needs.
Another, more distant solution, is to work on developing automated systems that can write advanced code.
Video 
Help
