With the end of the Cold War and the drying up of defence funding, military departments worldwide have seized upon the doctrine of "dual use". The idea that many of the technologies critical to the military are also critical to the economy, allows the .mils to act as a choreographer-cum- sugar daddy for industry while obtaining toys in the process. It is a beguiling idea because everyone seems to win. It is also thoroughly flawed.
A brief glance at the history of supercomputers demon-strates why dual- use research can't work. The development of high-performance computing has been driven by the military since the '50s. Every aspect of supercomputers - from how they are designed to how their performance is measured - has been shaped by the military's quest to design more powerful nuclear weapons and crack more powerful codes. A co-dependent relationship developed between the US military and supercomputing vendors like Cray and Thinking Machines. Vendors relied on government subsidies (it was easier than fighting for commercial sales) while the US Department of Defense relied on the vendors' ability to offer custom solutions (it was easier than customising general solutions). This resulted in increasingly expensive and exotic computers that required an army of technicians and scientists to use.
When DOD funding began to dry up in the late '80s, supercomputing vendors started to feel like a pack of poodles thrown into the wild. The outcome was a spate of bankruptcies that included the venerable Cray. While the vendors were lapping up government subsidies, microcomputer manufacturers were engaged in a Darwinian struggle for market share. The result has been a shakedown within the computer industry: the microprocessors used in personal computers and videogames have improved so much faster than supercomputers; a £500 chip can now beat a million-dollar Cray. The lesson of the supercomputer debacle is that nobody, not even the military, can beat the economies of scale. By encouraging products tuned for the specialised and relatively small defence market, the military's investment did more harm than good.
Special needs and bureaucracies are intrinsic to the military, and no amount of goodwill towards industry will make them go away. The military should stick to what it was designed for: warfare. If we try to pretend that military interests coincide with commercial interests, we will end up damaging both.
Steve G Steinberg (firstname.lastname@example.org) is an editor at US Wired.