With a science and technology budget that currently stands at about US$12 billion per year, the US defence complex is the world's largest investor in military research. Much of the money has gone into developing weapons of unprecedented lethality, but a large fraction supports 'dual-use' research, whose products — from the Internet to the Global Positioning System — have enriched society as a whole. And the trove of military data has proved surprisingly useful to scientists studying environmental change (see page 388).

Military efforts are also helping to improve public health. Studies of traumatic brain injuries inflicted by bomb blasts (see page 390) could aid in the diagnosis and treatment of brain injuries in civilians. And the need to keep troops healthy has resulted in advances ranging from a partially effective vaccine against HIV to a mobile-phone-based reporting system for disease cases (see page 395).

Such programmes have been strengthened by JASON, an independent panel of high-level scientists whose advice is often brutally frank (see page 397). But the Pentagon can and should do much more to support dual-use science — by, for example, minimizing the bureaucracy and secrecy that still make it far too difficult for outsiders to gain access to military data.

Defence officials should also insist that their public-health research be meticulously transparent about goals and methods — this is crucial to overcoming mistrust in the developing world. At home, the Pentagon could enhance its credibility among academics by funding discussions on the ethical, legal and social implications of its research — for example, the development of robotic warfare (see page 399).

Most fundamentally, Congress and the Pentagon should continue their strong support for military science. This is not as axiomatic as it was when the United States was in a decades-long, high-stakes technological race with the Soviet Union. Much of today's military research, in the United States and elsewhere, consists of shorter-term problem-solving, such as how to deal with low-tech roadside explosives, or the development of virtual worlds for training troops or aiding their post-injury recovery (see page 406). As the mission becomes more diffuse, high-level support for military science may wane, especially as the Pentagon's overall funding comes under scrutiny (see page 386). Yet cutting and narrowing military research would be short-sighted, especially when the concept of national security is itself expanding, to include not just military strength, but public health, economic vigour, dealing with climate change, and all the other factors that make for a strong society.