The US Department of Defense's Human Terrain System, an attempt to have social sciences inform military decision-making, is failing on every level.

In theory, it is a good idea. The Human Terrain System aims to embed anthropologists and other social scientists in military units in Iraq and Afghanistan to help improve understanding of local cultures and thus relieve tensions between civilians and soldiers. In practice, however, it has been a disaster. Questions have been raised about how well the programme vets its employees (see Nature 455, 583–585; 2008). Some scientists who have joined the system have complained about inadequate training. And qualified researchers have been dismissed for seemingly trivial reasons, even though much more questionable people seem to breeze onto the payroll.

A case in point is Issan Hamama. Under investigation by the Federal Bureau of Investigation since 2003 as a possible former spy for Saddam Hussein, Hamama nonetheless managed to secure a job as a translator for the Human Terrain System. Late last month, he was arrested in Maine and indicted for conspiracy; he is currently free on bail.

Another contractor, bodyguard Don Ayala, is also out on bail after being indicted for a murder committed in Afghanistan last month. According to a military affidavit, Ayala shot Abdul Salam at close range in the head after Salam doused his colleague, social scientist Paula Lloyd, with petrol and set her on fire. Lloyd had approached Salam on the street — he was carrying a fuel jar — to ask him about the price of petrol.

Lloyd returned to the United States to recover from her burns; some of her colleagues have not been so lucky. Social scientist Michael Bhatia was killed in Afghanistan in May; Nicole Suveges, a PhD student from Johns Hopkins University in Baltimore, Maryland, died in Iraq the following month.

Their names and sacrifices should be remembered. But the programme that employed them should not — except, perhaps, as an example of yet another good idea gone wrong on the war fields of Iraq and Afghanistan.

The immediate problems with the Human Terrain System can be traced to BAE Systems, the military contractor based in Rockville, Maryland, that screens potential employees, then trains those it hires. It has failed in every one of those functions, and army management has failed in its oversight of BAE.

But the larger question is whether the Human Terrain System is viable at all. Nature is not opposed in principle to academics working with the military; we have said before that social science can and should inform military policy (see Nature 454, 138; 2008). We continue to believe that the insights of science have much to offer strategies in a war zone — not least through training combat troops to understand the local cultures within which they operate.

But as currently constituted, the Human Terrain System is not the way to do this. Unless the programme can be reborn in a format less plagued by deadly mistakes, it needs to be closed down.