The New York Times reports on the difficulties the Obama administration is encountering in gauging progress in Afghanistan. According to the report:

Senior administration officials said that the president’s national security adviser, Gen. James L. Jones, approved a classified policy document on July 17 setting out nine broad objectives for metrics to guide the administration’s policy on Afghanistan and Pakistan. Another month or two is still needed to flesh out the details, according to officials engaged in the work. …

For instance, some of the measures now being devised by the Obama administration track the size, strength and self-reliance of the Afghan National Army, which the United States has been struggling to train for seven years. They include the number of operations in which Afghan soldiers are in the lead, or the number of Afghan soldiers who have received basic instruction.

The article goes on to note the difficulties the Bush administration had in devising metrics for Iraq. One such metric touted by the White House in 2005-06 was the “rat rate”—the number of good tips from Iraqis about insurgent activities. Only one problem: as the rat rate went up, so did violence. The article doesn’t mention, but could have, the 12 benchmarks Congress mandated in 2007—most of which proved utterly irrelevant to the success of the surge, which depended on better counterinsurgency tactics and the willingness of former insurgents to change sides.

Then there were numerous attempts earlier in the war by the Bush administration to measure progress based on inflated figures of Iraqi security personnel who were supposedly ready to fight insurgents—a particular obsession of Don Rumsfeld. This really shouldn’t be so hard. The problem with most of the faux metrics employed in Iraq—and now migrating to Afghanistan—is that they measure inputs, not outputs: the size of the local security forces, the number of operations they undertake, the willingness of civilians to phone in tips, the ability of local lawmakers to reach agreement on legislation—all of these are inputs. The ultimate output has to be a reduction of violence, and those inputs are only insignificant insofar as they contribute (or not) to that goal.

The only way the war will be judged a success is if the number of civilians killed goes down. By a stroke of good fortune, that figure happens to be relatively easy to measure objectively, whereas most of the other metrics are subject to interpretation and debate. (Just imagine how hard it would be to figure out if corruption in Afghanistan is up, down, or unchanged. Hard? More like impossible.) In Iraq the murder rate was rapidly increasing before the surge and fell just as rapidly thereafter. That’s how the world knew we were winning. Today in Afghanistan, the number of civilians killed is on the upswing. Only by reducing that figure will we reverse the downward slide of the war effort. Everything else is just noise.