Matthew Yglesias (who needs to get his comments working again soon) talks about clean hands.

In fact, the clean hands principle has a long and distinguished history in Western moral thought, being embraced by such seminal figures as Jesus Christ and Immanuel Kant. Indeed, many a classic ethical thought experiment is designed to provoke precisely the clean hands intuiton Drezner is so dismissive of here. Say you’re a doctor who has five patients who are going to die if they don’t receive organ transplants very quickly. Say there’s a nurse in your hospital such that if you killed the nurse and parceled out her organs you could save all five patients — net gain in human lives: four. Do you kill the nurse? Just about everyone says no.

You guys know how much I love analogies, so let's try another one, a modified version of one I'm mentioned here before, and compare it to Matthew's.

I've consistently contended that the situation in countries like Iraq, those ruled by a brutal despot who oppresses, tortures, and regularly kills civilians, is more akin to a hostage situation.

Let's say a madman had five hostages in a house. He is in clear violation of the law. Illegal weaponry has been found in his house before, and there is a strong suspicion that he still owns illegal weaponry. But he has enough stockpiles of food and water that he never needs to leave the house.

The hostages live in a state of fear. They are reasonably well-fed, but occasionally the madman tortures one of them for his amusement, and to keep the others in line.

Now hypothetically, let's say that the police have surrounded the house for days. There are snipers on every roof. Let's also say that if any of the snipers takes a shot, there is an extremely high likelihood of killing the madman, but also an equally high likelihood of killing one of the hostages.

If you're the police commander in charge, do you order the sniper to shoot?

Now before anybody starts nitpicking the analogy, please try to focus on the principle. It doesn't have to be a sniper. Say we have a grenade that will kill the madman but most probably kill one of the innocents. Or let's say we send in a SWAT team, and the madman's got a knife to one of the hostages throat, so we kill him, but not before he kills a hostage.

The idea is, do we severely risk the life of one, or perhaps more of the hostages, in order to liberate them from the hostage situation? Or do we pack up our equipment and go home?

In this case, there's a likelihood that you will probably kill two people, one innocent, one not. Plus, you are putting the other innocents, plus perhaps your fellow police officers, in harm's way. Still, it's probably a net gain of four lives, for the death of a criminal and an innocent.

I would say this example is qualitatively different from the nurse example. In Matt's hypothetical, he doesn't place any probabilities on the fates of any of the actors. The four patients live, at the expense of the nurse, who definitely dies. There's no talk about the relative risks of the procedures (these days transplants are very risky and uncertain).

But the most important element is intent. In the nurse situation, you are intentionally killing the nurse, for use of her organs. In the hostage situation, you are intentionally killing the bad guy, but unintentionally killing an innocent.

To suggest that these actions are morally equivalent is frankly grotesque.

But let's go back to Iraq. They have a population of roughly 24,000,000 people. As of this writing, even the most inflated estimates of civilian casualties as a result of this war are less than 1,500, and I believe still under 100 coalition military casualties.

According to a Human Rights Watch report, 50,000 to 100,000 Kurds were killed by Saddam's regime during the Anfal campaign, many thousands of Iraqis have either disappeared or been executed, millions of refugees have been displaced, and countless scores of Iraqis have been raped, tortured, and otherwise brutalized by the regime over the past three decades.

Was the price worth paying?