Alexander, an Air Force Reserve officer who conducted or supervised some 1,300 interrogations using traditional techniques, told me, "I was surprised that the people we thought would be the hardest were the easiest to interrogate."
He cites the case of a Muslim scholar, a high-level al-Qaida operative who was "as hard-core as you could find." Using a non-coercive approach, "in six hours I convinced him to cooperate."
How can it be that a violent, determined enemy of the United States could be persuaded to talk without extreme measures? "He's human, he's not a robot," says Alexander. By establishing a personal connection, interrogators can induce prisoners to open up. But if detainees are abused, he found, "they quit talking."
Sometimes, no doubt, torture can loosen a tongue. But once a high-value operative is brutalized, there is no way to know what he might have divulged under more patient, humane interrogation. If he spills secrets after being waterboarded, it "proves" that torture works. If he withholds information, it "proves" that nothing else would have sufficed.
But what if torture does sometimes work? Mere effectiveness is not enough to justify it. Yoo was once asked about the legality of "crushing the testicles of a person's child," and he did not rule it out.
Why should he? If torturing a terrorist failed, wouldn't we be justified in torturing his wife or his children to get the truth?
If waterboarding is OK, why not crushing testicles? Why not pulling out fingernails? Why not the most agonizing methods an evil mind could devise? The advocates of waterboarding are much more eager to declare what is allowed than what is forbidden -- if anything.
In the end, they don't really care about imposing limits, and they don't really care if torture is effective or not. Torture, in the minds of its apologists, is not a means to a good result. It's a good result all by itself.