The argument for not doing it at all has to rest on proving either that it's morally repugnant, or that there is no way to have an effective waterboarding policy, or that the costs exceed the benefits. Unfortunately, I seem to see too many opponents of current policy simply arguing that it never produces usable intelligence, so everyone else is a big fat moral cretin.
It feels slightly cheap, like supply-siders claiming there are absolutely no tradeoffs whatsoever to cutting taxes. The people who support waterboarding, and the Bush administration, perceive themselves to be wrestling with a genuine moral dilemma: how do you weigh the suffering of suspected terrorists against the suffering of innocent victims of terror? That's not an easy question, and if it were you who were trying to save, say, your child from a terrorist, your attitude about the utter impermissibility of torture might undergo a sea change.
That doesn't mean I agree we should waterboard--people will do lots of things for their children that should not be state policy. Only that some of the people I've heard saying they have to resort to these shaky arguments because their opponents are moral no-shows without a shred of decency seem to me to be awarding themselves vast moral credit for parroting, like a third-grader, the trivial truism that torture is bad. They find it easy to call their opponents immoral because they're ignoring a hard moral question. One that is, of course, easy to set aside if you seize on every piece of evidence suggesting that physical pressure is ineffective, and block out the people saying it's worked.
I'm against waterboarding. It's wrong. The state should not do this, for the same reason the state should not pull the wings off of flies. I've nearly suffocated from an asthma attack, and anyone who thinks it's not so bad is invited to come over to my house for some fun with a 3/4 full bathtub.
But I'm against it with the knowledge that this might, at least in some circumstance, result in some innocent person dying from lack of information. My defense is, first, that it is not okay to do purely morally repugnant things to save people--I wouldn't murder an innocent baby to save 1,000 people, either. And second, that a state which allows itself to do those things will do more harm to more people than any terrorist conceivably could.
As you can see, I do think there are ways to mix consequentialist arguments about the use of torture with the deontological argument about its morality. But I think it's easier to prove empirically that torture gets out of hand--as waterboarding clearly did, even though there's no evidence that the people administering it were specially bad, rogue agents--than that it never produces useful intelligence.
I haven't seen any particularly conclusive evidence either for or against the proposition that torture never generates useful intelligence (though as a practical matter, I can't see how it wouldn't work in some narrow situations where the information you're seeking is easily confirmable.) But also, as I've pointed out, we may well be very close to being able to make torture effective in many situations. If you can tell whether someone is lying by seeing which parts of their brain light up on a scan, then torture is an extraordinarily effective way to get information. Tell me what I want to know, or I will smash your fingers with this hammer. If you lie to me, I will know it, and smash your fingers harder. Some people will hold out, because they can resist pain, just as some people in the medieval era died rather than give their interrogators the satisfaction. But most people will break. If giving their interrogators good, useful information is the only way to make the pain stop, they will make. it. stop.
I don't want to live in a state that does that, for both moral and practical reasons. I think the greatest good for the greatest number lies in having a state that is forbidden to waterboard suspected (or known) terrorists. But I would feel just the same if those waterboardings were producing usable intelligence. And I don't know whether or not they are. So I'd like to keep my arguments to areas where I can make a really strong case.