The trouble with salami-slicing is that it doesn't stop just because you do. A judicious trade-off of competing considerations is vulnerable to salami-slicing from both directions. You can calibrate the viciousness of the torture as finely as you like to make sure that it matches the urgency of the situation. But you can't calibrate the torture candidate strapped down before you. Once you're in the torture business, what justification is there for banning (as Krauthammer would) the torture of official prisoners of war, no matter how many innocent lives this might cost? If you are willing to torture a "high level" terrorist in order to save innocent lives, why should you spare a low-level terrorist at the same awful cost? What about a minor accomplice?
Or what about someone wholly innocent? It's hard to imagine a situation where someone who refuses to supply life-saving information could be considered "innocent." But it's not impossible. (Suppose the terrorists have his wife. …) In this cold, hard world, allegedly facing a challenge greater than any the civilized world has faced before, would you torture an innocent individual for five minutes in order to spare a million innocents from death? These would be wartime deaths, many of them more painful and grotesque than the laboratory torture you are sparing one lone individual. If you say yes, go ahead and torture an innocent person, you have pretty much abandoned the various exquisite moral distinctions that eased your previous abandonment of an absolute ban on torture. But if you say no, my own moral hygiene, or my country's, forbids the torture of an innocent individual, even if the indirect but predictable consequence is a million human deaths, you are more or less back in the camp of the anti-torture absolutists whose simple-minded moral vanity you find so irritating.
So Krauthammer's second argument—that once you abandon an absolute rule against torture, there is no obvious moral stopping point—"proves too much" (in another lovely law-school phrase). It can be used to discredit any nonabsolutist torture policy, including Krauthammer's own.
Torture is like almost every other issue: It involves trade-offs between the rights of individuals and the needs of society. In his own proposed rules, Krauthammer makes some strange trade-offs. How many lives would he give up in order to relieve the military of the onus of torture? And where will he find morally pre-damaged patriots better suited to the task? Do CIA agents deserve to be told that torturing people is a "monstrous evil" that is too "inhumane" for uniformed soldiers, but just perfect for them?
It is not fatal to Krauthammer's or any other person's particular set of torture rules that they draw lines more exact than evidence or reason can justify. Drawing bright lines in foggy situations is what the law does. But good rules need to be defensible against salami-slicing in a more general way. The strength of an absolute ban on torture—or an absolute rule of any sort—is its relative immunity from salami-slicing, both in theory and in practice. It is hard to explain why you would torture a teenager abducted into a terrorist gang if this would save a dozen lives, but would not torture a uniformed military officer in order to save a thousand. It is not hard to explain why you would not torture anybody at all. The argument may be wrong, but at least it is clear. The policy—just don't do it—is hard to misunderstand, making it easier to teach and enforce. And the principle can be consciously abandoned but it can't easily erode.
But what about Krauthammer's conundrum? Will you eschew torture even when a few minutes of it, applied to a very bad person, would save a million lives? One answer is that the law wouldn't really be enforced in such an extreme situation. McCain himself has hinted at this, as Krauthammer points out, and Andrew Sullivan fleshes out the point in a reply to Krauthammer published in the New Republic. This may well be true as a prediction, and tempting as a moral argument, but ultimately not good enough. Surely every law should at least aspire to be enforced. Or—an even more modest standard—a law should not depend on unenforceability for its very justification. Furthermore, a law expresses a social norm even apart from its enforcement. If the hypothetical situation ever arises, something will happen. What do we want that something to be?
There is yet another law-school bromide: "Hard cases make bad law." It means that divining a general policy from statistical oddballs is a mistake. Better to have a policy that works generally and just live with a troublesome result in the oddball case. And we do this in many situations. For example, criminals go free every day because of trial rules and civil liberties designed to protect the innocent. We live with it.
Of course a million deaths is hard to shrug off as a price worth paying for the principle that we don't torture people. But college dorm what-ifs like this one share a flaw: They posit certainty (about what you know and what will happen if you do this or that). And uncertainty is not only much more common in real life: It is the generally unspoken assumption behind civil liberties, rules of criminal procedure, and much else that conservatives find sentimental and irritating.
Sure, if we could know the present and predict the future with certainty, we could torture only people who deserve it. Not just that: We could go door-to-door killing people before they kill others. We could lock up innocent people who would otherwise be involved in fatal traffic accidents. Civil libertarians like to believe that criminals get their Miranda warnings and dissidents enjoy freedom of speech because human rights are universal. But if we knew for sure that a newspaper column by Charles Krauthammer would lead—even by a chain of events he never intended and bore no responsibility for—to World War II, wouldn't we be nuts not to censor it? Universal human rights would make no sense in a world where everything was known and certain.
This is not to say that Krauthammer's killer hypothetical could never happen. It is to say that morality does not require us to build a general policy on torture around a situation that is not merely unlikely in real life, but different in kind from the situations we are likely to face in real life. What we would do or should do if this situation actually arose is an interesting question for bull sessions in the dorm, but not a pressing issue for the nation.
Every day American forces in Iraq and elsewhere probably inflict more pain on guilty and innocent people than officially designated American torturers would do in a year, even if Bush and company were free of any legal restriction. That pain is not necessarily unjustified (although I believe it is). But it makes the whole debate about officially designated "torture" artificial and symbolic, not to say deeply hypocritical. And yet supporters of the administration, the war, and the practice of torture have not leaped to embrace this argument, for some reason.