I am trying to understand the structure of the problem of dirty hands.
A clear example of a dirty hands situation is one in which a political leader authorizes the intentional slaughter of innocent non-combatants to demoralize the enemy and bring about the end of a war which, if it continues, could be reasonably expected to lead to the destruction of the leader's state. The leader must act, but he cannot authorize the actions necessary for the state's survival without authorizing immoral actions. He must act, but he cannot act without dirtying his hands with the blood of innocents. In its sharpest form, the problem arises if we assume that certain actions are absolutely morally wrong, wrong in and of themselves, always and everywhere and regardless of circumstances or (good) consequences. The problem stands out in sharp relief when cast into the mold of an aporetic triad:
A. Moral reasons for action are dominant: they trump every other reason for action such as 'reasons of state.'
B. Some actions are absolutely morally wrong, morally impermissible always and everywhere, regardless of context, circumstances, or consequences.
C. Among absolutely morally wrong actions, there are some that are (non-morally) permissible, and indeed (non-morally) necessary: they must be done in a situation in which refusing to act would lead to worse consequences such as the destruction of one's nation or culture.
It is easy to see that this triad is inconsistent. The limbs cannot all be true. (B) and (C) could both be true if one allowed moral reasons to be trumped by non-moral reasons. But that is precisely what (A), quite plausibly, rules out.
The threesome, then, is logically inconsistent. And yet each limb makes a strong claim on our acceptance. To solve the problem one of the limbs must be rejected. Which one?
(A)-Rejection. One might take the line that in some extreme circumstances non-moral considerations take precedence over moral ones. Imagine a ticking-bomb scenario in which the bomb planter must be tortured in order to find the location of the bomb or bombs. (Suppose a number of dirty nukes have been planted in Manhattan, all scheduled to go off at the same time.)
Imagine a perfectly gruesome form of torture in which the wife and children of Ali the jihadi have their fingers and limbs sawn off in the presence of the jihadi, and then the same is done to him until he talks. Would the torture not be justified? Not morally justified of course, but justified non-morally to save Manhattan and its millions of residents and to avert the ensuing disaster for the rest of the country? One type of hard liner will say, yes, of course, even while insisting that torture of the sort envisaged is morally wrong, and indeed absolutely morally wrong. I am in some moods such a hard liner.
But am I not then falling into contradiction? No. I am not maintaining that in every case it is morally wrong to torture, but in this case it is not. That would be a contradiction. I am maintaining that it is always morally impermissible to torture but that in some circumstances moral considerations are trumped by -- what shall I call them? -- survival considerations. These are external to the moral point of view. So while morality is absolute in its own domain, its domain does not coincide with the domain of human action in general. The torture of the jihadi and his wife and children are justified, not morally, but by non-moral reasons.
(B)-Rejection. A second solution to the triad involves rejecting deontology and embracing consequentialism. Consider the following act-type: torturing a person to extract information from him. A deontologist such as Kant would maintain that the tokening of such an act-type is morally wrong just in virtue of the act-type's being the act-type it is. It would then follow for Kant that every such tokening is morally wrong. A consequentialist would say that it all depends on the outcome. Torturing our jihadi above leads or can be reasonably expected to lead to the greatest good of the greatest number in the specific circumstances in question, and those on-balance good consequences morally justify the act of torture. So, contra Kant, one and the same act-type can be morally acceptable/unacceptable depending on circumstances and consequences. Torturing Ali the jihadi is morally justified, but torturing Sammy the jeweler to get him to open his safe is not.
On this second solution to the triad, we accept (A), we accept that moral considerations reign supreme over the entire sphere of human action and cannot be trumped by any non-moral considerations. But we adopt a consequentialist moral doctrine that allows the moral justification of torture and the targeting of non-combatants in certain circumstances.
Now we must ask: Do the consequentialist torturers of the jihadi and their consequentialist superiors who order the torture have dirty hands? Suppose the hands of the torturers are literally bloody. Are they dirty? I am tempted to say No. They haven't done anything wrong; they have the done the right thing, and let us assume, at great psychological and emotional cost to themselves. Imagine snapping off the digits of a fellow human being with bolt cutters or high-torque pruning shears. Could you do that to a child in the presence of his father and do it efficiently and with equanimity? Could you do your job, your duty, despite your contrary inclination? (I am turning Kant's phraseology against him here.) But you must do it because the orders you have been given are morally correct by the consequentialist theory.
Do the torturers have dirty hands? It depends on what exactly it is to have dirty hands which, of course, is part of the problem of dirty hands. On a narrow understanding, a dirty hands situation is one in which the agent acts, and must act, while both accepting all three limbs of our inconsistent triad and appreciating that they are inconsistent. A dirty hands situation in the narrow and strict sense is an aporetic bind of the practical and not merely theoretical sort. You must act and you must act immorally in violation of absolute moral prohibitions, and you cannot justify your actions by any non-moral considerations that trump moral ones. That's one hell of a bind to be in! Some will be tempted to say that there cannot ever occur such a bind. But if so, then there cannot ever occur a dirty hands situation. So maybe talk of 'dirty hands' is incoherent.
If this is what it is to be in a dirty hands situation, then a consequentialist cannot be in a dirty hands situation. He is not in an aporetic bind since he rejects (B). And the same goes for those who reject (A) or (C).
(C)-Rejection. A third solution to the problem involves holding that there is no necessity to act: one can abstain from acting. A political leader faced with a terrible choice can simply abdicate, or simply refuse to choose. He does not order the torture of the jihadi and and hence does not act to save Manhattan; but by not acting he willy-nilly aids and abets the terrorist.
For now I will conclude that if we leave God and the soul out of it, if we think in purely immanent or secular terms, then we are in a genuine aporetic bind, and the problem of dirty hands, narrowly construed, is a genuine one, but also an insoluble one. For rejecting any of the limbs will get us into grave trouble. That needs to be argued, of course. One entry leads to another, and another . . . .
Philosophia longa, vita brevis.
C. A. J. Coady, The Problem of Dirty Hands
Michael Walzer, Political Action: The Problem of Dirty Hands