The policy of nuclear deterrence is sometimes defended, especially by those who genuinely and honestly pursue it as the best means of preventing nuclear war, by the principle of ‘double effect’. Granted that you can't have an effective deterrent unless you are willing to use the weapons ‘if necessary’ in a crisis, the argument is that the deaths of any innocents who are killed, in the nuclear strike you have to be willing to undertake, will be excusable because they are only the unintended side-effects of the action. (The pompous latinate phrase ‘collateral damage’, much favoured by strategists, strictly means only the same thing – but is best avoided because it easily degenerates into signifying no more than ‘any extra damage’).
Is this ‘double effect’ excuse valid? This question leads us into some deep ethical waters, in which it is very important to learn to swim. In order to avoid raising political hackles, let me make my key point by considering a case which has nothing to do with war or its avoidance. It has been suggested by Sir Michael Quinlan.
Suppose somebody asks me: what are you doing just now this sunny evening? and my reply is: I am driving down the motorway. However, in doing this I unavoidably kill some flies who hit my windscreen. Regrettable as these deaths are, they are no part of my action of driving. They are clearly just unavoidable side-effects of it. So far so good.
However, even this excuse for killing the flies rests upon the appropriateness of the description of what I am doing. And of course, whatever I am doing can be truly described in a multitude of ways. Nevertheless, let us agree that the basic description of what I am doing simply is: ‘I am driving down the motorway’. Of course, if I am asked ‘what do you have to do in order to drive down the motorway?’ I will have to give further information, such as ‘I press my foot on the accelerator, move the steering wheel’ etc. But I do not have to include ‘killing flies on the windscreen’, since killing flies is not part of the action appropriately described as ‘driving down the motorway’. This is what it means to say it is only a side-effect of what I am doing. A side-effect is not part and parcel of the action itself.
So let us agree for the sake of argument that ‘I am driving down the motorway’ is the basic description of what I am doing. However, in addition to this basic action-description there will be other higher-level action-descriptions which must also enter into my answering the question: ‘what are you doing?’ I may, for example, be visiting my Aunt Edna. So when somebody asks me: ‘what are you doing?’ my reply will then be: ‘I am visiting Aunt Edna by driving down the motorway’. This may be called a ‘higher-level’ action-description of what I am doing. And there may be a host of other equally appropriate higher-level action-descriptions of what I am doing. Many of these will be true without being compatible with each other. I may, for instance, be trying to escape from Aunt Edna. And that truth may constitute a higher-level description of what I am doing in driving down the motorway. If so, it is clearly incompatible with the action-description: ‘I am visiting Aunt Edna’.
Now suppose I have been asked by my entomologist friend Fred to collect some flies for his research. I agree to do this by driving down the motorway, thus collecting some flies on my windscreen which I can give him when I get home. In this case, one true appropriate higher-level description of what I am doing will be: ‘I am collecting flies for Fred by driving down the motorway’. This action-description may be compatible with the action-description ‘I am visiting my Aunt Edna by driving down the motorway’ if I have agreed to do both actions together simply by driving down the motorway. Yet, in this case, my killing of the flies on my windscreen will simultaneously be a side effect of the action of visiting Aunt Edna, and a direct effect of the action of acceding to Fred's request. (A direct effect of an action is one for which I am answerable – to the law, to God or to whatever higher authority exists and to which we are all answerable. A side effect is one for which I am not so answerable). So whether something is or is not a side effect will depend on the description given of the action.
Now, the merely physical movements I have to make, e.g. pressing the accelerator etc., are identical in the two cases. But they are nevertheless two quite different actions because, although both are being done together, the two descriptions are distinct from each other. So whether the killing of the flies is a direct effect or a side effect will depend on the choice of the most appropriate higher-level action-description(s) of what I am (basically) doing.
Now, let us further suppose that killing bees has been forbidden by law, on environmental grounds, whereas killing flies is allowed. And let us suppose that driving down the motorway inevitably leads to the deaths on my windscreen of some bees as well as of some flies. And further let us suppose that, despite the ban on killing bees, my friend Fred has still asked me to collect some flies for his research. I decide to do this while visiting my Aunt Edna by driving down the motorway. Now, we have agreed that killing any insects on my windscreen will be only a side-effect of the action of my visiting my Aunt Edna. But killing the flies will still be a direct effect of the action of my doing what Fred has asked me to do. And both of these action-descriptions will be true of what I am (basically) doing. But now, while killing the flies will be excusable (because there is no ban against doing it even as a direct effect of my action), killing the bees is a criminal offence for which I may be guilty. This means that, under the description: ‘I am obeying Fred's request by driving down the motorway’ one effect of what I am doing will be excusable as a licit direct effect, while the other will be condemned as a criminal offence. Indeed, if I am stopped by the police en route, and they examine my car and notice the dead bees on my windscreen, I may be arrested, prosecuted and convicted. (Of course, if Fred knows that I shall unavoidably kill some bees while driving down the motorway, and that I may be prosecuted for an offence, he might modify his request to: ‘Please kill only flies as you drive down the motorway’. But given the initial conditions, including the presence of bees near the motorway, it is not even possible for me to do what Fred has now asked).
What should I do?
-
a) refuse to accede to Fred's request on the grounds that it will inevitably entail my committing a criminal offence?
-
b) try to prove that, while killing the flies will be a licit direct effect of what I am asked to do, killing the bees will be only an unintended side-effect, and is thus permissible under the doctrine of ‘double effect’?
If I attempt to argue b) I may be committed to a theory of intention according to which my ‘intention’ is simply a mental searchlight that can be directed to one part of what I am doing under the action-description ‘doing what Fred has requested’ while somehow preventing its light falling upon another part of it. This is clearly an incoherent account of what an intention is. But it is also clear that if killing flies is part of what I have to do in order to accede to Fred's request, so too is the killing of the bees, since killing them is part and parcel of what I have to do in acceding to it. I can't get out of the dilemma by claiming that somehow killing the bees is not part of what I have to do to accede to Fred's request, since it plainly is part and parcel of that action. So, as long as what I am doing is truly described as ‘doing what Fred has requested by driving down the motorway’ it is clear that if I obey his request I shall unavoidably commit the criminal offence of killing some bees.
Is it possible to justify killing the bees without being committed to an incoherent theory of intention? One way of doing this might be to argue that whether or not something is a side-effect or a direct effect depends on some further factors. For example, if what happens following the doing of an action as described under a certain action-description is sufficiently remote from the basic action itself, it may be excusable because it is unclear how far it is the effect of that action. Thus, if the deaths of the bees on my windscreen are not apparent until I come to wash my car a week later, I may be able legitimately to argue that these deaths were not the result of my driving down the motorway to do what Fred had asked of me. They may have been caused by some other action. But this does not turn the deaths of the bees into a mere side effect of my doing what Fred has requested. It may just be something for which I am not responsible because of some other reason. So the excuse of ‘double effect’ does not apply.
Again, I may try to excuse myself by claiming that at the time I did not know that there was a law against killing bees. But this would still not show that the deaths of the bees were only side-effects of my acceding to Fred's request. It might mitigate the punishment I have to undergo for killing the bees, but it would not turn that killing into a mere side effect. Further, I am answerable for not knowing the law.
Suppose Fred understands my dilemma and simplifies his request into: ‘just drive down the motorway’? I can certainly do this, and therefore I can also try to do it. But in doing this Fred knows (just as I do) that some flies and some bees will inevitably be killed if I accede to his request. True, while killing them is unavoidable, their deaths will be only side effects of this (basic) action of just driving down the motorway. But as we have seen, acceding to Fred's request is doing more than this basic action: it is a ‘higher level’ action, which has to be described differently. That is to say, it is precisely the action of acceding to Fred's request to drive down the motorway. It is not just the action of driving down the motorway, even though just driving down the motorway is what he has requested me to do. And the killing of the insects, whether flies or bees, is not a side effect of this action. i.e. of acceding to Fred's request. On the contrary, it is the direct effect of doing this action, since Fred knows the insects will be killed in my acceding to his request. Killing the insects is part and parcel of the higher-level action of acceding to his request, even though it is not part and parcel of the basic action of just driving down the motorway.
Of course, if Fred is not aware of this, then he is answerable for not knowing what he ought to know. And if he does not know killing the bees is forbidden, he is answerable for not knowing the law. But perhaps he is simply willing to permit them to be killed by me, despite this being forbidden by law, in order to get hold of his flies for his research. In that case, Fred is certainly answerable for the deaths of the bees as well as those of the flies, since he knowingly authorised their deaths in asking me to drive down the motorway with certainty of bees being killed alongside the flies. And I am also answerable for the bees' deaths in so far as I did what he asked me to do. And since we are both answerable, the deaths of the bees cannot be merely unintended side-effects of what I did or of what he asked me to do.
If I were prosecuted for killing bees the jury would have to decide whether what I was doing amounted to a criminal offence or not. My excuse that I was helping Fred with his research would not be likely to wash with them, for I would certainly be held responsible for killing the bees. Even though neither Fred nor I wanted them killed, and it was not the purpose of my driving down the motorway, I would still have criminally killed them because I knew that there were bees around at the time and place where I was driving, and went on driving regardless, and Fred would be guilty also for having asked me to commit this crime.
It has been suggested to me that the example of the flies and bees would be radically altered if it were possible to mount a net over my windscreen. This net would have a mesh which let flies through, but keep bees out. If such a net were practicable, this would indeed mean that killing any bees that unfortunately got through the net (it is assumed that the net cannot be 100% effective) would then be only a side effect of the action of ‘driving down the motorway with the net in place’. For in this case my action would be appropriately described as that of acceding to Fred's request to ‘drive down the motorway to collect flies by using a net which excludes the killing of (most of the) bees that would otherwise unavoidably be killed’. So in this case the appropriate description of the action I undertake would be different. But it is to be noted that in this case (and only in this case) the criterion of proportionality would then come into play. If too many bees got through, despite every effort to make the net effective, I could still be properly answerable for their killing. The double effect excuse would not suffice.
The key point of this analysis is that whether or not something is a side-effect of an action, and is thus something for which we are not answerable, depends on the appropriateness of the description given of the action. This means that, in this example, everything hinges on the wording of the request from Fred. If he asks me to ‘bring him some flies’, I can do this – but only by also committing a criminal offence, since knowingly killing bees in the course of doing this is unavoidable. If Fred asks me to ‘kill only flies’, I cannot even try to accede to this request, let alone carry it out, since it is impossible without the net in place. If his request to me is no more than to ‘drive down the motorway’ although he knows that in acceding to this request I shall unavoidably kill some bees as well as some flies, then he is asking me to commit a criminal act for which both he and I shall be responsible. Only if his request is: ‘drive down the motorway with the net in place’ could I legitimately plead ‘double effect’ in the killing of a (proportionate) number of bees.
Now one might be inclined to think we cannot decide which of several true action-descriptions is the most appropriate for a particular case, since there may be several equally plausible alternatives. Can we not choose the one that suits our case best? The answer surely has to be no. After all it is the business of juries every day of the week to decide, out of competing action-descriptions, which is the most appropriate. Did Bloggs accidentally drop the poison into the tea-cup (as he claims) or did he put it there deliberately (as the prosecution claims)? Which of two action-descriptions of what he did is the most appropriate one to adopt? Was what he did manslaughter or murder? Juries are constantly asked this sort of question, and are expected to come up with reasonable answers. It seems unreasonable therefore to suppose that in the case I have been discussing an equally appropriate answer cannot be arrived at.
The parallel with the nuclear deterrence case should be clear enough. Let us assume that the basic action commanded is: ‘let off a nuclear explosion at or near X’ (this is parallel to ‘drive down the motorway’) and X is a legitimate military target, such as the Soviet naval HQ near Murmansk (which, according to a BBC broadcast on ‘The Nuclear Button’ seems to have been one of the targets kept in mind by the cold-war strategists). Of course, doing this will entail some even more basic actions, like pressing the nuclear button on the submarine. But the basic action-description is bound to include something like: ‘let off a nuclear explosion at X’. (The submarine commander cannot get off his hook by claiming: ‘all I did was press the button as ordered’, as if he did not know that doing this would let the bomb off at X).
Killing the flies is like killing enemy combatants. Under just war criteria their deaths will be an allowable direct effect of letting off the bomb. But killing the bees is like killing innocents who are unavoidably within the range of the explosion and its inevitable aftermath (including of course being poisoned by radioactive fall out, contamination of the land etc.). Now, killing the innocent, except as a side effect of a legitimate action, is forbidden under international law, not to mention being intrinsically unjust and forbidden by God and the church. So the command given must not entail the submarine commander's killing the innocent as part and parcel of what he is ordered to do. Is it possible to frame an order to let off a nuclear bomb at X which does not include this?
The wording of Fred's request, on which everything hung, is like the order which the nuclear power will have to be willing to give to its agents (e.g. its submarine commanders) in ordering a nuclear strike. Whether the unavoidable killing of innocents by this strike is genuinely excusable as a side effect of the letting off the bomb, or whether it is a direct effect for which the nuclear power and its submarine commander are responsible, will depend on the wording of the order. As we have seen, some versions of the order (e.g. ‘kill only enemy combatants in letting off your bomb’) will be impossible to carry out. Others may be possible but will entail committing a crime (e.g ‘let off your bomb at X even though you will certainly kill some innocent people who are within range of the explosion's effects’). As with the bees, it is unreasonable to call these deaths side effects, and arguing this will probably involve an incoherent concept of intention. (A recent example of such incoherence came to light when a ‘Plane Stupid’ protester at Stansted airport in December 2008 claimed on TV that all he was intentionally doing was disrupt the airport, and that disrupting the travel plans of the would-be passengers waiting for their flights was not ‘intended’).
Finally, as far as I can see it is impossible to devise anything equivalent to the net which prevents the killing of bees except as genuine side effects. What would be needed is a ‘net’ which could ensure that the (majority of the) innocents at Murmansk (women, children, the elderly who live on the base or are within lethal range of a bomb dropped on it) are not killed along with the combatants. If such a ‘net’ is practicable, it needs to be described and shown to be effective. I have yet to hear of any such possibility. A strategy for avoiding killing the innocents, by the use of very discriminate targetting by very accurate equipment, is not nearly good enough. What is needed is some method which can ensure that the majority of the innocents are not killed. Planning to avoid the commission of murder is not at all the same thing as preventing its being committed. Without genuine prevention of the deaths of more than a small (i.e.proportionate) number of innocents at Murmansk, these deaths will not be side effects of the dropping of the bomb, and thus excusable as unfortunate consequences for which we are not answerable. They will be direct effects, for which we are answerable.
It could be argued that the deaths of those killed by radio-active poisoning some time later may be excusable, but this will depend on whether it can be clearly shown that they were killed by letting off the bomb or by something else (or both). (The effect of such evidence as that produced by Chernobyl will be relevant in deciding how appropriate it is to call their deaths ‘side effects’ or ‘direct effects’). If it is appropriate to call them direct effects, then they are plainly a matter of criminal responsibility, as well as of justice.
As I have said, everything depends on the wording of the order given to the submarine commander. The question is whether it is possible to frame the order in such a way as to ensure that the deaths of a (proportionately small) number of innocents killed under it will be only side effects because they will not be part and parcel of what the submarine commander does. I do not see how such an order can be framed. If I am wrong, perhaps somebody can suggest such a wording. If a candidate-wording is forthcoming it will then be a matter for a court to determine whether or not carrying it out would be criminal. In justice, it cannot be a matter for the nuclear power to decide on its own. If there is no such court, or it is unable to deal with the situation, this simply means that justice cannot be done, for it is intrinsically unjust for a party to be both judge and jury in its own cause.
NOTE: In this article I owe a good deal to Elizabeth Anscombe's essay Action, Intention and ‘Double Effect’ published in Human Life, Action and Ethics edited by Mary Geach and Luke Gormally, St. Andrews Studies in Philosophy and Public Affairs, 2005. I am also indebted to David Fisher, formerly of the Ministry of Defence, for helpful comments on my text. Sir Michael Quinlan's last book, Thinking About Nuclear Weapons (Oxford University Press, 2009) also discusses in Chapter 5 the ethical problems of nuclear deterrence. In it he claims that there can be only three choices: a) unconditional pacifist renunciation of nuclear possession, b) possession of nuclear weapons for deterrence only, without any permission for use, or c) possession for some limited use of nuclear weapons which could in extreme circumstances be ‘morally tolerable’. Allegedly no fourth option is possible. Option a) is rightly rejected as unjust because it would allow submission to a new Hitler or Stalin, and b) is rejected as strategically incoherent. Unfortunately, as regards c) there are two quite distinct meanings of the systematically unclear phrase ‘morally tolerable’:
-
1. it could imply that deterrence is a positive exercise of the virtue of justice, by making an unjust war between nuclear states unthinkable; or -.
-
2. deterrence is an intrinsically evil strategy, because it entails willingness intentionally to kill the innocent, but is ‘morally tolerable’ as the lesser of two evils.
What Sir Michael Quinlan meant by it remains unclear, but in any case his claim that, having rightly rejected a) and b), there can only remain choice c), cannot stand.
