The Problem with Prevention 1
The Problem with Prevention 1
Abstract and Keywords
This chapter argues that there are profound moral problems with the doctrine of preventive war. The consequentialist approaches to preventive war are first examined. The chapter then examines whether preventive action can be viewed as a legitimate component of the right of self-defence, and argues that it cannot. The right of self-defence has historically been grounded in a number of different theoretical justifications, including those that invoke a conception of psychological necessity and those that invoke human rights. On neither of these theories, however, can preventive self-defence be justified.
In September 2002, one year after the terrorist attacks on Washington and New York, the Bush administration released its ‘National Security Strategy of the United States of America’, the blueprint for the development of US military and security policy. This document contained a radical innovation in the grounds for waging war. It proclaimed a readiness to fight preventive wars against ‘emerging threats before they are fully formed.’2 Just over a year later, the United States, together with Britain and a number of other allies, put the policy of preventive war to dramatic effect by invading, defeating, and occupying Iraq in a matter of weeks.3 The doctrine of preventive war and its implementation in Iraq caused consternation among many long‐standing American allies, particularly in Europe. For its critics, the policy is a dangerous and misguided challenge to the legal and moral international order. For its advocates, the policy is a morally and strategically justified response to new forms of security threat.
In this chapter, I argue that there are profound moral problems with the doctrine of preventive war. The argument is structured in two parts. In the first part, I examine consequentialist approaches to preventive war. The case both for and against preventive war is often made in consequentialist terms. This seems to me a mistake. Consequentialism suffers from debilitating epistemological problems that renders it effectively useless as a moral theory of war. Consequentialism would require historical judgments about the long term overall effects of war that are intrinsically indeterminate and unknowable. Moreover rule‐consequentialist assessment of the laws of war suffers from what I call the impasse problem: the persistent presence of equally plausible rule‐consequentialist arguments for countervailing conclusions. Much of this part of the discussion uses (p.144) as its foil a recent and significant contribution to the debate on prevention by David Luban, which provides a vigorous defense of the consequentialist approach to war.4
In the second part of the chapter, I examine whether preventive action can be viewed as a legitimate component of the right of self‐defense. I argue that it cannot. The right of self‐defense has historically been grounded in a number of different theoretical justifications including those that invoke a conception of psychological necessity and those that invoke human rights. On neither of these theories, however, can preventive self‐defense be justified. Indeed, on what I believe to be the best overall account of self‐defense—the human rights‐based theory of self‐defense—the doctrine of prevention leads quite literally to a paradox: if a doctrine of prevention is morally justified, this can only be because fighting preventive wars are unjustified and vice versa. There are therefore strong ethical reasons for rejecting the doctrine of preventive war.
6.1. PREVENTION AND PREEMPTION
First, however, I want to address an important issue of terminology. In the National Security Strategy and in the speeches of the Bush and Blair administrations, the policy that is to be our concern is referred to as ‘preemptive’ rather ‘preventive’ war. But this is a rhetorical sleight of hand. According to well‐established legal usage, preemption consists in a first strike against an enemy who has not yet attacked but whose attack is clearly imminent. It involves ‘anticipating’ an aggressor who is literally poised to attack. Prevention, on the other hand, involves a first strike against a potential future aggressor who does not yet pose an imminent threat. In Michael Walzer's words it is ‘...an attack that responds to a distant danger, a matter of foresight and free choice’.5
It is sometimes said that a right of preemptive self‐defense is clearly and uncontroversialy recognized under international law.6 This is simply not true. Many scholars believe that even a limited right of preemption against imminent attack is ruled out under international law. The UN Charter is very clear that the right of self‐defense can only be invoked ‘if an armed attacks occurs’ (UN Charter, Article 51). Under traditional means of treaty interpretation this has been taken to unequivocally rule out a right to preemption.7 On the other hand there does (p.145) exist older customary law, in the form of the famous Caroline doctrine (of which more later), which encompasses a limited right to engage in preemptive strikes against an imminent attack. Those who support a right of preemptive self‐defense must argue that Article 51 does not replace, but merely supplements, this older customary law right. But this customary law argument is undermined by the fact that states have tended to avoid invoking a right of preemptive defense even in circumstances that might justify it (e.g. the right was not invoked by Israel to justify its first strike against Egypt in the Six Day War, which is often regarded as the locus classicus of justified preemptive action by the doctrine's supporters).8 Thus, while it is possible to argue that there exists a right of preemptive self‐defense in international law, it is by no means a settled question.
On the other hand, there is no uncertainty as to the legality of undertaking unilateral military action against ‘emerging threats before they are fully formed’. Such action against a nonimminent threat is properly referred to as ‘prevention’ rather than ‘preemption’, and it clearly contravenes international law as it currently stands.9 The ‘Bush doctrine’ articulated in the National Security Strategy, while not historically unprecedented, does represent a genuine innovation in the grounds for war, one which would transform the moral and legal framework on the international use of force.
6.2. THE DISUTILITY OF CONSEQUENTIALISM
It is not difficult to construct a consequentialist argument for preventive war, particularly in light of the much‐discussed potential nexus between international terrorism and WMD. If undertaking a particular preventive war is an effective way of averting a significant terrorist attack or act of future aggression from a hostile state, and if the war does not itself cause more harm than it prevents, and if there is no less costly way of achieving the same good, and if the opportunity costs of the expended blood and treasure do not outweigh the projected goods, then the war would be prima facie justified on consequentialist grounds.
These are all big ifs, and much of the plausibility of the consequentialist argument rests on its ability to fulfill the evidentiary burdens they impose. I argue that the consequentialist approach to war quite generally suffers from two traditional vulnerabilities of consequentialist theory. The first is its tendency to generate counterintuitive ethical conclusions and assessments. The second, and far more serious, is a severe epistemological weakness resulting from the radical indeterminacy of the historical judgments required to yield a meaningful consequentialist conclusion about the justification of war. This criticism is not new, it was made by (p.146) Walzer in Just and Unjust Wars,10 but I believe that the full force of the objection has not been fully appreciated by consequentialist writers on war.
Of course, we may expect the case of preventive war to generate special difficulties for any ethical theory. Let us begin, therefore, with what should be an easy case: the Allies' war against Nazi Germany, which many feel is the clearest modern example of a justified war. The first problem is that it is not clear that a consequentialist analysis supports this judgment. The Nazi regime was brutal and aggressive: it systematically murdered approximately six million innocent people, and subjected many millions more to a regime of extraordinary moral repugnance. But the costs of the war against Germany were in many ways more horrific still. Those who lost their lives to the war outnumber the victims of German murder and genocide, by approximately ten to one. This does not yet include the suffering of the approximately forty million persons displaced or made homeless by the war and the incalculable destruction to the world's cultural and material wealth. This was a horrific price to pay for stopping Nazi Germany. Even allowing for the expanded genocide that would certainly have taken place if the war had not been fought, it is not obvious that on strict consequentialist grounds it was a price worth paying.11
What is most striking about the assessment of war from a consequentialist perspective, however, is not that particular wars fail or pass the test in a counterintuitive way, but rather the near impossibility of accurately determining the long‐term balance of consequences of engaging in war. To return to World War II, there is a very real sense in which we still do not know the long‐term balance of consequences of this war six decades after its end. This is the case for a number of reasons. First, we do not, even now, know the actual casualty figures of the war with any degree of precision. Estimates of the Soviet War casualties alone have ranged from seven to forty million persons.12 That is a margin of error of 470 per cent—an almost incomprehensible thirty three million human lives!
Chairman Zhou En‐lai famously responded that ‘its too early to tell’ when asked about the impact of the French Revolution. In a similar way, it may be literally true that it is too early to tell what the ultimate balance sheet of World War II's costs and benefits is, because the war continues to have a profound and ongoing impact on the world today. This is true both on the macro and on the micro level. The war and its aftermath continue to shape patterns of political alliance and economic exchange within Europe and around the world (e.g. it continues to play (p.147) a significant role in the development of the European Union). On a micro level the war continues to affect the lives and well‐being of countless individuals in both subtle and profound ways. To take but one example, the trauma of experiencing the war has left legacies of mental illness and emotional abuse that continue to ripple down through the generations of many families.
Secondly and most damagingly, because consequentialist assessment requires a comparative judgment between the consequences of fighting and not fighting a given war, it always depends upon questions of counterfactual history that are intrinsically unknowable. For example, if the Allies had not chosen to fight Nazi Germany we can assume that it would have continued to expand its genocidal and aggressive policies—but by how much? Would the Nazi state have collapsed under the weight of its own inefficiency and inequity as the Soviet Union did 45 years later? Would National Socialism have survived the death or assassination of Hitler? There is simply no reliable way to answer these and countless other relevant counterfactual questions like them.
These profound epistemological difficulties arise even when conducting a retrospective analysis, with the benefit of hindsight and decades of painstaking academic research. But if consequentialism is to function as a guide to action for policymakers rather than simply as a form of backward‐looking moral assessment, then it is necessary to undertake an immeasurably more difficult task still. That is an accurate prospective analysis of the likely long‐term consequences of engaging in war before the conflict has even begun. To return to World War II, it is almost impossible to believe that any of the leaders in 1939 could have reliably foreseen the long‐term consequences of the war that was about to begin. As Carl von Clausewitz said of war: ‘no other human activity is so continuously or universally bound up with chance’.13
Moreover, World War II commenced with clear and unambiguous acts of aggression on the part of Germany. But the case of preventive war generates further difficulties still for consequentialism. This is because the feared act of aggression lies sometime in the future and indeed may never eventuate. In such a case, the problems of indeterminacy and the dependence on unknowable counterfactual judgments, present in all war, are magnified many times over.
Now, I certainly do not want to claim that we can never make an accurate and well‐founded determination of whether the balance of consequences favors war or not. In fact it is often very easy to make this determination in the negative. Because initiating a war always has very significant moral costs, it is obvious that going to war to prevent a minor infringement or slight injury will not lead to better over all consequences.14 This points to an important asymmetry in our ability to assess the consequences of war. In general, it is easier to fulfill the evidentiary burdens of the negative judgment that engaging in war will not lead to better (p.148) overall consequences than it is to fulfill the burdens of the positive judgment that fighting will lead to better consequences.
There is a structural factor inherent to war that explains why a positive assessment of the likely consequences of war will always be more problematic and less reliable than the purely negative assessment. As Clausewitz observed, war is subject to an inherent tendency to escalate to levels of violence beyond those that combatants would rationally have chosen or indeed predicted at the outset: ‘If one side uses force without compunction, undeterred by the bloodshed it involves, while the other side refrains, the first will gain the upper hand. That side will force the other to follow suit...’. ‘Each side, therefore, compels its opponent to follow suit; a reciprocal action is started which must lead in theory to extremes’.15
Clausewitz's structural observation about the logic of escalation in war is, moreover, backed up by empirical evidence which shows that historically most war leaders have assumed, prior to conflict, that the war on which they were about to embark would be short and relatively easy. Sometimes they are correct. More often they are not, and the war proves to be more protracted, difficult and costly than generally expected.16 If this is correct, this points to a general and persistent tendency for leaders and war planners to underestimate the costs and overestimate the benefits of war.
Act‐consequentialism appears to offer an obvious strategy for justifying preventive wars (or wars more generally). But the appearance is illusory. There are many straightforward cases in which a decision to go to war would unambiguously lead to morally worse consequences than other policy alternatives. But it is exceptionally difficult to meet the epistemological burdens for demonstrating that going to war would lead to better consequences than alternatives. Wars are events of enormous historical magnitude, and extreme unpredictability. Action in war is subject to a pervasive tendency toward escalation, that can make even the most careful and good‐faith assessments of its projected costs wildly inaccurate. Consequentialist analysis necessarily involves issues of counterfactual history that are quite literally unknowable. In the case of preventive war, the relevant counterfactual questions include whether the feared attack would even have taken place without war—in other words whether a preventive war has ‘prevented’ anything at all.17 For (p.149) governments and citizens seeking to justify a preventive war, act‐consequentialism has almost no use.18
6.2.1. Rule Consequentialism and the Impasse Problem
When philosophers talk of a consequentialist approach to war, they are often thinking about rule‐consequentialism. Unfortunately, rule‐consequentialism faces even greater difficulties than act‐consequentialism as an ethical theory of war. This should not surprise us. Rule‐consequentialism requires assessing the long‐term consequences of iterated patterns of behavior, and hence it amplifies further the epistemological weaknesses inherent in act‐consequentialism. I argue that rule‐consequentialism suffers from two forms of epistemological weakness. First, we have no sufficient empirical data for assessing the truth of rule‐consequentialist judgments. Second the rule‐consequentialist arguments invoked by philosophers and International Relations theorists to support their conclusions persistently generate what I call the impasse problem: the existence of equally plausible rule‐consequentialist arguments for countervailing conclusions.
The rule‐consequentialist argument for preventive war is very old. It dates back at least to Cicero who argued in favor of preventive war against the internal threat of Marcus Antonius by observing that ‘every evil is easily crushed at its birth; become inveterate it as a rule gathers strength.’19 Alberico Gentili, the Oxford jurist of the sixteenth century echoed these sentiments: ‘it is better to provide that men should not acquire too great power, than to be obliged to seek remedy later, when they have already become too powerful.’20 Michael Walzer takes as the starting point for his discussion of prevention the argument of Edmond Burke, which develops this theme by invoking the notion of balance of power. According to Burke an effective balance of power between the main states of Europe preserves European liberty by preventing any one state from becoming too dominant, and secondly that fighting early, before the balance has become unstable, is less costly than waiting till the threat becomes imminent.21
As David Luban points out, the Bush doctrine as spelled out in the National Security Strategy turns this argument on its head, by claiming that the imbalance of power (i.e. US dominance) is worth preserving because it is necessary to protect (p.150) American liberty. But Luban suggests that both variants of the argument can be assimilated to a more general argumentative scheme:
1. that some state of affairs X (balance of power, U.S. dominance, whatever) preserves some important value V (European liberty, U.S. liberty, whatever) and is therefore worth defending even at some cost; and
2. that to fight early, before X begins to unravel, greatly reduces the cost of the defense, while waiting doesn't mean avoiding war (unless one gives up V) but only fighting on a larger scale and at worse odds.22
This rule‐consequentialist argument seems plausible.23 However as both Walzer and Luban point out, there is a rule‐consequentialist rejoinder that is at least as plausible. The rejoinder is that even if (1) and (2) are true, it would be better on consequentialist grounds for states not to accept them because doing so would expand the grounds for going to war and thereby lead (in Burke's words) to ‘innumerable and fruitless wars.’ The ultimate consequences of a general rule permitting preventive war would therefore be worse than one that prohibited it.24 What we are faced with is what I call the impasse problem: two competing and apparently equally plausible rule‐consequentialist interpretations of the preventive‐war norm. One interpretation claims that a rule permitting preventive war would lead to better overall consequences because wars fought early would have less human cost, and would protect valuable goods such as national or international liberties. The other interpretation claims that a rule prohibiting preventive war would lead to better overall consequences because a more permissive doctrine would multiply the number of wars fought.
Which interpretation is correct? Luban has three rule‐consequentialist arguments for why a general norm permitting preventive war should be rejected. I suggest that each can be matched by an equally plausible rule‐consequentialist interpretation for the opposite conclusion, hence the impasse problem is not avoided. I want to stress that my intention is not to endorse these counterarguments as the correct rule‐consequentialist assessment of prevention. Rather by showing the ease with which rule‐consequentialist arguments can be matched by apparently equally plausible counterargument I wish to draw attention to the spurious character of these arguments' apparent plausibility.
Luban's first argument is that a general right of prevention would make war more likely because it would broaden the category of permissible war. In particular it would justify first strikes by both sides in all the world's most dangerous hotspots. For example, it would justify a first strike by both India and Pakistan, North Korea and Japan, Israel and Syria, and most frighteningly during the Cold (p.151) War it would have justified preventive strikes by both the United States and the Soviet Union given a favorable opportunity.25
But the counterinterpretation of these cases runs as follows: the existence of such hotspots itself imposes substantial moral cost because of the risk and instability they bring to international affairs, and the way they impede valuable economic and cultural exchange. Over the long term, a rule permitting prevention would have positive consequences, by transforming the strategic environment itself in ways that make such dangerous strategic standoffs less likely. Thus, if a norm of prevention had permitted dominant states to hit early and hit hard when a potential rival emerged, many of the hotspots referred to by Luban might never have developed. For example, had the United States used its nuclear monopoly at the end of World War II to act preventively against the Soviet Union, there might never have been a Cold War with the associated risks of global nuclear annihilation. Similarly, one may argue, the other hotspots only pose a danger to the world because the emerging standoff was not nipped in the bud by early preventive action.26
Luban's second argument is that a general doctrine of prevention introduces too much ambiguity into the rules of war, which will in turn make war more likely. The reason the doctrine introduces ambiguity is that the judgment as to whether a state constitutes a sufficient threat to justify preventive action necessarily involves what Rawls calls burdens of judgment [the inevitability that different (reasonable) people's judgment will diverge] and infirmities of judgment (the fact that people's judgment about matters of great moment and high risk is seldom rational).27
But it is not clear that increased ambiguity in the rules of war would increase the incidence of war. Michael Byers has argued that ambiguous laws increase the latitude of action for strong states while having little effect on the ability of weak states to engage in military action. This is because ambiguity ‘allow[s] power and influence to play a greater role in the application of the law to particular situations.’28 Strong states can bring to bear various forms of political, economic, and military pressure to influence the assessments pertaining to the application of the law, while still retaining the law as a diplomatic tool to be deployed against weaker states. If this assessment is correct, then one can see how an ambiguous law could increase the overall stability of the international system, especially when (as is now the case) the international system is dominated by one powerful state. By freeing the hand of the strong state to take decisive measures against potential rivals and (p.152) disruptive smaller states, the rule could have a strong deterrent effect that would mean fewer wars in the long run. Moreover, because the strong will disproportionately influence how the ambiguous criteria for the rule are applied, there is little risk of a general increase in violence among weaker states. Luban's third argument is linked to the first two. The doctrine of prevention ‘actually makes rival states into potential threats to each other by permitting preventive invasion of potential adversaries based on risk calculations whose indeterminacy makes them inherently unpredictable by the adversary—and then it licenses attacks by both of them, because now they are potential threats to each other’.29
But here we need to know much more about the specific incentive structures in play and how they would cash out in real strategic negotiations. According to standard assumptions of International Relations theory, the primary interest of states is to maintain their own security. Typically states will seek to achieve this through building up their military capability and developing alliances to deter neighbors or rivals from attacking them. This deterrence strategy can result in reasonably long periods of stability and security. But the strategy is not cost free. It can also contribute to the initiation of disastrous conflicts that diminish the security of all states involved, as for example the armament and alliance strategies of the major European powers did in the lead up to World War I.
Luban's argument suggests that a general doctrine of prevention would radically increase the costs of the deterrence strategy by making it much more difficult to develop and maintain a stable relationship of military deterrence. This conclusion contradicts the arguments I have just given which suggest that stabilizing deterrence effects could indeed be increased by recognizing a general doctrine of prevention. But suppose Luban's conclusion is correct. In this case, the effects of a doctrine of prevention could play out in other ways.
In essence, if Luban is correct then the deterrence strategy of military armament and alliance formation would cease to function as a rational strategy for states to maintain their security under a general doctrine of prevention. Given this, states would have to look to alternative strategies to maintain their security. Plausible alternatives would include strategies of mutual assurance and disarmament, for example through the implementation of strong nonaggression treaties and mutual disarmament initiatives backed up by independent inspections and monitoring. It is moreover plausible that the mutual assurance and disarmament strategy would results in less cost overall than the deterrence strategy. Therefore by changing the incentive structures so as to make the current deterrence and armament strategy untenable (if indeed this was the effect), a general doctrine of prevention could still lead to better overall consequences.30
Once again we are faced with an impasse between differing and opposed rule‐consequentialist interpretations of the doctrine of prevention. As I have said, I (p.153) certainly do not want to claim that my counterarguments should be accepted as the correct and authoritative rule‐consequentialist interpretation of prevention. Nor, indeed, do I need to establish this for my argument to succeed. I simply need to show that they are as plausible as the countervailing interpretations given available information. The relevant question is of course: is there a conclusive empirical basis for ruling out one set of rule‐consequentialist arguments or interpretations in favor of those that support their contradictory?
What is most striking in the rule‐consequentialist literature on war is the general absence of reference to empirical evidence that could answer these questions.31 On reflection, it seems obvious why this is. There does not seem to be any empirical evidence capable of verifying rule‐consequentialist interpretations of the rules of war. It is certainly possible to subject the historical record of war to statistical analysis (although it is interesting how little interest most rule‐consequentialist philosophers show in sustained analysis of the historical record). But at best this can tell us what the incidence of war was during periods when the rules of war were interpreted differently. It cannot tell us what the incidence of war would have been, had the rules of war been different. For we have no way to control for the effects that differing social, economic, technological, climatic, administrative, and institutional conditions have on the incidence of war. Not only do we not have the required empirical evidence to settle such rule‐consequentialist questions, it is unclear even what could count as empirical evidence.32
In light of this absence of empirical evidence, rule‐consequentialists are compelled to resort to general arguments of the form Luban offers. It is not that these arguments seem implausible. On the contrary they seem very plausible. It is just that there are rule consequentialist arguments for the countervailing position that are equally plausible. Once we recognize this, we can see that the apparent plausibility of these rule‐consequentialist arguments is spurious. The arguments are a kind of armchair conjecture, conducted without empirical data or even a clear conception of what would count as empirical data.
Rule‐consequentialist arguments, like their act‐consequentialist counterparts, can provide no useful justification of preventive war. There is no empirical evidence that could settle the question of which interpretation of the laws of war will lead to the best overall consequences, and the arguments that philosophers and legal theorists use to defend their preferred consequentialist interpretation of the (p.154) rules of war can be matched, point for point, with equally plausible arguments for the exact opposite conclusion.33
But one might have a worry about the argument at this point. If the epistemological critique of act and rule‐consequentialism in the ethics of war is valid, then surely it proves too much. If the long‐term consequences of complex decisions and the rules that might govern them are always indeterminate, then how can we explain the role that consequentialist cost‐benefit analysis clearly, and appropriately, plays in many spheres of life, most notably in the realm of public policy, administration, and economic management? How is future oriented practical judgment possible at all?
The short answer is that there are distinctive problems in assessing the consequences of decisions and rules relating to war that are not present in many other areas of practical judgment. These problems have to do with the factors I have already surveyed: the immense temporal and geographical scale of war's effects, the poorly understood, but very real dynamic of escalation in military conflict, the pervasive role played by chance in military outcomes and the persistent psychological tendency for planners to underestimate costs and overestimate the benefits of war. These are factors that are either unique to war, or are present in war to a far higher degree than in most other human activities. Moreover, in war we have no empirically tested predictive theory of outcomes in the way that we have, for example, in economics. Although there have been attempts to create predictive models of military outcomes there are many variables, both human and natural, that simply cannot be accounted for. As the United States discovered to her cost in Iraq, even possessing a military superiority virtually unparalleled in the history of warfare is no guarantee of successfully achieving military objectives in a war.
But the concern might be restated in another way: if there is a distinctive epistemological problem about judging the consequences of warfare, then surely this will be an equal problem for any plausible deontological approach to war including the just war theory. After all, several aspects of the just war theory clearly require assessment of the consequences war, for example the jus ad bellum criterion of proportionality. If accurate assessment of the future consequences of war is impossible then just war theory must suffer from the same indeterminacy as act and rule consequentialism.
(p.155) I think this objection can also be met. There are two ways we might interpret the criterion of proportionality. On the first interpretation, proportionality states that for a war to be justified there must be reason to believe that the good consequences of the war will exceed the bad. On the second interpretation, proportionality articulates a less demanding condition that takes the form of a defeater condition. It tells us that even if a war is prima facie justified because the other conditions of jus ad bellum are met, the war will still be unjustified if the expected moral costs are grossly disproportionate to, that is significantly greater than, the expected benefits.
Neither interpretation makes just war theory indeterminate in the way that consequentialist approaches to war are indeterminate. I have already argued in section 6.2 that the kind of negative and minimal judgment required by the second interpretation of proportionality raises far fewer epistemological problems than the positive judgment required for a consequentialist justification of war. In those cases where war would be grossly disproportionate because the goods it is intended to achieve are trivial, this fact is often manifestly obvious especially compared to the open‐ended and escalating harms of war. There is little epistemological difficulty with this kind of judgment.
What then of the first interpretation of proportionality? We must concede that this does require making essentially the same analysis of a war's costs and benefits as we find in consequentialism, and I believe that the task suffers from the same epistemological problems and indeterminacy as consequentialism. But because satisfying proportionality is a necessary condition for the overall justification of war, this fact does not serve to make just war theory itself indeterminate. For in any case in which it is impossible to determine whether good consequences will outweigh the bad (because of epistemological difficulties), the war is ipso facto unjustified. On this interpretation of proportionality, just war theory is not indeterminate, it is simply highly restrictive.
While it is an important and open question as to which interpretation of proportionality is correct, this discussion brings out one of the key strengths of just war theory compared with consequentialism. In just war theory, proportionality functions as one condition within a network of other considerations that collectively serve as a potential justification for war. In contrast, when consequentialism is employed as a justification for war, indeterminate positive assessments of the long‐term consequences of fighting must carry the full justificatory burden on its own. This, I have argued, they are simply unable to do.
6.2.2. Compliance Effects and the Development of International Law
I now want to suggest that the preceding rule‐consequentialist arguments for preventive war are all vulnerable to a deeper objection still. Each assumes a causal connection between the content of the rules of war and the behavior of states and other international actors. The arguments turn on what a given interpretation of the rule of prevention would ‘lead to’ or to what consequences the rule would ‘produce’. But of course these causal effects only follow if one makes two (p.156) assumptions: (a) that compliance levels for the rules of war are high and (b) that compliance levels are largely invariant to changes in the formulation of the rules. Unfortunately both assumptions are obviously false.
In connection with this point Luban makes a critical move.
If [the critic of rule consequentialism] means that no evidence can show that a ban on preventive war would save lives, the reply is that no evidence can show that any doctrine of just war saves lives, simply because states so frequently disregard moral and legal norms. The right test for a moral norm should not be whether the norm will be efficacious, but rather whether it would be efficacious if states generally complied with it.34
But Luban's interpretation of the rule‐consequentialist test is precisely the wrong one. Rule consequentialism is defined by the claim that moral rules are to be selected solely on the basis of their expected consequences. Given this, the correct rule‐consequentialist test is what the consequences of endorsing a rule would be under expected levels of actual compliance. The consequences a rule would have under hypothetical (and as Luban himself explicitly acknowledges, unrealized) compliance conditions are strictly irrelevant.35 Quite simply, hypothetical consequences are not consequences. Therefore, they should play no role in rule‐consequentialist analysis.36
This is a crucial point and once we acknowledge it, several things become clear. The first is that the rule‐consequentialist arguments we have considered so far, both those for and against preventive war, are all defective. For they all assume high compliance levels for the rules of war that do not actually exist, and they unrealistically consider the consequences of different formulations of rules while holding the assumed compliance levels constant.
Once we recognize that these two propositions are false and that the way in which a rule is formulated can have a reflexive effect on that rule's compliance level, then a whole new and important area of rule‐consequentialist argumentation is opened up to us. Indeed, we are now in a position to formulate the rule‐consequentialist argument for preventive war in a different and more powerful way. Moreover, this argument links directly with Luban's discussion of whether, in today's strategic environment, it may not be reasonable (p.157) to grant to the sole superpower exceptional permissions not applicable to other states.37
The argument runs like this: Suppose a supremely powerful state is bent on a course of action that is inconsistent with the current rules of war as enshrined in international law (e.g. the prosecution of a preventive war). There are three likely outcomes. The first is that the international community reinterprets the rule in a way that is consistent with the strong state's desired course of action (e.g. by permitting preventive action under the norm of self‐defense). The second is that the international community maintains the original interpretation of the rule, but grants an exceptional permission to the strong state that is not granted to the other states. The third is that the strong state simply abandons the system of rules by acting outside the accepted framework of international institutions and international law.
Each of these three outcomes would have serious consequences, but there is an argument to be made that the third eventuality would be the most damaging of all, because it threatens the viability of the system of international law itself. For no system of international rules can plausibly function without the participation and support of the dominant states. This is because, firstly dominant states are among the most important international actors, and secondly if dominant states do not participate in the legal order it reduces the incentives for smaller states to do so. If all of this is correct, then on consequentialist grounds one may judge that it would better overall to either (a) adopt an interpretation of the rule consistent with the dominant state's desired course of action even if one believed that such an interpretation would produce more immediate harm than good, or (b) to grant to the strong state an exceptional permission within the system of rules in order to (c) maintain the participation of the dominant power with the system of rules and hence maintain the integrity of the international legal order itself.
This argument has particular force if one views international law as a work in progress. That is to say one views it as a nascent system of norms, with some potential to develop into a robust legal system with high compliance levels and real sanctions impartially and consistently applied. On this view the international legal system has value as much for what it might one day become as for what it already is. For those who take this view, who care deeply about the development of international law, who believe that it is an imperative of the most urgent kind to foster and protect the emerging international legal order, and who were dismayed (p.158) and appalled at the damage wrought on international institutions by diplomacy in the lead up to the war in Iraq, this argument generates a painful dilemma.
How should we respond to this consequentialist argument that explicitly links the formulation of rules, or the granting of exceptional permissions to particular states within the system of rules, to the viability of the system of rules as a whole? The first thing to note is that the argument is framed from the perspective of those outside the policymaking organs of the dominant state. Because it takes the actions and intentions of the dominant power as a brute fact, this argument cannot be used by US policymakers themselves to justify their demands that the United States be granted an exceptional permission. But what about those of us outside the corridors of power of the dominant state? What should we make of the argument?
An obvious response is to push home the epistemological critique of consequentialist arguments developed in the preceding sections. How are we to assess the relative costs and benefits of the different outcomes of rule amendment, the granting of exceptional permissions, or the abandonment of the system of the rules by the dominant state? Over what time frame? And holding which assumptions constant? How are we to assess the counterfactual consequences of the strategies not chosen and the opportunity costs of those chosen?38 All of these questions are deeply obscure and once again we are reminded of how little traction consequentialism gives us on the issues involved.
But a different and instructive form of response is suggested by Michael Byers in a challenging recent paper. He argues that it might actually be better for the system of international law as a whole if the ‘United States simply violated international law without advancing strained and potentially destabilizing legal justifications’.39 The reasoning behind this claim is that the actions of a state will only have a precedent‐generating effect in customary international law if the state claims a legal justification. He therefore advocates that the US approach to prevention should be one of ‘exceptional illegality followed by mitigation’40—essentially to explicitly violate the law while making an extra‐legal claim of reduced responsibility. He writes:
It follows that if a state admits—explicitly or implicitly—that it is violating international law, the overall effect is as likely to strengthen the rule as weaken it. The situation is in some (p.159) respects comparable to the prohibition of torture, which remains a strong rule despite a pattern of violation by some 120 states. This continued strength would seem to flow from the fact that those states consistently deny and seek to conceal these acts.41
Yet this solution seems as unsatisfactory as the problem it is formulated to address. The difficulty is apparent if one asks in what sense a rule can be said to be ‘strong’ if there is ‘a pattern of violation’ by 120 states (roughly 60% of international community)? The argument seems to belie a fetishism about the formalistic integrity of the law, without sufficient concern for the ability of the law to provide an effective remedy (in this case to the continuing problem of torture). Why should we care about the integrity of the law, if that integrity masks, and to a certain extent is dependent upon, widespread noncompliance? This critical observation can be generalized into a robust response to the reformulated rule‐consequentialist argument for prevention. The response runs like this: a primary function of legal systems is to protect the weak from predatory or arbitrary action by the strong. But if the participation of the strong within a system of law can only be secured by amending the law, or by granting an exceptional permission so as to allow potentially predatory or arbitrary behavior by the strong that would otherwise have been ruled out by the law, then the legal system has substantially failed to fulfill this function. There must be an appropriate relationship between means and ends in moral argument and action. In this case the means employed to support the system of international law are antithetical to the deeper purposes and principles of the law itself. The integrity of the system has been purchased, to some degree, at the expense of its effectiveness and impartiality: two of the most important features of any just legal system.
If the proposed exceptions or amendments to existing law were minor and concerned matters of peripheral importance, then it may be plausible to modify the legal system to ensure the general participation of the strong. But if the proposed modifications concern—as they do in the case of preventive war—a central norm regarding the fundamental issue of when force may be used in international relations, then the situation is very different. If we are prepared to bend the law or grant exceptional liberties to strong states concerning norms of fundamental importance, then we may wonder what is the value of the legal system whose integrity we are aiming to preserve. Or to phrase the argument in more consequentialist terms, it is difficult to see where the beneficial consequences of the system of law will come from if the strong get to do what they want either way.
In this first part of the chapter, I hope to have shown that consequentialist theorizing about war is defeated by the profound epistemological difficulties of assessing the long‐term effects of individual wars on the one hand, and the rules of warfare on the other. What is fatal to the consequentialism of war is not simply that we do not know how to answer the questions that consequentialism asks, nor even that we cannot know, but that we do not even know what it would mean to know (in the sense of having a clear conception of what would count (p.160) as attainable empirical evidence in the real world one war or the other). Given this we should abandon the fruitless and unproductive attempt at second guessing the construction and interpretation of the rules of war on consequentialist grounds and concentrate our efforts on an activity in which philosophical and legal analysis can gain some genuine traction: that is answering the question of whether, and under what conditions, preventive war could be right or just. It is to this task I now turn.
6.3. PREVENTIVE WAR AND SELF‐DEFENSE
I believe that the best way of investigating the justice of preventive war, is to look at the question of whether preventive war can be justified as an act of legitimate self‐defense. The notion of self‐defense is fundamental to the way we think about the ethics of war, and the National Security Strategy itself casts the issue of prevention as one of self‐defense.42 The right of self‐defense is normally taken to require that three conditions be met.
1. Necessity: defensive action must be necessary in the sense that there is no less harmful way to avert the harm or attack.
2. Proportionality: harm inflicted in the course of defensive action must be proportionate to (or not disproportionately greater than) the harm or attack one is seeking to avert.
3. Imminence: the harm or attack one is defending against cannot be a distant or remote one, but must be truly immanent.
The National Security Strategy addresses, at least implicitly, each of these conditions. The condition of immanence is explicitly acknowledged, but it is argued that ‘we must adapt the concept of imminent threat to the capabilities and objectives of today's adversaries.’43 This adapted conception of immanence is clearly meant to include more than the ‘visible mobilization of armies, navies, and air forces preparing for attack’44 and would permit action against ‘emerging threats before they are fully formed’45 and ‘even if uncertainty remains as to the time and place of the enemies attack’46.
But this talk of ‘adapting’ the concept of imminence is, once again, mere rhetoric. Any doctrine which allows the use of force against potential threats that (p.161) have not yet fully formed and where it is unclear when, or even if, an attack will occur has not adapted the traditional condition of immanence, but has abandoned it. The real challenge of the Bush doctrine, therefore, is not whether the concept of imminence needs reform but whether it needs to be abandoned entirely. To put the question another way: can we can make sense of the right of national self‐defense, without the traditional condition of immanence? It is this question I propose to discuss and I hope to show that the notion of imminence lies very deep in the idea of self‐defense. Correspondingly, it is very difficult to understand preventive war as a legitimate part of self‐defense.
I used to believe otherwise: that the concept of immanence is not fundamental to the right of self‐defense. I previously argued that the condition of immanence is subsidiary to the condition of necessity.47 Imminence, I argued, is derived from necessity because of epistemic constraints on human action. The point is that one cannot know that defensive action is necessary, unless the attack one is defending against is really imminent. If on the other hand, one had superhuman knowledge and could foresee that the only way to prevent some future wrongful harm or attack was to engage in defensive measures, then one would not be required to wait till the attack was imminent: necessity would be enough.
If such an interpretation were correct, then it would open a way to allowing prevention as a legitimate form of self‐defense. The view would certainly place an enormous, and probably unsustainable, burden on the intelligence used as a basis for going to war, for one would have to be certain that the defensive measures were really necessary to prevent the harm in question. But the idea of a preventive war of self‐defense against a nonimminent threat would not be ruled out at the level of moral theory.
I now think that this view is wrong, and that the condition of immanence is much more fundamental to the right of self‐defense. To see this we must look into the theoretical underpinnings of the right of self‐defense. The right of self‐defense as we currently have it in law and moral theory, has been influenced by at least three distinct traditions of thought. The first is the consequentialist tradition. The second tradition views self‐defense as action conditioned by a kind of psychological necessity. The third tradition views the aggressor as morally liable to harm by an act of legitimate self‐defense and explains this liability on the basis of his fault or culpability for initiating the attack. As I have discussed consequentialism above, I will say no more about it here. But I do intend to examine the role played by imminence in the other two traditions, both of which have been influential on modern ideas of legitimate self‐defense.48
The connection between self‐defense and necessity is an old and deep one. In fact, in some jurisdictions self‐defense is referred to simply as ‘necessary defense’. Necessity in this context is close to the legal concept of duress. It is a form of excuse stemming from the fact that a wrongful action is viewed as psychologically necessary or unavoidable in situations of extreme danger or pressure.49 It is probably this conception of necessity that Cicero had in mind in this famous passage:
there is a law not of the statute‐book, but of nature, ...that should our lives have fallen into any snare, into the violence and the weapons of robbers or foes, every method of winning a way to safety would be morally justifiable. When arms speak, the laws are silent ...50
What underlies this necessity approach is a compassionate stance towards the frailty of human nature. Given the kinds of creature we are, there are some things (such as abstaining from taking action sufficient to save our own life) that we simply cannot be expected to do.51 As Aristotle says they ‘strain human nature to breaking‐point and no one would endure.’52 If it is to be just and humane, the law must take account of this, and not hold us responsible for actions taken under the enormous psychological pressures of the shadow of death.53 This psychological necessity view of self‐defense was reflected in the medieval law of se defendendo. According to this law one was not to be punished for killing in the course of a fight, if one quite literally ‘had one's back against the wall’, and taking the opponent's life was the only way of saving one's own.
Now one might think that such a conception of self‐defense can have little connection with the laws of war, for states are not psychological entities that feel fear in the way that individual persons do. But it is precisely this account which is suggested by the most famous legal text on the right of national self‐defense: the Caroline doctrine. This text states that a nation's right of self‐defense is based on a ‘...necessity of self‐defense, instant, overwhelming, leaving no choice of means, and no moment for deliberation’.54 Although states do not experience (p.163) the psychological fear of a real person, perhaps they can experience something analogous.55
Viewing self‐defense in this way, does however, have a number of significant and somewhat unpalatable consequences. The first is that contrary to modern legal theory, self‐defense is viewed not as a justification but merely as an excuse. An act is justified if it is the right or permissible thing to do in the circumstances. An act is excused if it was the wrong thing to do, but the agent ought not in the circumstances be held fully responsible for the act. Cicero captures this implication astutely with his claim that when arms speak ‘the laws are silent.’ It is not that the laws approve or permit the actions in question; they simply abstain from passing judgment.
The second consequence is that self‐defense on this view takes a form which is rather peculiar to the modern eye. The modern right of self‐defense encompasses, in most jurisdictions, the defense of others including strangers. But because on the psychological necessity view self‐defense is grounded in the idea of action under extreme fear or pressure, the excuse is only available for defending certain kinds of person. The medieval claim of se defendendo was only available if one was defending one's own life or that of close friends or family. Only then were the psychological pressures considered sufficiently grave to provide an excuse for murder.
Thirdly, the necessity view makes no discriminations based on fault as to who can claim the excuse of self‐defense. The excuse is available to anyone facing the extinction of their life, be they innocent victim or culpable aggressor of the original attack.56
What is clear, however, is that the psychological necessity view of self‐defense is quite incompatible with the idea of a preventive war. For on this view it is precisely the imminence of the potentially fatal attack and the extreme psychological pressure this imposes that makes defensive acts excusable. When the attack is remote and merely potential there is no psychological necessity of this form. What is required on the necessity view is precisely suggested by the language of the Caroline doctrine: ‘...necessity of self‐defense, instant, overwhelming, leaving no choice of means, and no moment for deliberation’.57 In contrast (and almost in contradiction) to this the National Security Strategy explicitly states: ‘We will always proceed deliberately, weighing the consequences of our actions.’58 This brings out an important feature of preventive wars; they are wars of choice. Their function, strategically, is to engage the enemy at a time and on terms of one's own choosing. As such they are antithetical to the tradition of thought on self‐defense (p.164) which views that right as grounded in the overwhelming psychological necessity of facing an imminent lethal attacker.
If a doctrine of preventive war is to be integrated into self‐defense then it must be on the basis of a very different theory of self‐defense to that of psychological necessity. In fact, the most plausible theory of self‐defense is very different to the necessity view. It is a theory which views self‐defense as a justification grounded in a conception of human rights and concomitant notions of liability to harm.
6.3.2. Self‐Defense and Rights
Here I do no more than sketch out the rights‐based theory of self‐defense which I have defended in detail elsewhere.59 One starts with the observation that human beings have rights. Paramount among these is the right not to be killed or subject to significant bodily harm. But killing and harming another human is precisely what is done in acts of self‐defense including the fighting of a defensive war. This raises a simple but profound question for the theory of self‐defense: what is it that makes the person against whom one is using defensive force liable to be harmed in this way? Some part of our theory of self‐defense must account for the fact that an aggressor has lost or forfeited or fails to possess the right to life when he is killed in defense.
The best way to account for this moral liability to harm on the part of aggressors is to point to the fact that they commit wrong by engaging in an act of aggression. One way to understand this is to observe that many rights are implicitly reciprocal in their nature. Thus on a plausible understanding of rights, one only has the right to life so long as one respects the right to life of others. At the moment one wrongfully attacks the life of another, one becomes morally vulnerable or liable to being killed in self‐defense. This explains why we believe that it is permissible to kill an aggressor in self‐defense but not to kill an innocent bystander even if this were the only way to save one's life. For an innocent bystander has done us no wrong and therefore has the right not to be killed by us, even if this were the only way to save our life.
If one accepts that the permission to kill in self‐defense is tied to some wrongdoing on the part of the aggressor, then it is easy to see why there is a problem with preventive acts of self‐defense.60 In a preventive war, one attacks and kills those (p.165) who have not yet committed an act of wrongful aggression against you. Without the presence of active aggression it is difficult to see how there can exist the liability to harm which seems to be such a crucial part of the classic model of self‐defense.61
Against this, it might be objected that in cases of preventive self‐defense one has good reason to believe that the persons one is using force against are about to engage in wrongful aggression. Is not this sufficient to establish their liability to be harmed? If you have good reason to believe that they will act aggressively, why must one wait till they actually engage in aggression before using defensive force?
To answer this question, we may consider an analogy with liability for punishment in criminal law. The analogy is not perfect because self‐defense is not itself a form of punishment. Nonetheless there are sufficient similarities in the grounds of liability for punishment and self‐defense to make the comparison instructive. In order for there to exist liability for punishment one must establish both a criminal intent and the existence of a criminal act—both mens rea and actus reas. Without criminal intent and the existence of a criminal act a person may not be punished even if we have good reason to think that he will commit a certain crime in the future.
In the movie ‘Minority Report’, magical clairvoyants have the ability to see infallibly into the future. Law enforcement agencies are able to use these predictions to arrest and punish those whom the clairvoyants had predicted will commit murder, before they ever have the chance to commit the crime. This procedure is viewed as efficient law enforcement in this imaginary future world. But most of us feel a deep moral repugnance toward such an idea (this indeed is the point of the movie), and this stems from the fact that it is unjust to punish someone who has not yet done anything to merit punishment (even if we grant some supernaturally reliable prediction that they will do it).
On the rights‐based theory of self‐defense a similar problem arises with preventive self‐defense. A Minority Report style supernatural example is not required to see this. Imagine that I know you have a violently jealous nature and that if you ever discovered that I have been sleeping with your wife, you would almost certainly take your revenge by killing me. Suppose now that I see you reading through my private papers and that among those papers is a letter proving my infidelity. As soon as you read that letter you are almost certain to form and act upon a murderous intent. But it seems clear in this case that I would be justified in using defensive force only after this had happened. Until such time, there is nothing you have done that makes you liable to my defensive force. Preventive self‐defense is therefore ruled out, because you retain your full rights not to be attacked or harmed.
(p.166) Examples like these show us that the conditions of self‐defense (necessity, proportionality, and imminence) function in a fundamentally different way within a forfeiture‐of‐right theory of self‐defense than they do within a consequentialist theory of self‐defense. Within a consequentialist theory, the conditions of self‐defense are interpreted so as to ensure that the balance of harms in a self‐defense situation is maximally conducive to overall welfare. It is therefore appropriate to construe the conditions in a way that is probabilistic. For example, a highly harmful potential attack with a low probability of occurring might yield the same defensive justification as a less harmful attack with a high probability of occurring. But within a forfeit of right account of self‐defense, the function of the self‐defense conditions is quite different. Although on this account the necessity, proportionality, and immanence conditions often function to reduce harm, they are not intended to reliably minimize overall harm, as on the consequentialist model (e.g. one is entitled to kill virtually any number of culpable aggressors if this is necessary to save one's own life). Instead their central function is to ensure that the person or persons against whom defensive force is used are really morally liable to the infliction of force, and that the level of harm inflicted on each aggressor is proportionate to his or her individual moral liability.
The imminence requirement, then, does more than simply bolster the necessity requirement in conditions of imperfect knowledge or ensure that the outcome of conflict situations are welfare maximizing. By requiring that self‐defense cannot be undertaken until an unjust attack is imminent, the condition guarantees that those who are subject to defensive force are morally liable to it because they are currently engaged in an unjust aggressive attack. It would seem then, that the condition of imminence has an essential not derivative role in the theory of self‐defense and that preventive self‐defense is thereby ruled out at the level of moral theory.
6.3.3. The Conspiracy Paradox
There is, however, a complication. In certain circumstances one may be liable to defensive force, and even to punishment, without having committed a criminal act, if one has engaged in sufficiently clear and serious preparations for performing such an act. These cases are covered by the laws of criminal conspiracy. On the surface these laws seem to contradict the general principle that punishment and self‐defense require liability grounded in active wrong doing on the part of those against whom force is used. But this appearance is deceptive, for the liability to punishment and harm in these cases is grounded in the fact that conspiracy to commit a crime is itself a criminal offense. It is the wrongdoing of engaging in a criminal conspiracy that underlies the liability to harm in these cases. This is demonstrated by the fact that penalties for conspiracy offenses are different to (and normally lower than) those for the corresponding crime.
But this observation seems to open up a new route to justifying preventive self‐defense. If conspiracy to engage in an attack is itself a culpable wrongdoing, then (p.167) there may be a right of preventive self‐defense in cases where the attack is not imminent but in which there is nonetheless a proven conspiracy to commit it. This seems to be the view of Michael Walzer:
the line between legitimate and illegitimate first strike is not going to be drawn at the point of imminent attack, but at the point of sufficient threat. That phrase is necessarily vague. I mean it to cover three things: a manifest intent to injure, a degree of active preparation that makes that intent a positive danger, and a general situation in which waiting, or doing anything other than fighting greatly magnifies the risk.62
I will assume that ‘intent to injure’ together with ‘active preparation’ is roughly equivalent to a conspiracy to attack. Walzer's view could then be paraphrased as follows: conspiracy to attack together with necessity equals a prima facie just cause for war. Call this thesis the conspiracy account of preventive war.
This account seems on the surface plausible. Unfortunately, the position is subject to an unpalatable paradox. Everything hinges on whether engaging in a ‘conspiracy to attack’ is a wrongful act sufficient to ground a right of preventive defense. For Walzer this consists of two elements: (a) manifest intent to injure and (b) active preparation. But (and here is the problem) these two elements precisely describe an articulated doctrine of prevention such as the National Security Strategy. That document affirms quite explicitly both the intention to attack (‘...the United States will, if necessary, act preemptively’.63) and the active preparations for doing so (‘to support preemptive options, we will: build better, more integrated intelligence capabilities...coordinate closely with allies...transform our military forces...’64). But if a conspiracy to attack (consisting in a manifest intent and active preparation) is a form of wrongdoing sufficient to ground preventive military action, then doctrines of prevention such as the National Security Strategy are ipso facto morally wrong. The paradox implicit in the conspiracy account of prevention now emerges clearly. If manifest intent and active preparation together constitute a wrong sufficient to ground preventive war, then any doctrine of preventive war is impermissible. If, on the other hand, doctrines of prevention are permissible, then the combination of manifest intent and active preparation are presumably not in themselves wrong and this implies that there is no sufficient moral ground for preventive war.
The paradox is similar in structure to the famous liar's paradox, which consists in a statement that is true if it is false and false if it is true. On the conspiracy account, if preventive wars are permissible, then doctrines of preventive war such as the National Security Strategy are impermissible. If doctrines of preventive war are permissible, then preventive wars are impermissible. This paradox severely undermines the conspiracy account of preventive defense.
How might one evade this paradox? There are a number of possibilities. The first is based on a distinction suggested by David Luban in his contribution to (p.168) this volume. Luban argues that the kind of threat that could justify preventive action must be a large‐scale threat to basic human rights most clearly evidenced by a war of subjugation. An acceptable doctrine of prevention, on the other hand, will be strictly limited to planning for the use of minimum force directed solely against the enemy's military capabilities: ‘...unless the preventive war itself aims at a large‐scale attack on basic human rights, planning for it is not a conspiracy to commit the wrongs that would justify preventive war’. Referring to my own argument he concludes: ‘Paradox lost’.65
I believe that this distinction, on its own, does not take us very far in avoiding the paradox. The problem is that the distinction is question begging. Even an attack that is strictly limited to proportionate and necessary action against legitimate military targets can constitute a large‐scale violation of basic rights. It will constitute such a violation of rights if the attack is itself unjustified (as for example an aggressive war conducted solely against military targets violates human rights in the victim society). But of course whether engaging in a preventive war against a nonimminent threat is justified as a form of defense or is an unjustified act of aggression is precisely what is at issue. Luban's distinction does not help us to escape the basic quandary at the heart of the paradox: if the manifest intent and active preparation to carry out a preventive attack against the military capability of the enemy is wrong, then doctrines of prevention are also wrong. If such intention and preparation is not wrong then the justification for preventive war evaporates because (so long as the enemy is not planning to attack nonmilitary targets) there is no prior wrongdoing to ground and explain the enemy's liability to harm. Paradox regained.
But there is a further objection to the paradox which is also implicit in Luban's chapter.66 A distinction might be drawn between conspiracies to attack which are reactive and conditional, in the sense that they are formed in response to a preexisting conspiracy to attack on the part of an enemy (whether this attack is directed solely against the military of the broader society), and conspiracies which are not conditional in this way. The thought is that there is a form of conspiracy to attack that is not morally objectionable because it has the form: ‘we are only planning to attack them because they are planning to attack us’. One could then imagine a rule that permitted planning and preparation for preventive war and whose structure mimics the current structure of the international law on the use of force. This states that a first use of force is prohibited, but defensive force in response to a first use is permitted. In a similar fashion, one might say that engaging in a first conspiracy to attack is prohibited whereas planning a preventive attack in response to a first conspiracy is permitted.
I think that the conditional intention response to the conspiracy paradox does identify a theoretically valid construction for the law, but there are two problems with it. The first (which actually applies more generally to any form of the conspiracy account) is that the realm of liability generated by a wrongful conspiracy (p.169) will most often be too narrow to justify a full‐fledged preventive war. For who is responsible for the wrongful conduct that constitutes an unconditional conspiracy to attack? If the threatening conspiracy is to be kept a secret (as presumably it must be) then it will be the ruling executive together with a small cabal of military and political elite who are responsible for planning it. It is this group, rather than the ordinary soldiers and state officials who may be doing nothing more than their regular and unobjectionable duties, who have taken steps sufficient to make themselves liable to preventive defensive force. How can we then justify a broader preventive war which involves the killing of soldiers who have done nothing to assume liability to lethal force?
Luban responds that all wars involve the killing of innocent people, and that ‘unless we accept pacifism, the destruction of innocent life (tragic and senseless as it surely is) does not by itself make an otherwise just war unjust, provided it satisfies proportionality.’67 While Luban, earlier in this chapter, provides a subtle and persuasive discussion of the many ways in which war necessitates killing persons in ways that would be prohibited in the strongest terms under normal domestic morality, he is surely wrong to conflate these cases with that of killing a soldier whose leadership is covertly conspiring to start a war.68 For example, when we say that a conscript soldier forced at gunpoint to serve in an aggressive war is morally innocent, what we mean is that the soldier has an exculpating excuse for action that is objectively unjust. It is well established in law and moral theory that engaging in an excused but objectively unjust harmful act can generate liability to defensive force. In contrast, a soldier sitting in his barracks or going on exercise while his leadership, largely unbeknown to him, plots an aggressive war has not engaged in any use of force at all, let alone an objectively unjust one.
Similarly, we know that all wars will cause the collateral killing of innocent civilians at a level which would never be tolerated under domestic law or morality. But a state that uses force against soldiers in a preventive war is directly targeting soldiers who are morally innocent of the conspiracy to use force. One does not have to be a full adherent to the doctrine of double effect to believe that there is something particularly repugnant about directly targeting the innocent in order to achieve one's political and policy objectives.
The second difficulty with the conditional intention proposal is that it does not seem plausible or viable given the nature of the international political system. Modern states have standing armies, and the strategic planning organs of those armies constantly engage in planning and preparation for military action that is both defensive and offensive in the tactical sense. In the modern world, therefore, active preparation for war can be taken as a given for most states. Because of this, a rule which prohibited an unconditional conspiracy to attack but permitted conditional conspiracies would in effect turn exclusively on the second of Walzer's conditions: the intention of the state in question. The justification of preventive war would be based on a subtle assessment of the intentions of other states: were their plans to attack us conditional on our declared intention to act preventively (p.170) toward them? Or were they planning to attack us anyway, irrespective of our military intentions?
But how could such assessments function as part of a legitimate rule of war? Part of the problem is once again epistemological: what would count as evidence for the intentions of international actors? Who would make the assessments? And how could they be verified or tested in an impartial way? More fundamentally, to attempt to distinguish between conditional and nonconditional intentions to use force seems to misconstrue the nature of the international system. As thinkers from Thomas Hobbes to Hedley Bull have pointed out, the international sphere is an anarchical system, in which mistrust, fear, and hostile intentions are endemic and virtually universal. Except in those cases where formal or informal alliance structures exist, the ill intent of most states, most of the time can and should be assumed. Given this harsh political reality, a rule that justified war on the basis of an active unconditional intention to use force would be quite literally unworkable. It is true that this argument can itself be given a consequentialist interpretation in the claim that any such rule would be likely to lead to a disastrous proliferation of conflicts around the globe.
Yet I have tried to show that the moral objection to preventive war runs deeper than this and stems from considerations that avoid the profound difficulties that bedevil consequentialist theorizing about war. If wars of defense can be justified at all then this must be because those that we kill and maim have done something to make themselves liable to such treatment. It is this fundamental moral constraint that explains the centrality of the conception of self‐defense to the ethics of war. For it is precisely the function of self‐defense in the theory of war to explain how military action can be consistent with this constraint. The constraint itself places an exceptionally high burden of justification on those who would choose recourse to war. But it is one that is required by our conception of humans as beings with moral dignity and basic rights that demand respect. The fact that someone (or his state) may engage in a potential future aggressive attack is not sufficient to discharge this burden. Preventive wars are not morally justified.
(1) I am indebted to many people for careful and helpful criticism and comments particularly Henry Shue, Jeff McMahan, Alan Buchanan, Bob Goodin, members of the Princeton Center for Human Values research seminar especially Charles Beitz, Michael Smith, Stephen Macedo, Philip Petit, and most especially all the other participants of the Changing Character of War workshop including the fellow contributors to this volume.
(2) National Security Strategy of the United States of America, September 2002, p. 2.
(3) The doctrine of preventive war was not the stated legal justification for the war, which instead invoked rights granted under earlier Security Council resolutions dating from the termination of the first Gulf War. Nonetheless, the doctrine of prevention and the reasoning behind it figured strongly in the debate over the war and its public justification.
(4) David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48.
(5) Michael Walzer, Just and Unjust Wars, Basic Books, 3rd edn, 2000, p. 75.
(6) See, e.g. David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, pp. 212–14.
(7) See Ian Brownlie, International Law and the Use of Force by States, Clarendon, Oxford, 1963, p. 275–8; Ian Brownlie, ‘The U.N. Charter and the Use of Force, 1945–1985’, in Cassese A. (ed.), The Current Legal Regulation of the Use of Force, Martinus Nijhoff, Dordrecht, 1986, p. 499; Yoram Dinstein, War, Aggression and Self‐Defense, 4th edn, Cambridge University Press, Cambridge, 1988 pp. 182–7; Michael Byers, ‘Preemptive Self‐Defense: Hegemony, Equality and Strategies of Legal Change’, The Journal of Political Philosophy, vol. 11, no. 2, 2003, 171–90, p. 172 n. and p. 180.
(8) See Michael Byers, ‘preemptive Self‐Defense: Hegemony, Equality and Strategies of Legal Change’, The Journal of Political Philosophy, vol. 11, no. 2, 2003, 171–90, p. 172 n. and p. 180.
(9) Unilateral preventive action in this context refers to the use of force by a state or group of states without Security Council authorization. The Security Council is mandated to authorize preventive force in order to maintain international peace and security under Chapter VII of the UN Charter.
(10) Michael Walzer, Just and Unjust Wars, Basic Books, 1977, p. 77.
(11) Of course a sophisticated form of consequentialism will consider a range of morally relevant consequences beyond the crude metric of lives lost or saved. It would consider, for example, the moral disvalue of the potential subjection of many millions of people to a regime as deeply unjust as Nazism. It may also be the case that genocide is morally worse than the murder of an equivalent number of unrelated people. But even on a sophisticated form of consequentialism it is very difficult to see how a consequentialist could justify World War II given the magnitude of its actual moral costs. Consider how one of the ultimate consequences of the war was to usher in the Soviet Union's brutal half‐century dominion over the states of Eastern Europe.
(12) See Eric Hobsbawn, The Age of Extremes, Michael Joseph, London, 1994, pp. 43–4.
(13) Clausewitz, Carl Von, On War, Michael Howard and Peter Paret, (eds), Princeton University Press, Princeton, 1976, b. 1 ch. 1, sec 20.
(14) My argument concerns a consequentialist assessment of war. I make no claims about international use of force that falls below the threshold of war.
(15) Clausewitz, Carl Von, On War, Michael Howard and Peter Paret (eds), Princeton University Press, Princeton, 1976, b. 1, ch. 1, §3 pp. 75–6 and 77.
(16) See, e.g. Geoffrey Blainey, The Causes of War, Macmillan Press, London, 1988.
(17) As Walter Sinnott‐Armstrong points out in this volume (p. 210), if the effects of war are as indeterminate and unpredictable as I have suggested, this raises the difficult question of how any rational military or political planning is possible. What precisely are planners doing when they assess that going to war would be, for instance, in the national interest? I think that much military planning is subject to similar difficulties, and many policymakers, if they are honest will say as much. But it is also important to note that the criteria according to which most policymakers assess military decisions are substantially narrower in scope than those proposed by consequentialism. For example, judging what is in the national interest is a much narrower (and hence easier) question than judging what will generate best overall long‐term consequences (consider: what is in the national interest need not be in the interest of all or even the majority of the members of the nation). Often the consequences that matter to policymakers are more narrow still—the consequences of a given military action for their own political party, department, or career.
(18) An alternative way of interpreting this argument is to view it not as a critique of consequentialism per se, but as a strong rule‐consequentialist argument for pacifism, or at least a moral presumption against war close to pacifism. Walter Sinnott‐Armstrong considers this possibility in his chapter in this volume (p. 211) and I think that if the epistemological difficulties of consequentialism could be overcome, this would indeed be its likely conclusion.
(19) Cicero, Works, XIV, Loeb Classical Library, London and Cambridge, MA, 1979, 16–17 (Pro Milone, 10–11).
(20) Quoted in Richard Tuck, The Rights of War and Peace: Political Thought and the International Order From Grotius to Kant, Oxford University Press, Oxford, 1999, p. 18.
(21) Michael Walzer, Just and Unjust Wars, Basic Books, 3rd edn, 2000, pp. 76–7.
(22) David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, p. 220.
(23) Luban sometimes presents the contrast between the consequentialist argument for prevention and his consequentialist reading of Walzer's reply as a move from act to rule‐consequentialism (pp. 226 and 227). But in fact both arguments are rule‐consequentialist in structure. For the argument that it is better to fight earlier rather than later can be generalized into a universal rule of prevention.
(24) Michael Walzer, Just and Unjust Wars, p. 77; David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, pp. 223ff.
(25) David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, p. 226–7.
(26) Of course now that these standoffs exist, any norm of prevention would have to be phrased so as not to trigger avoidable wars (at least in the short term until the expected strategic transformation has time to develop). In reality, the danger that wars would be triggered by a norm of prevention in existing strategic standoffs would be limited by the existing proportionality requirement within the rules of war.
(27) David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–248, p. 227.
(28) Michael Byers, ‘preemptive Self‐Defense: Hegemony, Equality and Strategies of Legal Change’, The Journal of Political Philosophy, vol. 11, no. 2, 2003, 171–90, p. 182.
(29) David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, p. 227.
(30) An interesting analogue of this argument is the way in which the mutual acquisition of weapons of mass destruction can actually stabilize relations between rival states. e.g. India and Pakistan have taken more tangible steps to resolve their dispute over Kashmir after they both became nuclear armed.
(31) Luban does gesture at one point to what ‘experience has taught us’, without explaining either which experience he has in mind or how he supposes that it has taught us. (David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, p. 225.)
(32) Bob Goodin has suggested to the author that one might try to obtain a relevant comparable data‐set from economic behavior, e.g. by looking at the incidence of corporate takeovers under different regulatory schemes. But there are a number of problems with this suggestion. The first is that it is difficult to know what would constitute the corporate analogue of a preventive war. The second is that even if one could construct a plausible economic analogue of preventive war, it is highly doubtful whether these observations could be generalized to cases of military action in which human lives and national sovereignty are at stake.
(33) I actually think this radical indeterminacy of consequences is true of the rules of war more generally. For example it is often claimed that, whatever other grounds they may have, the rules of jus in bello are clearly justified on consequentialist grounds. But this is by no means obvious. To take the issue in its most general form, how do we know that having jus in bello rules leads to better consequences than not having any in bello rules at all? It could be that by civilizing the conduct of war, those rules make it easier for states to take their people to war, thus increasing the frequency of wars in the long run. Consider, for example, how the horrific experience of the two world wars leads the major states of Europe to renounce war with each other and embark on the extraordinary collective project of the European Union. Again, I do not want to claim that abandoning the rules of jus in bello would lead to better overall consequences, but simply that this argument is as plausible a speculation as the countervailing interpretation. Furthermore, there does not seem to be any empirical data to which appeal could be made, even in principle to settle the issue one way or the other.
(34) David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, p. 226 (emphasis in original).
(35) As Luban himself points out: ‘For the most part the historical record shows that states fight wars regardless of reigning doctrines of just war, because they regard those wars to be in their interests.’ (David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, p. 236.)
(36) Consider the example of alcohol prohibition. If we assumed high general levels of compliance, then a rule outlawing the consumption of alcohol would probably have beneficial overall consequences, given alcohol's associated health and social costs. But we know that compliance with such a law in the actual world would be very low under any tolerable regime of enforcement. Given this, it is plausible that criminalizing alcohol would have many deleterious effects, such as strengthening and enriching criminal organizations who would take over manufacture and distribution of alcohol, and making it more likely that alcohol users will transition to harder and more dangerous drugs. I take it as obvious that the reduced benefits, and increased costs of such a rule under actual expected compliance conditions is what is relevant to any rule consequentialist assessment of a proposal to prohibit the consumption of alcohol.
(37) The argument that Luban considers for this conclusion is particularistic in form, focusing on US hegemony. The question, he asks, is whether it would be better for the world if the United States enjoyed an exceptional permission to undertake preventive war. His conclusion is that those who would answer this question in the affirmative have not proven their case. But if one's general perspective is rule‐consequentialist (as I believe Luban's position is best interpreted), then this does not seem the right question to ask. For if exceptional permissions are granted to US hegemony, it will be very difficult to deny them to future hegemonys in a similar position of global dominance. The correct rule‐consequentialist question is therefore: is it better for the world if uniquely powerful hegemonic states (whoever they may be) are granted exceptional permissions within the international legal order. It is this question that my argument addresses. (See Luban, David, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, pp. 236–48.)
(38) These problems generate further iterations if we consider the status of an actor with only a small influence on decision‐making within institutional contexts, e.g. a voting citizen, a political commentator, or a moral philosopher. Such situated actors must presumably ask a more complex set of questions still: what effect would my action (voting, putting forward an argument in the public domain) have on the actions of fellow citizens or participants in the political process? What effect would this in turn have on the actions of my state? And thereby on the international system itself? An additional layer of complexity still is added by the fact that the three probable outcomes I have considered are not discrete alternatives. Each is subject to shades of variation, so that the actual range of outcomes may be substantially more variegated and complex.
(39) Michael Byers, ‘preemptive Self‐Defense: Hegemony, Equality and Strategies of Legal Change’, The Journal of Political Philosophy, vol. 11, no. 2, 2003, 171–90, p. 186 (emphasis in the original).
(40) Ibid., 188.
(41) Ibid., 187.
(42) Luban says: ‘Whatever moral force the argument possesses derives from the inherent right of self‐defense—provided, of course that the case for prevention as a legitimate form of self‐defense can be sustained.’ (David Luban, ‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, p. 221.) In his preamble to the National Security Strategy, President Bush asserts the doctrine of prevention ‘as a matter of common sense and self‐defense...’ (National Security Strategy of the United States of America, September 2002, p. ii).
(43) National Security Strategy of the United States of America, September 2002, p. 15
(45) Ibid., v.
(46) Ibid., 15.
(47) See David Rodin, War and Self‐Defense, Oxford University Press, Oxford, 2001, p. 41.
(48) A fourth tradition which I do not discuss here is the Catholic doctrine of double effect initiated by Thomas Aquinas. On this view, it is never permissible for a private individual to intend the death of an assailant in self‐defense, however it is permissible to engage in acts which one foresees will result in the death of assailant as long as their death is not intended either as a means or as an end in itself. Summa Theologica II‐II, Q. 56. This tradition is not prominent in the modern conception of self‐defense, though the double effect doctrine has had a profound influence on broader moral theory.
(49) The term ‘necessity’ can have two meanings in moral and legal theory. The first is the excuse under discussion here. The second is the condition found inter alia within just war theory and the law of self‐defense which requires that an action may only be undertaken if there is no less costly or harmful way of achieving the same result.
(50) Cicero, Pro Milone (10–11) in Works, XIV, London and Cambridge, Mass., Loeb Classical Library, 1979, 16–17.
(51) George Fletcher, Rethinking Criminal Law, Little, Brown, Boston, MA, 1978, p. 808.
(52) Aristotle, Nicomachean Ethics, Crisp, R. (tr.), Cambridge: Cambridge University Press, 2000, 1110a.
(53) Law and morality seem only partially persuaded by this line of thought. The excuse of necessity is available under certain circumstances, but it does not cover all cases in which one is in fear for one's own life. It does not cover for instance the use of lethal defensive force against innocent bystanders. Nor is duress an excuse for certain kinds of crime, e.g. treason (J. C. Smith and B. Hogan, Criminal Law, Butterworths, London, 1992, pp. 238 and 240. See also R. Stone, Offenses Against the Person, Cavendish, London, 1999, p. 66).
(54) Quoted in Yoram Dinstein, War, Aggression and Self‐Defense, p. 249.
(55) Some philosophers have spoken of the ‘artificial fear’ that can be experienced by states as political entities. See T. Airakson, ‘The Whiteness of the Whale: Thomas Hobbes and a Paradox of War and Fear’ in M. A. Bertman (ed.), Hobbes: War Among Nations, Avebury Press, Aldershot, UK, 1989.
(56) It may indeed be the necessity model of self‐defense that underlies Luban's somewhat counter‐intuitive claim that ‘a state can legitimately defend itself against an armed attack even if that attack is lawful’ (‘Preventive War’, Philosophy and Public Affairs, vol. 32, no. 3, 2004, 207–48, pp. 211–12 note.).
(57) Quoted in Yoram Dinstein, War, Aggression and Self‐Defense, p. 249.
(58) National Security Strategy of the United States of America, September 2002, p. 16.
(59) David Rodin, War and Self‐Defense, Oxford, Oxford University Press, 2002, ch. 4.
(60) It should be noted that there is disagreement among those who defend a rights‐based view of self‐defense as to what kind of wrongdoing is required to ground liability to harm in self‐defense. Some theorists have argued that one can use defensive force against an ‘innocent aggressor’ or ‘innocent threat’ as long as the aggression or threat is ‘objectively wrong’, that is to say excused, but not justified (see Judith Jarvis Thomson, ‘Self‐Defense and Rights’, The Lindley Lecture, Lawrence University of Kansas Publications, 1976, Suzanne Uniacke, Permissible Killing, The Self‐Defense Justification of Homicide, Cambridge University Press, Cambridge, 1994.). My own view is that ‘objective wrongdoing’ is not a sufficient ground for self‐defense, and that the person against whom one uses defensive force must have some measure of responsibility or culpability for the wrongful action defended against (see David Rodin, War and Self‐Defense, Oxford University Press, Oxford, 2002, ch. 4, esp. pp. 83ff.). However, the argument in the following sections does not rely upon this thesis and is in fact agnostic between the two views. One may therefore read reference to ‘wrongdoing’ in the following sections as including the broader class of ‘objectively wrong’ but potentially excused actions.
(61) Jeff McMahan pursues a similar line of argument to that employed here in Jeff McMahan, ‘Preventive War and Killing the Innocent’, Richard Sorabji and David Rodin (eds), The Ethics of War: Shared Problems in Different Traditions, Ashgate, London, 2005.
(62) Michael Walzer, Just and Unjust Wars, Basic Books, 3rd edn, 2000, p. 81. Interestingly, Walzer's language of ‘sufficient threat’ is adopted (whether consciously or not) by the National Security Strategy (see NSS, p. 15).
(63) NSS, p. 15.
(64) Ibid., 16.
(65) David Luban, ‘Preventive War and Human Rights’, p. 193.
(66) Ibid. pp. 190–2. This formulation of the objection has also been strongly articulated in discussion by Henry Shue.
(67) Ibid., p. 194.
(68) Ibid., see esp, pp. 175–80, ‘Standing on the Threshold of Pacifism’.