Preemptive War and the Epistemological Dimension of the Morality of War


by Randall R. Dipert

C.S. Peirce Professor of American Philosophy

University at Buffalo, NY


I.  Introduction:  Not Just a Matter of Language


Preemptive or preventive war is in a class of hard cases for the traditional philosophy of war.  All appear to violate one or more of the Just War criteria.  Preemption appears to violate Last Resort and Just Cause.  (I will use "preemption" also to cover cases in which an enemy's attack is not known to be imminent, which others have called preventive war.)   A deterrent strategy of massive nuclear retaliation appears to violate Chance of Success and Proportionality.  Humanitarian Intervention appears to violate Just Cause, especially insofar as sovereignty is salient.[1]  Within mathematical game theory, tit-for-tat, which sometimes will violate Chance of Success and possibly other criteria, has been proven to be one of the few successful and stable strategies of prolonged conflict in situations strongly resembling the actual world.  While Deterrence, Humanitarian Intervention, and Tit-for-Tat have received a great deal of discussion in recent literature, the philosophical literature on preemption is remarkably and deplorably scant—almost non-existent c. 2003.[2]

      We cannot hope to solve all of these issues here.  In this paper I am going to make the following points.  (I) First, and contrary to some pronouncements, there is a great deal of support for the claim that some preemptive wars, even in the absence of an imminent attack from an enemy, are morally justified.[3]  There is also a large and important literature that rejects preemptive and preventive wars of all sorts.[4]  (II) Second, a great part of the difficulty of preemptive war arises because of the widespread assumption, without much argument, that the Just War criteria are not only jointly sufficient but are also jointly necessary.  (III) Third, a neglected element in the philosophy of war is what I call the epistemological dimension: the extent to which a criterion can be reasonably known to be fulfilled.  Considering this epistemological dimension both makes preemptive war morally justified in some comparatively rare cases and grounds a special condition that is usually attached to preemptive war by those who consider it justified, namely what I will call an Urgency Condition.  (It also grounds a weak "internationalist" condition for moral preemptive war, a rather surprising result.)  (IV) Finally, I will argue that preemptive war is restricted by a universalist metaprinciple with similarity to principles of Kant and R.M. Hare. This metaprinciple, which governs when one ought or prudentially should engage in preemptive war, also grounds the Nuclear Deterrence and Tit-for-Tat, as well as some instances of Humanitarian Intervention.  It requires that one's policy for engaging in preemptive war is such that its practice by every nation, in light of a priori facts of game theory, and by better established empirical facts of human history and human nature, is conducive to fewer wars and less loss of human life than is a world in which they are not universally practiced.

      I agree with many critics who regret the widespread recent use "preemptive" war to describe what had more often carefully been called "preventive" or "anticipatory" war.  This is especially a problem in the National Security Strategy (2002) and the discussion that has followed it.    Unfortunately, I think that criticism of this verbal error has often resulted in a simplistic and impoverished philosophical discussion, even if highly respected scholars (e.g., Tuck 1999) have used "preemptive war" in a broad sense. A preemptive war, narrowly and precisely speaking, is an attack, a first-strike and often a surprise attack, on another nation which is itself about to attack, and in which waiting may seriously imperil the attacked nation.  A morally justified preemptive war requires the danger of an imminent attack—likely (all but certain) and very soon, typically a matter of hours or days.  A preventive or anticipatory war, although both are clumsy terms, is an attack launched either far in advance of an expected aggression, or for which the time of the expected aggression is unknown.[5] 

      The usual logic of critics of the recent colloquial use of preemptive war goes something like this.  A war such as the recent war in Iraq is not a real preemptive war in the strict sense, but is instead a preventive or anticipatory war.  All preventive or anticipatory wars are morally unjustified, and the use of the term "preemptive" is a clumsy attempt to disguise this fact.[6]  Implicit in this criticism is also that it is "morally obvious" or "widely argued" in the existing literature that preventive or anticipatory war is unjustified.  This last point however is badly mistaken.  No lesser authorities than Hugo Grotius, Kant, a whole list of authors considered in Tuck 1999, as well the distinguished contemporary author Michael Walzer[7]  have argued that some anticipatory or preventive wars are morally justified.  John Rawls, in one of his last major essays, The Law of Peoples (1999), makes careful remarks that avoid requiring actual enemy attack for morally permissible war.[8]  As I observed above, there is little other traditional philosophical literature clearly arguing that such wars are never justified and no more extensive, recent treatment c. spring 2003  than Michael Walzer's.[9]    And regardless of how religious and philosophical authorities line up,[10] we should ask what exactly the argument is that shows that all preventive or anticipatory wars are morally unjustified.    In what follows I will follow popular usage in including what have been called preventive or anticipatory wars—wars without knowledge imminent aggression—as varieties of preemptive war. I will focus on cases in which there is no knowledge of imminent aggression, but at most knowledge of non-imminent or indefinitely timed future aggression.


II.  The Problem of Preemptive War without Knowledge of Imminent Aggression


In this paper, I will argue that some preemptive wars in which there is no danger of imminent attack are morally justified.  Going to war against a nation is not "punishment" of that nation, its soldiers, or civilians for misdeeds.  In the modern conception of the morality of war, our military action is exclusively justified by the threat another nation poses to our citizens in the future.[11]   We can see this since: (a) This condition of threat guides jus in bello moral constraints once war starts, e.g., the fact that those who do not pose a threat may not themselves be intentionally harmed, such as civilians and disarmed  soldiers.  (b)  If the nation should withdraw and no longer constitute a threat, such as when there is a sudden, permanent change of government or the military ceases to obey its regime,  then our continued attack would be unjustified.   And (c) a conception of the attacking enemy's motive in attacking us is inessential in the articulation of our right to counterattack.[12]  A conception of motive is always necessary for the more usual moral assessment of an action and for punishment.     

       As is well-known, there are three forms of threat that another nation may pose, and that have been argued morally to justify military action as a Just Cause:

                       (1)  An actual large-scale attack, such as "armed attack" in the sense

                              of Article 51 of the U.N. Charter, which allows self-defense.

      Observe however that it is the future enemy attacks that pose the threat

      and justify military actions; the past attack, I would claim, is simply

      strong evidence for the imminent (further) danger to our civilians.[13]

      At least as regards loss of life,  there is no repairing through counter-

      attack the damage the first attack.

                       (2)  The threat of imminent large-scale attack, that is of a first-strike or

                              even a surprise attack.  Wars launched with such a justification have

                              traditionally been called "preemptive."

                       (3)  A threat of (a) distant future, or (b) indefinite, but

      all-but-certain large-scale attack.  These two separate classes of cases

      have been called "preventive" or "anticipatory" wars.

     I argue that the primary moral considerations in all of these scenarios are just two factors, namely the extent of the threat  (the "Threat Condition") and the degree to which this threat can be known or reasonably appraised (the "Knowledge Condition").  A third factor, the "Urgency Condition,"  giving a reason to attack now rather than waiting, has frequently been added, such as by Walzer (Just and Unjust Wars, the third condition for legitimate first-strikes p. 81), but I will later  argue shortly that the Urgency Condition is morally derivative from the second, Knowledge Condition.  In order to focus on the most interesting cases of preemption, I will assume that all reasonable nonmilitary options have been exhausted (Last Resort), and that Proportionality, Success, Proper Authority, etc., are not an issue.  


III.  The Metaphysics and Epistemology of Relevant Ethical Facts


    The basic question is whether any such preemptive war is morally justified,  especially if an enemy's attack or aggression is not imminent.  I will argue that there clearly are cases in which preemptive war, without a danger of imminent attack, is morally justified.  My main argument is that the widespread presumption against preemptive war arises because we have ignored what I will call the epistemological dimensions of the criteria for a morally permissible war.   I will argue that almost all previous discussions of the moral justification of war have discussed solely what I would call metaphysical dimensions of conditions for war, that is, what is the case and what is not case, as opposed to what can be known, is reasonable to believe, and so on.  Once epistemological dimensions are introduced, it becomes obvious, I will argue, that some preemptive wars without imminence are justified, if any wars are justified at all. 

     There are two other issues involved in the discussion of preemptive war that I would also like to disentangle.  First, there is widespread agreement, at least among those who are not absolute pacifists, that something like the traditional Just War conditions are sufficient to justify the moral permissibility of war.  However, there is scant literature, few or no intuitions, and no widely accepted moral foundations that make meeting the Just War conditions necessary for the moral permissibility of war.  Part of the widespread moral presumption against the morality of preemptive war without imminence arises because of a confusion of sufficient with necessary conditions for the moral permissibility of war.[14]  Those who even consider that Humanitarian Intervention might be permissible are admitting the lack of argument that Just War conditions are necessary criteria.   Second, and this is perhaps where the flavor of the English words "just" and "justice" invite confusion, there is no sense in which an enemy deserves to be attacked.  It is not an issue of justice as we ordinarily consider it.  Rather, the only source for the moral permissibility of war lies in a nation's right to exist (or more precisely, in its citizens' rights to existence and to a polito-economic order of their choosing), and in a threat to that existence, not in any theory of international "punishment" that applies a model of rules of law, order and justice to the order among nations.  Even if the misleading model of intra-national policing and justice were to be applied to the order among nations, then it would quite clearly support, not go against, some sort of preemptive war--as we will see.

      We can sharpen the description of one of the most interesting and telling example of preemptive war.  Suppose a nation cannot determine in advance when a devastating and unjust attack will occur, but knows that it will occur sometime.  Is it morally doomed first to absorb the attack before it may do anything in its own defense?  If a nation and its people have any right to peaceful existence, and this right places highly in an order of rights,[15] this seems unlikely.  Consequently, I would argue that there is clearly some moral right to preemptive attack.   An opponent of this view might argue that we can in fact never be in the position that we know an attack will come but not know, or be able to find out, when.  This criticism is far-fetched, although the justifiability of preemptive war turns on the fairly rare cases where we would be strongly justified in believing that a distant threat obtains.

         Presumably a preemptive attack is justified if all of these conditions are met: (1) the threat posed to the existence of the nation about to launch a preemptive attack is substantial, (2) there will eventually be an attack, (3) the attack, when it comes, is unjustified, (4) there is seriously increased risk to the nation, or its military forces, by waiting to launch an attack after an enemy's first-strike, and (5.-5.4) beliefs that conditions (1)-(4) hold are each strongly (and objectively, not subjectively) justified.  Conditions (1)-(4) are metaphysical conditions on facts, (5.1)-(5.4) are epistemological conditions on their justification to the central authority.[16]  From the point of view of actual decision-making, 5.1-5.4 are sufficient conditions for the moral rightness of a decision to engage in preemptive war; the moral justifiability of the action requires all of (1)-(4) and (5.1)-(5.4).   We must remember that these describe the bare moral permissibility of preemptive war, or the decision to go to war, and not the wisdom, other prudential desirability, or obligation to go to war.     


IV.  Thought Experiments and Intuitions


        Although they are not conclusive, I wish to argue further for the intuitive plausibility of some right to preemptive attack by sketching two fictional cases concerning not nations but conflict and morality between individual person.  These provide a more fuller description of the "telling case" of devasating attack by an enemy, and thus better grip out moral intuitions and sympathies.   First, imagine an implacably belligerent neighbor, who has attacked you, your family and property on prior occasions.  He has sworn to do so again.  For whatever reason, no third party (such as police) are likely to aid in your defense or even punish the neighbor and thus deter the neighbor.[17]  You have knowledge, say from conversations with the UPS delivery man, that your neighbor has received a devastating weapon—such as a mortar or RPG—that is certainly intended for an attack on you, for which you have no available countermeasures, and is such that once this attack begins, you and your family will certainly be immediately destroyed.  You do not know when this attack will come—tomorrow, next week, next month, etc.  Because of the neighbor's own defenses, there is no way to remove or disable this weapon without using deadly force that is likely to kill the neighbor.   I would argue that you do have the right to attack your neighbor first, although this right hinges on the extent and reasonableness of your strongly justified belief in the threat.[18]

       In a different case, let us imagine that while traveling you come into contact with individuals whom you have overwhelming evidence to believe are terrorists intent on killing hundreds or thousands of individuals.  You do not know when they will launch their plot and the terrorists are not now breaking the law or doing anything immoral (say, the boxcutters they are carrying are not forbidden).  We can imagine separate cases in which you are a law enforcement agent, and in which you are simply a civilian.  Once the terrorists begin to execute their plot, it will be too late; instead, you must act now.   Furthermore, let us suppose that the only way you can stop them is by using deadly force.  I would argue that if you are civilian, you are permitted to interfere in the terrorists' plan, and if you are a law enforcement agent, you may be obligated to do so.  Obviously again the degree of required evidence of the facts for this moral permission or obligation is very high.  This discussion raises a question of whether policing agencies should be in the business of apprehending criminals after the fact or, in some cases, of preventing the commission of a crime in the first place;  it also raises questions about the legitimacy and desirability of conspiracy and "intent" laws.[19]  These issues in turn raise difficult questions about whether mandates and statutes can be crafted such that they justify the prior restraining only of (future) guilty parties.

    My question however is strictly the moral one of whether, once you know certain facts with whatever degree of reasonableness and certainty is required, you may use force on an individual who, ignoring mental states and verbal indications thereof, has not yet done anything wrong.  I think it is obvious that there exist such cases, even though the standards for knowledge are very high and thus such cases are very rare.

     Even more telling would be a case in which a nation knew, with a certainty that is usually only accorded mathematical proofs or even God's foreknowledge, that another nation was preparing to attack in, say, two years.[20]  Loss of life and property would be considerably lessened if we attacked now, rather than waiting two years or even one.  There exists no other way to avoid the enemy's attack.  From this God's-eye perspective, especially if there is increasing and unjust harm by delaying action, preemptive action in advance of a "clear and present" danger would not only be permissible but likely obligatory.  If the clarity of the danger is sufficiently great, then whether or not it is "present," that is imminent, is irrelevant.[21]  

     But human beings typically lack such knowledge.  As we wait to act, new information can arrive or old information can be revised; we can even make new inferences from the information we had, or discover the fallaciousness of our old inferences.  This is ultimately the source of the Urgency Condition and the "present" in the phrase "clear and present [danger]":  our fallibility almost always decreases with time, and we have more rather than less information and can make more leisurely and careful inferences.  Our human fallibility entails that the threshold for preemptive attack increases with the amount of time in advance of the expected aggression against us, since the risk of wrong assessment of such calculations as Just Cause and Chance of Success is typically proportional to the amount of time we have to gather more information or correct earlier factual or judgment errors.  We should realize, however, that requirements of "imminence" or "clear and present danger" are wise rules of thumb, but are themselves grounded in more detailed epistemological considerations.

      In establishing the moral justifiability of going to war, the only factor that diminishes this effect of time-increased fallibility is the amount of harm that will befall us if we wait.  Observe however that Urgency is ultimately not a separate condition, but a factor that is dependent on the epistemological status of our beliefs, combined with the general likelihood of generally decreasing fallibility.  Very high levels of objective certainty of a future attack on us would require only small amounts of harm in waiting (for God, no harm to waiting at all), in order to justify a preemptive attack by us, while small objective uncertainty in our judgments of future attack would require comparatively greater amounts of harm in waiting to attack in order to justify a preemptive attack.  In the calculus of justified war, there may be other interactions as well:  the likelihood of our total destruction in an enemy first-strike would require slightly less objective certainty than would the likelihood of at worst losing a city.[22]  Very roughly, the justifiability in going to war, J, is positively proportion to the probability of harm of an enemy surprise attack, P, the amount of this harm, H, and the danger in waiting, W, crudely,

J µ P ´ H ´ W—although to be "minimally reasonably justifiable" each of P, H, and W must be above a certain threshold.[23] 


V.  Historical Examples


One particular subset of cases is I think of interest: cases in which the development of particular technology increases greatly the applicability of the conditions for justified preemptive war listed earlier: (1) Substantial Threat, (4) Increased Risk in Waiting, and also (5.1)-(5.4) the degree of justification of each of (1)-(4), or what we might describe altogether as Intelligence.  Roughly, the implacable belligerence was already present, but the other nation is about to develop or deploy a first-strike weapon.  While they all marked incredible advance in weapons, and changed the nature of warfare, such developments as firearms, artillery, automatic weapons, armored sea and land vehicles, poisonous gases were all initially of limited effectiveness, were mainly of use only on front lines (rather than devastating or threatening the whole military or economic infrastructure of a nation), and countermeasures were rapidly developed and these weapons were quickly replied to in kind.  Furthermore, few if any major nations in the history of warfare have possessed the military or technological means effectively to stop the development or deployment of such weapons by a preemptive war, or possessed reliable information about the enemy's development of them or about these weapons' effectiveness in advance of their use.  So the necessary conditions for the moral justification of non-imminent preemption have been extremely rare.  The possession of, effort to develop, or ability and likelihood secretly to develop nuclear weapons and a long-distance method of delivery (by missile or smuggling into the country), with undeterrable motives (e.g., ideological or religious), and in the absence of countermeasures, together with sufficient evidence to this effect, almost certainly does meet these conditions, at least unders some circumstances.   

     As for condition (5), the epistemic threshold in justifiably believing (1)-(4), observe how technology has improved this.[24]  While human intelligence is as variable as ever, electronic surveillance of communications, and visual/aerial, overhead radar, infrared, magnetic and motion sensors are considerably improved.  We can surmise that in many cases of war before modern communication, the belief by a nation's central, war-authorizing command that the nation had been attacked at a [25]distant locale was often comparatively unjustified.   It was a hand-carried but counterfeitable message on paper, word of mouth across several people, terse semaphore or telegraphic message, and so on.    In some cases, we today will be justified in believing that a nation will attack us in the future with greater epistemic justification than nations of the past had to believe that they had already been attacked.   Consequently, if wars of the past were morally justified despite a lack of complete certainty that they had been attacked, then present-day wars are justified in which we have a comparable degree of evidence that an enemy will attack us in the future.   

     There are three relevant historical cases that are readily apparent and a fourth that is less likely to come to mind.  Each evokes an intuition that is widely shared.  The three more readily apparent cases are (1) the Soviet acquisition of nuclear weapons in 1948, (2) the Soviet acquisition of thermonuclear ("hydrogen") weapons in 1953, and (3) the Israeli destruction in 1981 by F-15's and F-16's of a plutonium-producing nuclear reactor in Iraq.  An attack in 1948 was not seriously considered.  An attack on the Soviet Union was quickly rejected by Eisenhower in 1953—although the main obstacle seems to have been the feasibility of removing permanently the threat in one attack i.e., in Just War Theory the "Probability of Success" condition was not met.   Eisenhower himself stood almost alone against his civilian and military advisors in believing what has in fact proven to be the case, namely that the Soviets were not implacably belligerent and were deterrable; he was also of the general opinion, in WW II and during his Presidency (in Korea, Indochina, Lebanon, or in removing a Soviet atomic threat) that air power alone could not achieve strategic objectives—and that land wars against substantial foes were too costly for all but the worst evils (i.e., in violating Proportionality).

     The limited Israeli attack on Iraq has seemed to most observers to be justified (e.g., Walzer)[26] —although the U.S.and U.N. at the time formally condemned the attack and the Israeli policy.[27]  By my own criteria of moral permissibility, this would depend on the likelihood, given reliable intelligence estimates, of an eventual unjust nuclear attack by Iraq on Israel.  It would also depend on the state of development of the reactor—such as how many civilians were at what level of risk  from such an attack earlier as opposed to later—and on a reasonable estimation of further escalation.

     The fourth and most interesting case, which might not immediately come to mind as a case of preemptive war, is the Cuban Missile Crisis of 1962.[28]  The general history is well-known, but what is worthwhile to point out here is that the Soviet missiles in Cuba by no means posed an imminent threat (indeed, we thought they had not even been fitted with nuclear warheads), but at most a growing threat.  Intelligence was excellent, in constituting sufficient evidence of e.g., (1) Substantial Threat, and  in the intelligence's reliability.[29]   However, the U.S. committed what are universally regarded as "acts of war" (that are typically taken morally to justify counterattack or even war), namely a blockade, interception of foreign freighters on the high seas, and threatening to fire upon, and indeed actually shooting canon "over the bow" of, an unarmed foreign ship.   To the extent one believes that the Kennedy administration was justified—and most apparently do—then this is an acknowledgement of the justification of preemptive "acts of" war where the threat is not imminent.  I myself am not sure whether the U.S. blockade was morally justified, since the Soviets were otherwise rationally deterrable—even if Castro, in much later conversation with McNamara, indicated he was not.  Furthermore, diplomatic possibilities had not been exhausted, such as the later offer to withdraw similar middle-distance missiles from Turkey.[30]  Within its larger historical context, it is possible that principles such as the Monroe Doctrine have moral status because they tend to diminish the frequency and severity of certain kinds of war.[31]  Furthermore, it is also possible that Kennedy and his civilian administration were justified since the U.S. military at that point (notoriously Curtis LeMay) would not have settled for anything less than some use of force, and this was the least force the U.S. military would abide.  (What this shows is I believe the intermittent fragility of civilian control of the military, even in the U.S. and which even Eisenhower himself as president decried.)    


VI.  The Epistemic  Threshold for Preemptive Attacks

We have seen a number of reasons why some preemptive wars, even in advance of an enemy's imminent attack, are morally justified.  Nevertheless, this justification hinges on what I have called an epistemological threshold.  This threshold is the minimum amount of "objective certainty" about the enemy's intentions, bellicosity, and present and future military resources necessary to justify preemptive war.  It is not the subjective certainty of feeling strongly about the extent of evidence for these factors.  To be morally justified, one must have, and appreciate, extensive evidence for these factors and the other usual criteria for Just War except Just Cause; one must lack substantial evidence that goes against one of these factors, after a reasonable effort to acquire such evidence.  A "second order" objective certainty is also necessary: one must be justified in believing that one's past record of judging intentions, resources and so on, from the information sources one is now using (e.g., satellite imagery), has usually been correct.

      It may be instructive here to reflect on the 2003 Iraq War.[32]  The fact that Iraq actually did not have weapons of mass destruction, nor did it even have quickly constructable technology and facilities to produce them, shows that the Bush administration did not have knowledge of the weapons or facilities.  It does not however alone entail that it was not objectively certain to the extent required by the epistemic threshold criterion for preemptive war.  In fact I believe that it was highly rational to believe, objectively certain, and in Grotius' words "morally certain,"  that Iraq had chemical weapons despite its falsehood.  (This is a consequence of permitting defeasible or non-absolutely-certain justification or warrant for knowledge that is now almost universally accepted by epistemologists.)  This is debatable, to be sure.  However, I am not convinced that having chemical weapons of the kind they supposedly possessed alone posed a sufficient threat to justify preemptive war.  (The case for morally justifying preemptive war with regard to biological or nuclear weapons almost certainly did not meet the epistemic threshold.)  This is not to suggest that there were not other morally sufficient reasons, or that there might be some accumulative effect of arguments that are separately, in various respects, weak.  Grotius, for one, diminishes the importance of intent, and allows one to change intents in mid-war, while retaining its morally justified character.  

     Especially in the recent 2003 Iraq War, there was a constant refrain for international approval of the coalition efforts.[33]  Intuitively, some international assent, especially by sympathetic nations if not the Security Council of the U.N., is desirable.  Yet it is very difficult to see how this fits into the moral theory of the permissibility of war.  If a nation is somehow morally justified in going to war, then consultation or approval is not necessary—as suggested by the U.N. Charter itself.  Likewise, approval by the U.N. Security Council is not alone sufficient to render a military action morally justified: if it weren't independently moral, then adding to it U.N. or international approval of any sort does not make it morally justified.   Indeed, there is a vicious circularity for excessive internationalists:  what principles of just war are to be used in international discussions of whether to approve a nation's military action?  These cannot be just the international approval itself since that is the outcome of deliberation.  The principles of morally justified war, to be more than procedural,  must be wholly independent of international approval and agreement.

    However, this reasoning, contrary to our intuitions, seems to leave no place at all for internationalism in the moral justification of war (at least as regards its moral permissibility).  I would propose that considering the epistemological dimension of morally justified war does give a proper place to our internationalistic inclinations.  As is now all too well known, political discussions of the conditions of just war are prone to being blinded by already firm geopolitical worldviews, as well as by past political rhetoric that tend to chain politicians to certain views for the sake of "consistency."  The facts of the case, such as intelligence on WMD's, are likewise prone to a certain institutional conformist tendencies—and this tendency was well-known long before the supposed influences of neoconservatives on the U.S., and apparently also on foreign intelligence services.  For example, when critical policy decisions rest on intelligence, the legendary Sherman Kent,[34] proposes that we critically examine existing intelligence, and apply in my terminology "second order" principles, explicitly attaching the probability that various truths are mistaken, based on past incidents of the type of information from such sources.

     International approval, plays a role in the moral justification of war primarily in this epistemological dimension.  I do not think approval of the oddly chosen U.N. Security Council[35] is necessary for a morally justified war, even if it is desirable and should often be sought (for various prudential reasons).  Rather, the underlying principle is something like this: a failure to persuade numerous like-minded nations of both the relevant facts (e.g., the existence of WMD's), when these nations preferably have some independent intelligence capability, or failure to persuade them of the relevant moral principle  embodied in a policy (e.g., that if a nation is as chronically belligerent as Iraq, and has such a WMD capacity, then it can be preemptively attacked), is strong evidence against one's having met the epistemological threshold for preemptive war.  In the recent situation, the opposition of Russia and France, especially Germany and Mexico, and the unenthusiastic acquiescence of China gave prima facie evidence against having met this threshold; the support of the UK, Italy, Spain, and Poland were however probably sufficient to meet my condition.   In any case, it is in this epistemological dimension of the philosophy of war, and not anywhere else, that international or international-organization approval plays a role in moral justification.[36]

     It is difficult to say much about what precisely this epistemic threshold is.  It need not be "warrant" as it is used by epistemologists when discussing conditions for knowledge.[37] Roughly, I think that the evidence at hand both for bellicosity and for the enemy's possession of military resources constituting, or soon to constitute, a threat (and of their probable offensive nature) must be overwhelming and "all but certain."  I do not think that "manifest preparations" for an attack (in Walzer's terms) are necessary, whatever this means.[38]  Additionally, our second-order assessment of this evidence must be such that we have good reason to believe that it constitutes good evidence: this source has not mislead us in the past, etc.  A second-order assessment is our reasonable estimate of the probability of evidence for our first-order assessment of harm, bellicosity, etc., being correct.  The military resources must be such that they are likely, if used in a first-strike, to endanger our nation itself or to pose a severe threat of incapacitating our own military resources.  It seems to me—although I have not studied this matter at all thoroughly—that chemical and biological weapons are indeed terrifying, but are unlikely to be serious in this precise sense.  Their dispersal problems as well as the existence of countermeasures tend to lessen their military danger.  Nuclear weapons, including dirty bombs, are almost certainly in the "severe threat" category. 

      Several factors raise and lower this threshold.  One is the seriousness of the threat.  Another is the amount of time until these military resources pose this threat.  Still another is a kind of proportionality: minimizing civilian and even military deaths.  The epistemic threshold never gets so low that, for example, one may launch a preemptive war based on evidence of a nation's bellicosity or resources that is "somewhat likely."




     One final but crucial consideration in preemptive war is the application of my proposed "metaprinciple."  The sufficient conditions for the permissibility of preemptive attack have already been discussed.  I see no strong argument to suppose that these, together with more usual Just War criteria, exhaust the sets of sufficient conditions and thus that their disjunction is a necessary condition of war.  However, it remains to be considered when one should launch a preemptive war.  That is, when is it prudentially or morally wise to do so, up through when it is obligatory to do so? At this second stage of consideration I propose this.  Sketch in general terms the circumstances in which your nation finds itself, with respect to a potential belligerent—both in terms of strength of evidence, amount of threat, urgency, and so on.  Suppose now that every nation were always to launch preemptive wars in those circumstances.  How much conflict would likely result, including a diminution or increase of war-provoking conflicts over the long term?

     One might think that this principle would give little guidance in recommending preemptive wars, but au contraire.  Let us suppose that John Rawls, following Raymond Aron and others, is correct in claiming that democratic states ("liberal constitutional democracies") have very few except legitimate reasons to go to war, and consequently rarely do go to war for "bad" reasons.[39]  Some wars might still occur because of epistemic mistakes or (legitimate) mutual fear and distrust trust—something Rawls seems not to consider.  Let us further suppose that this general level of warfare in a region or in the world gradually decreases in those places where there exist nothing but constitutional democracies.  (Rawls speaks of "lasting peace" but that may not be possible and such utopianism is unnecessary for the argument.) Let us further suppose (the biggest "if") that democracy can be imposed, or the conditions for democracy can be created, by the correct occupation force.  Then there are circumstances in which, if the conditions for the permissibility of preemptive of war are met, then preemptive war is further recommended by this second principle.

     There is an interesting question here, beyond philosophical considerations, about whether a nation should formulate policies of exactly what conditions will, and what conditions will not, trigger preemptive war.  There is even more of a question about whether one should publicly announce this policy—even in a democracy.[40]   But there is another and telling side of this coin:  what if we should have a policy of never engaging in preemptive war or, worse still, announce our policy of never engaging in any preemptive or preventive war?  Here I think we are encouraging a hostile enemy to prepare an offensive, including weapons development, right up an actual attack.  If there do exist, or can possibly exist truly devastating weapons, this is to invite annihilation.  Even a small nuclear power with ballistic missiles (perhaps positioning missiles on ocean freighters on the high seas) would be free to inflict devastating attacks.  While large, stable countries such as China and the former U.S.S.R., have historically been deterred by the policy of massive nuclear retaliation, it is unlikely that all nuclear nations with ballistic missiles (including terrorist organizations), will remain deterrable.  I believe that such a policy of banning or foreswearing preemption would almost certainly result in more, rather than fewer, wars and deaths, because it would embolden more nations to believe that they could succeed in an unjust war, especially in ideological wars whose "success" consists simply in inflicting harm on the West, democracy, capitalism, etc.

     To announce a policy of rejecting any preemptive or preventive war is thus almost certainly mistaken and violates my second principle insofar as it increases possible threats.[41]  The rare and careful use by us of restricted preemptive war, under unspecified conditions, in the world we are likely to have for centuries—without, for example, militarily dominant international organizations willing to punish with force the illegitimate use of force—is actually likely to make the world more safe.[42]  This is not a conclusion that I am especially happy with, but it is my conclusion.



[1] Earlier instances probably violate the contract between volunteer soldier and nation, although that is not a matter for the philosophy of war.  Since events in Bosnia and Kosovo (and court cases contesting soldiers obligations to take part in UN peacekeeping missions), volunteering soldiers have however received fair notice of the extension of their duties beyond national defense strictly construed.

[2] Especially before the U.S. National Security Strategy of September 2002 and the Iraqi War of spring 2003.  An exhaustive search in Philosopher's Index in early 2003 for substantive references to "preemptive or "preventive"  war,  as well as other variants, brought just two articles—plus Walzer's Just and Unjust Wars, which I knew of but was not referenced.   Philosopher's Index is an index of all major and many minor philosophical journals in English, as well as major journals in foreign languages, 1940 to 2003.   The American Philosophical Association in a resolution passed overwhelmingly in spring of 2003 stated ""Members of the Eastern Division of the American Philosophical Association express our serious doubts about the morality, legality and prudence of a war against Iraq led by the United States.  Both just war theory and international law say that states may resort to war only in self-defense." (Eastern Division: 1,202 in favor, 263 opposed; Pacific Division 283 in favor, 78 opposed).  This was a peculiarly definitive position for professional philosophers to take, given the lack of literature (and apparently, lack of extensive and recent thought about the issue), and also given that figures as well known as Grotius and Walzer have permitted preemptive war. (See later remarks about Grotius and Walzer for references.)
   In the Ethics of War and Peace (ed. by Terry Nardin, Princeton University Press), there is no sustained discussion of preemptive war, and although author David Mapel approves permitting preemptive and even preventive war (pp. 59-60), his defense is from a Realistic perspective that rejects a role for moral thinking in the philosophy of war (p. 60).  Jeff McMahan permits both in rare cases (p. 85) but gives what I  later call an epistemological presumption against it, "a state's ability to understand the intentions and predict the future behavior of its adversaries is notoriously weak."  This treats epistemological criteria in a piecemeal way and it is difficult to see why states' knowledge of (i)  adversaries'  (ii) intentions is a special case, or why this fallibalist constraint does not also apply to other Just War criteria or to individual psychology.

[3] A major theme of Richard Tuck's historico-political essay, The Rights of War: Political Thought and the International Order from Grotius to Kant (Oxford University Press, 1999) is in fact a survey of how many political philosophers in the Modern era endorsed versions of preemptive war, why they did so, and the relationship of this issue to wider issues in ethics and political philosophy. See especially pp. 18-31.

[4] Extreme early critics of preemptive war include Luis de Molina (1593) and other Spanish Thomists.  See Tuck 1999: 51f.

[5] There is a third kind of justification for an attack, a "balance of power" attack in which one seeks to preserve a status quo of military power in a region, even when a enemy's future attack is not certain, or even likely, and in which the enemy's implicit threat of the use of that power will merely result in undesirable economic and social consequences.  I will not consider it here, but suspect that such an attack is never morally justified.  Emmerich Vatel (1751) , and perhaps Cicero, are two of the few philosophical authors to endorse this criteria (see Tuck pp. 191-5), although its endorsement is more common within Realism and Realpolitik.

[6] See, for example, E.g.,   Jeffrey Record, "The Bush Doctrine and War with Iraq" in Parameters, Spring 2003, pp. 4-21.  Compare Michael Walzer, Arguing about War (New Haven: Yale University Press, 2004)  writing in September 2002 of Pres. Bush's West Point speech of May 2002,  "preemption is not an accurate description….The war that is being discussed is preventive, not preemptive—it is designed to respond to a very distant threat."  He goes on to write that "International lawyers and just-war theorists have never looked on this argument with favor…" but admits that the gulf between preemption and preventive wars might be narrowed by WMD's and their delivery systems. pp. 146-147.

[7] Grotius, On the Law of War and Peace trans A.C. Campbell (Kitchener Ontario: Batoche Books, 2004) Book II Ch. 26 "apprehensions from a neighbouring power are not a sufficient ground for war. For to authorize hostilities as a defensive measure, they must arise from the necessity, which just apprehensions create; apprehensions not only of the power, but of the intentions of a formidable state, and such apprehensions as amount to a moral certainty." (p. 227).  M. Walzer Just and Unjust Wars. From a New York Times Editorial: "We know that Mr. Kerry does not rule out preemptive strikes if a country poses a clear and serious danger to the United States or its allies -- that's longstanding American policy, and it's in the U.N. charter." July 29, 2004.  The casual and apparently mistaken invocation of the U.N. Charter was retracted Aug. 4, 2004.  A massive Christian, Niebuhresque defense of deterrence and preventive war is contained in Paul Ramsay's The Just War: Force and Political Responsibility (1968, reissued with a foreward by Stanley Hauerwas, 2002).

[8] "Well ordered peoples… [initiate war] only when they sincerely and reasonably believe that their safety and security are seriously endangered by the expansionist policies of outlaw states" pp. 90-91. "Well-ordered peoples do not go to war against each other, but only against non-well-ordered states whose expansionist aims threaten the security and free institutions of well-ordered regimes…." p. 94  While these passages do not precisely offer sufficient conditions for the morality of preemptive wars, they put all their emphasis on the reasonably known "policies" and "aims," rather than actual deeds.

[9] Just and Unjust Wars, 1977, 1992, Chapter 5, "Anticipations."  For a morally justified first strike, he rejects imminent enemy attack as a necessary condition, arguing that "sufficient threat" is the relevant necessary condition, "by which I mean 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…greatly magnifies the risk" (p. 81).  He contrasts his first condition with Emmerich Vattel's condition (The Laws of Nations, 1758) of "injustice, rapacity, pride, ambition, or of an imperious thirst of rule" (quoted by Walzer, p. 78 and in footnote 5 p. 340 ).  "Injustice" would seem to suggest grounding humanitarian intervention, but I agree with Walzer that the remaining proposed sufficient conditions are too weak.  However, Walzer's "intent" is too strong: as the term is used in philosophy of action and law, "intent" would have some reference to time of performance or circumstances.  I replace Vattel's and Walzer's notions with a condition of bellicosity, which must usually be supported by extensive evidence from past incidents.       

[10] Stricter adherents of Just War Theory within the tradition of Augustine and Aquinas, through to recent Vatican pronouncements, have generally condemned anticipatory wars of most sorts—except Humanitarian Intervention.

[11]  Or at the very least, the Future Threat aspect of permitted war can be separated from Punishment for past deeds and from morally and emotionally neutral Tit-for-Tat (or other forms of rational deterrence to this and other nations in the future). There may be other justifications for military action, such as humanitarian intervention, in which a nation by action or inaction poses a threat to its own citizens.  I do not think an international consensus or UN mandate is alone sufficient for the permissibility of military action, nor (more controversially) is it necessary.  See below for this argument. The counterattacking nation does not constitute a legitimate and acknowledged legal authority for the administration of justice, and its military and defense institutions do not have that "purpose," a point made by Vattel (Tuck 1999 p. 194), Kant, and others.

[12] This has interestingly been disputed by Jeff McMahan in "War as Self-Defense," Ethics and International Affairs 2204 (18:1): 75-80 largely from idead fisrt developed in "Self-Defense and the Problem of the Innocent Attacker" Ethics Ja 1994 (104:2): 252-290, but the usually granted right of national self-defense is absolute.

[13] A threat to the defenders of those citizens (such as to military forces), rather than directly to the citizens, is a variant of this.  

[14] There is also frequently a blatant fallacy ad ignoratium, namely that since there is no or little literature justifiying preemptive war, therefore preemptive war is morally unjustified. (Observe that the premise that there is little literature defending preemptive war is mistaken.)    Cf. the APA condemnation of the Iraq War in Dec. 2002.  There is shred of rationality in this maneuver, however, since the onus is on justifying war: one cannot simply retort to a critic of a war, "Why not go to war?"  Observe that even if a set of jointly sufficient conditions for moral are met, this is not sufficient to make going to war wise or morally obligatory. 

[15] A right to existence, and to a stable political order or to what Rawls calls a "well-ordered" people, and more precisely a "decent" people (The Law of Peoples  1999 pp. 4ff), is after all necessary for exercising almost any other rights.

[16] I require that the central authority have justifying evidence in hand, not that it be justifiable, or later turns out to be justifiable.  This is partly because I intend these conditions both as describing both moral truths and as moral guidance for agents, a distinction not widely enough noted.

[17] Or you know that your neighbor, because of psychological or religious factors, is undeterrable by any future punishment, no matter how severe or likely.

[18] Legally excused or justified by an assertion of necessity, although that is not the issue here.

[19] While state of mind (usually, intent), mens rea, is required for many crimes, it is contrary to metalegal traditions and intuitions to make mens rea sufficient for the crime.

[20] Such as in the film Minority Report or involving historical-psychological laws as in Isaac Asimov "Robot" novels.

[21] The Catechism of the [Roman] Catholic Church requires as one necessary condition for war that "the damage inflicted by the aggressor on the nation or community of nations must be lasting, grave, and certain" (#2309) PART THREE: LIFE IN CHRIST /SECTION TWO: THE TEN COMMANDMENTS /CHAPTER TWO: "YOU SHALL LOVE YOUR NEIGHBOR AS YOURSELF" /ARTICLE 5: THE FIFTH COMMANDMENT /III. Safeguarding Peace/Avoiding War" and does not literally require that the damage be actual or past.

[22] Observed by Secretary of Defense Donald Rumsfeld as quoted in the Washington Post January, 2003.

[23] We can term this the "Rumsfeld-Kershner Principle," since one component of the proportionality comes from a remark by Donald Rumsfeld and Stephen Kershner urged me (while not endorsing it ) to make something like this formula explicit.

[24] Even if the single most recent example of overestimating such prowess, in the 2003 Iraq, is disaterously obvious. 

[25] Stephen E. Ambrose, Eisenhower: Soldier and President (NY: Simon & Schuster, 1990, originally published 1983-4), p. 369.  The first Soviet thermonuclear ("hydrogen") bomb was tested August 12, 1953.  A curiously related opinion was an earlier suggestion by Bertrand Russell to prevent by military means the Soviet Union from developing atomic weapons.  This is discussed in Ray Perkins, "Bertrand Russell and Preventive War," Bertrand Russell on Nuclear War, Peace, and Language: Critical and Historical Essays," Schwerin, Alan (ed), 3-13 (Praeger : Westport, 2002) ABSTRACT : Contrary to what many critics have charged, Bertrand Russell did not urge preventive war against the Soviet Union in the 1945-49 period. He did advocate threatening war in order to get the Soviets to agree to international controls on armaments, but he did so only as long as he believed it likely that the Soviets would accede to the threat." See also David Blitz, "Did Russell Advocate Preventive Atomic War against the USSR?" Russell: The Journal of the Bertrand Russell Archives Summer 02, 22(1): 5-45

[26] At least if we ignore the violation of Jordanian airspace, and French and Italian involvement in the reactor's construction.

[27] United Nations Security Council resolution 487 (1981) of 19 June 1981; General Assembly Resolution 39/14 (November 16, 1984)

[28] The Cuban Missile Crisis, the War of Spanish Succession, and a number of examples and European Union representatives that support the legitimacy of preventive war are discussed in Max Boot, "The Bush Doctrine Lives," The Weekly Standard (Feb. 16, 2004).

[29] Richard Holbrooke in reviewing Max Frankel's High Noon in the Cold War (2004), NY Times Oct 15 2004: "In 1962, unlike 2003, American intelligence and analysis was excellent. High-altitude photographs found and identified the missiles before they were deployed."  Evidence or belligerency was more lacking, and while intelligence about the missiles was good to excellent, intelligence about the status of their possibly nuclear warheads was dangerously in error.

[30] Holbrooke dismisses this, "By the time they were installed in early 1962 they were already obsolete; President Dwight D. Eisenhower said they should have been dumped at sea rather than sent to Turkey, and American nuclear submarines made them superfluous."  What Eisenhower thought is however not relevant, but rather what was important was the Soviet's perception of their threat.

[31] Here I allude to what I believe is an extension of the conditions justifying war (i.e. sufficient conditions that are alternatives to the traditional ones).  Namely, principles or conditions for going to war which, in the long run and with near certainty diminish the overall quantity of destruction through war over history when applied by all parties, are themselves morally justified sets of sufficient conditions for a just war.  What I have in mind is the seemingly ruthless tit-for-tat strategy that can be proven in game theory to be (one of) the most stable and least destructive such strategy to all parties if it is universalized and given certain facts of human nature and conflict.  This would also justify the proven deterrent concept of mutualized assured destruction ("MAD"), which has been historically effective but cannot be dealt with easily within traditional conditions for just war, such as those of Grotius.  Extended into jus in bello, such a meta-rule for conditions of just war might justify the otherwise oddly arbitrary and contractual nature of such "customs" as wearing uniforms and distinguishing between those who wear them and those who do not.  It is moral not just because we have agreed to them in the Geneva and other conventions, but because such practices are generally destruction-minimizing.   Unfortunately, such rules may also justify widely ridiculed rules of war such as fighting battles only in rows, in brightly colored uniforms, and in open fields.  The conclusions of employing my metaprinciple have similarities with views in the use of force between nations that are sometimes called (geopolitical) "realism."  However,  my justification for these endpoints is ultimately philosophical and moral, rather than claiming that moral reasoning does not apply.

[32] Walzer's position is less helpful than one might think.  He did not to my knowledge point out clearly that, presuming the possession of WMD's, Iraq in 2002-2003 fulfilled the criteria for anticipatory war in Just and Unjust Wars.  His position in the New Republic, various discussions that were transcribed, and later in Arguing about War seems to have been that a Last Resort condition was not met.  Although there is a great deal of policy and popular literature on the U.S. National Strategy, and later on the wisdom of attacking Iraq, most of it both pro- and con- has a partisan, Internationalist, canonical Just War (i.e. the position of the Roman Catholic Church), narrowly legal, or Realistic flavor.

[33] See for example, Michael Ignatieff, The Lesser Evil: Political Ethics in an Age of Terror (Princeton: Princeton University Press, 2004, pp. 162-167.  While this is a subtle and careful discussion, Ignatieff does not seem aware of the possibility that I am advocating, namely, that international approval is not a fundamental, moral condition for morally justified war.  Rather, its logical position is in support of the epistemic threshold, and it is this condition that is fundamental; even here it is also more limited than many, including Ignatieff, propose.

[34] See New York Times, July 11, 2004 discussing Kent's lecture notes and articles by his aides.  Interestingly, recent history suggests more that CIA and other U.S. intelligence services have tended to underestimate military potential of rogue nations (Cuban missiles, Libya, North Korea) than to overestimate them.  Iraq seems to be a fluke.  And while a lack of human intelligence was a notable failure, depending instead on the analysis of aerial and satellite surveillance, various figures high in Hussein's regime, including division commanders, believed that Iraq indeed had chemical WMD's; we could have had many such high-placed human intelligence sources and still been mistaken.

[35]   Namely, in its representation of veto-bearing members that is wildly undemocratic, but also in the ulterior economic and geopolitical motives that seem to more often than not play an unhelpful role.  The role of the Security Council in this epistemological dimension would have been greater if had not avoided its possible role in crises such as Kosovo and Rwanda, and more generally shunned the use of force in enforcing mandates such as disarmament.

[36]   This does not speak to prudential and legal issues, and does not address the moral issue in complying with international agreements, such as the U.N. Charter.  Article 51 clearly avoids stating that self-defense is permitted, without Security Council sanction,  only in response to armed attack.  See Anthony C. Arend,  "The War that Changed Everything," New Statesman March 15, 2004 for a discussion of the legal issues.  I believe preventive and maybe even preemptive (imminent) war is indeed contrary to the spirit or "implicit" text of of the U.N. Charter and thus probably intended to discourage it.  If one looks to legislative history for evidence of intent, it is clear that for major powers c. 1945, this ambiguity was intentional. 

[37]   The quantity and nature of evidence for any such actionable threshold differs with the domain of the subject and the associated risks of being mistaken in acting or of inaction.  Epistemic appraisal almost certainly has more dimensions and gradations than simply being knowledge or "mere belief" once it is regarded with respect to its impact on actions.  The recent history of epistemology and its intuitions about what constitutes one sort of "true knowledge" smacks of a lack of sophistication with regard to the uses and functional nature of beliefs in all of cognition. 

[38]  Observe that with "manifest,"  Walzer is tacitly introducing an unexplained epistemic condition.

[39] The Law of Peoples: 47f and earlier in "The Law of Peoples" (1993) in Collected Papers.

[40]   This is an invitation to a bellicose enemy to arm itself just up to the point where their resources would trigger preemptive war by this policy, or to conceal exactly those assets or plans that are likely to trigger preemption.   Announcing preemption policies in detail thus increases dangers to a nation.

[41] If all nations that constitute a substantial threat are deterrable, and means of deterrence such as mutually-assured-destruction are available, then announcing and following a policy of never initiating preemptive war is advisable.  It reduces the perceived threat to other nations by the nation that has announced such policies, assuming some modicum of believability and of observable weapons development and deployments consistent with those policies.  I assume however that we have entered a new world in which not all nations and organizations that constitute a substantial threat are deterrable. 

[42] I thank Stephen Kershner, David Hershner, Peter Sinden, Kenneth Shockley, James Hurtgen, and Alan Dipert for comments on earlier versions of this paper.