Just war theory and just assassinations

I use just war theory precepts to make my argument for several reasons. First, the same issues are at stake for war and assassination. Intentional and unintentional killing, violations of state sovereignty, who has the right authority to kill, the risk of powerful actors abusing accounts of when it is permissible to use force, and so on are all present in determining when wars and assassinations are permitted or required. Although both war and assassinations concern the same issues, I take no position on whether assassinations are wars. Just war theory provides guidance about who can legitimately use force, when, why, and how, and for this reason just war theory is a useful means to assess the morality of assassination. Second, I use just war theory precepts to argue that assassinations are required under some circumstances as a means to revise the just war theory tradition from within. The most important revision to just war theory regarding assassinations that follows from just war theory precepts is that many wars that just war theorists now believe are permissible would be impermissible whenever a just assassination would achieve the just cause and meet the other just war theory precepts. Whenever an assassination is just, a war that would achieve the same end is impermissible.

I differ in a number of important ways from others who have made related arguments. Andrew Altman and Christopher Wellman (Reference Altman and Wellman2008) argue that assassinations are morally permissible whenever humanitarian interventions are permissible. An armed intervention is permissible if a state inadequately protects human rights, according to them (Altman and Wellman Reference Altman and Wellman2008, 233). The first way I differ from Altman and Wellman is that I show that not only are assassinations permissible when the just war conditions are met, but also they are required instead of other policies under some circumstances. Second, I show that the conditions for a just assassination can be met when individuals commit severe non-violent severe harms as well as violent ones. Third, I show that even if one believes in more demanding interpretations of key just war theory precepts than Altman and Wellman, just assassinations are still permitted and required whenever they would likely cause the fewest severe harms to innocents. Even if one accepts Walzer's higher bar for the just cause of humanitarian intervention, for instance, just assassinations are still required under some circumstances. Just assassinations may also be required instead of a war of collective self-defense.

Daniel Statman and Fernando Tesón argue similarly that some just war theory precepts allow some targeted killings (Statman Reference Statman2012; Tesón Reference Tesón2012). Neither Statman nor Tesón argues that just war theory precepts require targeted killings, however. Rather, Statman argues that ‘individualist’, ‘collectivist’, and ‘contractualist’ accounts of just war theory permit targeted killing under some circumstances. By focusing on the principle of discrimination and assuming that jus ad bellum precepts are met, he fails to show that just war theory generally allows or requires targeted killings or assassination (Statman Reference Statman2012, 95). Fernando Tesón argues that four conditions from just war theory – roughly, just cause, proportionality, necessity, liability to defensive harm – sometimes make targeted killings permissible. None of these scholars argues that a diversity of just war theory precepts require assassinations instead of other policies that would likely severely harm more innocents.

Just cause

A jus ad bellum requirement is that for a resort to war to be permissible, there must be a just cause. This requirement should set a relatively high bar because it grants an exception to the strong general moral prohibition against killing and severely injuring others. Because assassinations also require an exception to the prohibition against killing, there must be a just cause for an assassination to be permissible. In this section I argue that at least whenever there is a just cause for war, there is a just cause for assassination if it will likely achieve the same ends. This is because if the exception to the prohibition against violence permits many killings, it must also permit one or a few killings, assuming all targeted individuals are liable to lethal defensive harm.

There are numerous possible just causes for war, including collective self-defense, defense of others, preempting aggression, preventing aggression, punishment, and recovery of lost goods (McMahan Reference McMahan2005a). Traditionally, resisting aggression by individual states or by a collection of states has been the central just cause for war, and it is the only just cause for war recognized under international law that does not require UN Security Council approval (Walzer Reference Walzer1977, 62; Rodin Reference Rodin2002; McMahan Reference McMahan2005a; Fletcher and Ohlin Reference Fletcher and Ohlin2008; May Reference May2008, 56). Aggression is using military force to violate the territorial integrity or political independence of another state without a just cause. Article 51 of the UN Charter permits any country that is illegally attacked to decide to and wage a defensive war. It furthermore allows other countries go to war in aid of the victimized state.

Another widely accepted just cause for war is humanitarian intervention. A humanitarian intervention is the violation of state sovereignty by foreigners for humanitarian reasons where there is some substantial evidence of this intention (Pogge Reference Pogge2003, 93; McMahan Reference McMahan2010, 44). The debate about what constitutes a just cause of humanitarian intervention centers on the relative importance of human rights and collective self-determination. I briefly explain the two leading positions of the just cause criterion of humanitarian intervention of Michael Walzer and his critics (Walzer Reference Walzer1977, chap. 6, Reference Walzer1980, Reference Walzer2007, chap. 14; Beitz Reference Beitz1980, Reference Beitz2009; Luban Reference Luban1980a, Reference Luban1980b; Doyle Reference Doyle2009).

Walzer values highly the collective right to self-determination. Because he believes that collective self-determination exists in ‘cases of “ordinary” oppression’, only when there is genocide, mass killings, mass slavery, or ethnic cleansing, is there just cause for humanitarian intervention (Walzer Reference Walzer1977, Reference Walzer1980, 217–218). Whereas Walzer highly values collective self-determination, his critics such as David Luban, Charles Beitz, and Christopher Wellman argue that the protection of some fundamental human rights should trump the group right to collective self-determination and state sovereignty (Beitz Reference Beitz1980, Reference Beitz2009; Luban Reference Luban1980a, Reference Luban1980b; Wellman Reference Wellman2012). Although the details of their accounts differ, their central contention is the following: ‘if we take human rights as seriously as we should, then even a legitimate state has no principled objection to outsiders’ intervening in its internal affairs if this interference will prevent just a single human rights violation’ (Wellman Reference Wellman2012, 119).

Each of these just cause principles allows the cross boarder use of lethal force in different circumstances and for different reasons. Because each of these accounts permits the use of lethal force against numerous individuals, each permits the killing of just one or a few persons.

Proper authority

I argue that those entities that can legitimately carry out a just war can also carry out a just assassination because any actor that possesses the authority to use great lethal force against many across state borders is permitted to use lethal force against one or a few. There are two central questions regarding who can carry out a just assassination (Fabre Reference Fabre2008, 963). First, who or which institutions are permitted to authorize the use of force? Second, who or which institutions are permitted to execute the authorized use of force?

Traditionally, leaders of states have been the only proper authority able to decide to wage war and soldiers of states have been the only individuals authorized to carry out wars (Frowe Reference Frowe2011, 59; Fabre Reference Fabre2012, 142–143). Just as an individual victim need not wait for the police or a judge to decide that she can defend herself, a country that is the victim of aggression need not wait for the UN Security Council or any other institution to authorize and enact a just war against the aggressor. Some believe humanitarian intervention is different. To use force legitimately to halt human rights abuses, some argue that the UN Security Council must authorize it (Kahler Reference Kahler2011). According to Altman and Wellman, however, there is an ‘absence of consensus on the question of authority’ among just war theorists regarding which actor if any must authorize a humanitarian intervention in order for it to be legitimate (Altman and Wellman Reference Altman and Wellman2008, 232). The commissioners of 2001 ICISS report argue that UN Security Council is the most legitimate legal actor to authorize humanitarian intervention (ICISS 2001, 49–53). Yet even they allow regional organizations and the UN General Assembly to authorize a humanitarian intervention if the UN Security Council fails to act when mass atrocities occur (ICISS 2001, 53–55). My argument is compatible with revisionist accounts of proper authority as well. Cécile Fabre, Magnus Reitberger, Deane-Peter Baker, Uwe Steinhoff, and James Pattison for instance propose a variety of alternatives to restricting proper authority to states or intergovernmental organizations (Fabre Reference Fabre2008; Steinhoff Reference Steinhoff2009; Pattison Reference Pattison2010b; Baker and Pattison Reference Baker and Pattison2012; Reitberger Reference Reitberger2013).

For whichever actor or set of actors one accepts as a proper authority to assess and execute a just war, developing accountability mechanisms to prevent them from abusing their authority to assassinate is vital. Because the financial and human costs to those actors that perform assassinations are much lower than wars, they may be more willing to abuse such authority. Although it is beyond the scope of this paper to develop detailed suggestions for how to prevent the abuse of proper authority to wage just assassinations, I briefly comment on two options. The Rome Statute of the International Criminal Court could be amended or the text with the updated amendment on aggression could be used to try individuals who commit and order the commission of unjust assassinations. In 2010, states agreed to outlaw one of the most controversial actions, aggression, by amending the Rome Statute (Scheffer Reference Scheffer2010). Whether or not such global institutional reform or use would be possible, states should develop internal checks such as judicial review by a domestic court for possible assassinations as has recently been proposed in the United States regarding foreign drone strikes (Shane Reference Shane2013). States should also develop laws that could try individuals for carrying out unjust assassinations, just as police can now be held legally accountable for abusing their special authority to use force. While such ex ante and ex post accountability mechanisms would not prevent all cases of unjust assassinations, they would provide realizable legal rules to curb the abuse of the proper authority to perform just assassinations. Whether one accepts the traditional or revisionist views of legitimate authority for a just war, the same actors should be able to decide about and execute a just assassination.

Right intention

Another jus ad bellum precept is right intention. A right intention is acting to achieve or being motivated to realize a just cause. One may accept that just assassinations are permissible only if the proper authority has a right intention, because both just wars and just assassinations involve the use of force to achieve a just cause. There is some disagreement about what constitutes the right intention precept. Although intention and motive are often conflated, Terry Nardin and Pattison among others argue that an intention is one's goal or aim, whereas a motive is the reason one holds for wanting to achieve that goal (Nardin Reference Nardin2006, 9–11; Pattison Reference Pattison2010a, 154–155). For instance, the leaders of Tanzania, who ordered the invasion of Uganda in the late 1970s, were likely motivated in part by self-interest, even though the invasion also achieved the humanitarian outcome of ending the murderous reign of Idi Amin (Wheeler Reference Wheeler2000, chap. 4; Ellis Reference Ellis2003, 18).

Just war theorists disagree about whether there must be a right motive, a right intention, both, or neither (ICISS 2001, 36; Ellis Reference Ellis2003, 18; Nardin Reference Nardin2006, 11; Pattison Reference Pattison2010a, chap. 6). Whichever interpretation of the just war theory precept of right intention one favors, it also meets this criteria for a just assassination because both aim at ensuring that the proper authority is acting for the right reasons.

A reasonable expectation of success and epistemic uncertainty

There must also be a reasonable expectation of success for a war to be permissible. Likewise, only if an assassination is likely to achieve a just cause can it be permissible. This raises the vexing question of what constitutes a reasonable chance of success and how anyone can make such an assessment. No consensus exists among just war theorists regarding this requirement or how to apply it. In fact, there is surprisingly little written addressing directly what constitutes a reasonable expectation of success. Many just war theorists mention the requirement in passing only (Walzer Reference Walzer1977, 107; ICISS 2001, 37–38; Coady Reference Coady2003, 290–291; Lucas Reference Lucas2003, 88–89; Hurka Reference Hurka2008, 127–129; Frowe Reference Frowe2011, 56–59; Fabre Reference Fabre2012, 5). As with the other precepts I argue permit but do not require a just assassination, I accept that many just war theory account of reasonable expectation of success meets this criterion for a just assassination.

There are two components of a reasonable expectation of success. First, there must be some moral assessment of what qualifies as a reasonable probability of success. Second, there must be adequate information to make such an assessment. This second component requires specifics on the case at hand and more general probabilities of success in similar cases. I consider each in turn.

First, there is a moral assessment component of the likelihood of success requirement. Second, a reasonable expectation of success entails a practical judgment about likely outcomes (Brown Reference Brown2003, 44–46). The second component requires two types of adequate information to make such judgments. First, one must have knowledge about the probabilities of outcomes given past similar situations. There is some literature on the likely outcomes of military interventions (see, for instance, Fortna Reference Fortna2004; Krain Reference Krain2005; Murdie and Davis Reference Murdie and Davis2010; Kathman and Wood Reference Kathman and Wood2011; Peksen Reference Peksen2012), targeted killings (Carvin Reference Carvin2012; Johnston Reference Johnston2012), and assassinations (Jones and Olken Reference Jones and Olken2009) on which one can base such assessments. Those considering using force against one or many should draw on these and future findings of scholars to inform their decision making in addition to relying on knowledge regarding a specific situation.Footnote ⁴ Second, one must have good information of the specific case regarding the likely outcomes of various options (which also apply to proportionality considerations).

Some might object that making such judgments requires perfect or unrealistically good information. It does not require either of these. Nor do just war theorists require such an unrealistically high bar for information. Making such an assessment simply requires the same level of certainty of information regarding the likelihood of success as just war theorists require for assessing whether a war will likely be successful. Like those deciding on whether a war would likely achieve a just aim, those deciding on whether an assassination would be just must base their choices on the best information available, including intelligence. Tragically, individuals will inevitably make mistakes. Some may decide that the risk is too great and favor pacifism, or hold that assassinations are never permissible. But all just war theorists accept that executing a just war, or as I argue here, an assassination, given good information that suggests there is a reasonable expectation of success is permissible whenever all other just war precepts are met.

Liability to harm, excuses, and justifications

What, if anything, can make someone liable to be killed in order to protect others or oneself? By liable to be killed, I mean that one is not wronged and one does not have one's rights violated if attacked intentionally (McMahan Reference McMahan2009, 8–10). Throughout this article, I am only referring to liability to harm for reasons of self-defense or the defense of others (Quong Reference Quong2012, 45). An innocent person is one who is not at all liable. An attack against a liable person is generally justified only if it is necessary to prevent proportionate harm to one or more innocent people. An intentional attack against one or more people that is morally permissible or justified is a rare exception to the general moral and legal prohibition against intentionally causing serious bodily harm to adults who do not consent to such harm or the risk of such harm. The bar for liability should be set high because of how serious a wrong it is to attack an innocent person and because the powerful sometimes stretch and abuse justifications to fit their own purposes. The rights one can forfeit because one has made oneself liable to defensive harm include some of those to bodily security and due process, among others (Altman and Wellman Reference Altman and Wellman2008, 252). Individuals forfeit their rights not to be harmed on the basis of self-defense or in the defense of others, according to leading theories of liability, because the individual poses a serious threat to at least one person, among other possible conditions (Anscombe Reference Anscombe1961; Walzer Reference Walzer1977; Thomson Reference Thomson1991; Rodin Reference Rodin2002; McMahan Reference McMahan2005b, Reference McMahan2009; Frowe Reference Frowe2010; Mapel Reference Mapel2010; Quong Reference Quong2012). Liable people need not be attacked or killed to be made non-threatening. They can sometimes be arrested or persuaded to not pose a threat, for instance. These less harmful options must generally be taken instead of intentionally harming a liable individual, if the distribution of the risk of harm does not fall disproportionately on innocent third parties or on the innocents who are attempting to carry out the less harmful options.

In this section, I discuss several accounts of liability that are roughly laid out from the most to least permissive. I briefly summarize these prominent positions to show that all allow killing of civilian or military leaders under some circumstances. The main criteria that scholars use to argue for various accounts of liability include posing a threat, whether a threat is objectively or subjectively unjust, and to what degree someone is morally responsible or culpable for posing a subjectively or objectively unjust threat. These are different accounts of how a right against being harmed is lost or forfeit and thus of how one becomes liable to such harm. After discussing the leading theories of liability, I then show that liability can sometimes be annulled by excuses or justifications.

The first account of liability is associated with one interpretation of Walzer and holds that anyone who has made herself dangerous or a threat to others is liable (Walzer Reference Walzer1977, 145–146; Mapel Reference Mapel1998; McMahan Reference McMahan2009, 11). McMahan interprets Walzer as saying that ‘just posing a threat’ is sufficient for liability (McMahan Reference McMahan2009, 11). This account does not require any moral responsibility of an individual to make herself liable. Proponents of this interpretation of Walzer can point to passages of Just and Unjust Wars such as the following. Individuals lose ‘their rights because of their warlike activities’ (Walzer Reference Walzer1977, 145). A soldier ‘can be personally attacked only because he already is a fighter’, that is, he is a ‘dangerous man’ (Walzer Reference Walzer1977, 145). Walzer writes, ‘most soldiers most of the time have not chosen the combat and discipline they endure’ (Walzer Reference Walzer1977, 305). The war convention accepts that soldiers may be ‘victims’ (Walzer Reference Walzer1977, 45) because soldiers often are ‘coerced moral agents’ (Walzer Reference Walzer1977, 306; Mapel Reference Mapel1998). Responsibility requires being able to make a choice. Walzer believes that even if soldiers do not freely make a choice to join a military, and that therefore they are not responsible for their actions, they can still be liable if they post a serious threat to others. In addition to directly posing a threat as a soldier on a battlefield might, individuals including civilians can too make themselves liable by directly contributing to posing a threat (Walzer Reference Walzer1977, 144–146). For instance, a civilian that works at a tank factory, but not one that works at a food factory even if the food is only being produced to feed soldiers at war, could be liable (Walzer Reference Walzer1977, 146). On this account of liability, it does not matter whether one's threat is just or unjust, unlike the second account.Footnote ⁵

The second account of what is sufficient for liability to attack is posing an objectively unjust threat against one or more people. On this account, like the first, causal responsibility, but not moral responsibility, is sufficient for liability. This account is less permissible than the first because only those that pose objectively unjust threats, and not those that pose any threat at all, are liable. Associated with J.J. Thomson and G.E.M. Anscombe, this account is intuitively attractive (Anscombe Reference Anscombe1961; Thomson Reference Thomson1991). The basic idea is if someone is going to violate your right to not be unjustly physically attacked, you and often others have a right to harm and even kill the attacker, if doing so is necessary and proportionate to annul their threat. On this account, a sadistic mass murderer who poses an imminent lethal threat to innocents is just as liable to attack as a nun who through no fault, negligence, or recklessness of her own loses control of her car which runs a red light and is about to hit an innocent pedestrian because of an unforeseeable brake failure.Footnote ⁶ Both are liable because both pose an objectively unjust threat to an innocent person. The reason the nun can be killed on this account is not because she knowingly took a small risk that driving might result in such a freak accident; it is simply that she is posing the objectively unjust threat.

A third view, advanced by McMahan, holds that ‘moral responsibility for an objectively unjustified threat of harm’ is sufficient for liability (McMahan Reference McMahan2005b, Reference McMahan2009, 35). On this account, intentionally causing an unjust threat is sufficient for liability, but so too are negligent actions. Knowingly taking a risk that could result in, and by bad luck does result in, an unjust threat is enough to make one liable to some degree for McMahan. The conscientious car driver example from above is similar to one McMahan uses to illustrate this argument (McMahan Reference McMahan2009, 162–167, 176). According to McMahan, people who are morally responsible and not in any way excused are most liable to attack. Those who are responsible but partially excused are liable to some degree. Those whose excuse fully annuls their responsibility are not liable at all, just as innocents are not at all liable. McMahan argues that soldiers that fight on the side of a just war do not forfeit their rights because they have done nothing wrong and therefore cannot be justifiably targeted. There is no moral equality of combatants on this account of liability. Nor can jus ad bellum be divorced from jus in bello. For those who pose unjust threats, however, he sets the liability bar low by arguing that anyone who is to some degree morally responsible is liable even if that person does not intend to pose the threat. McMahan believes that culpability implies blame and is a higher standard than moral responsibility (McMahan Reference McMahan2009, 162).

Jonathan Quong proposes a fourth account which holds that someone can be liable only if she is morally responsible for treating someone as if they are liable when in fact they are not, based on the available evidence and accurate moral beliefs (Quong Reference Quong2012, 62–77). This view differs from McMahan's responsibility account because it is based in part on the evidence available to the agent, not on whether an act is objectively unjust. It precludes liability of the nun driver because she does not wrongly treat the pedestrian as liable when in fact the pedestrian is not, acts as best she can given the evidence available to her (by stepping on the brake and trying to swerve), and holds accurate moral beliefs (that cautious driving is permissible).

The fifth account of what conditions are sufficient for liability requires a higher standard of responsibility, namely culpability for posing an unjust threat, which is associated with David Rodin and Helen Frowe (Rodin Reference Rodin2002, chap. 4, 189; Frowe Reference Frowe2010). Rodin writes it is ‘the fault of the aggressor and the innocence of the victim’ that creates the moral asymmetry allowing the victim and in some cases third parties to kill the aggressor in order to protect the victim (Rodin Reference Rodin2002, 77–79, quote from 77). Frowe writes that ‘whether or not the target is liable to be killed depends on objective facts about their culpability’ (Frowe Reference Frowe2010, 269). On the responsibility account but not on the culpability account, someone is liable to defensive harm even if someone knowingly takes a reasonable risk (such as carefully driving a well maintained car to work) that through no fault or negligence of hers becomes an unjust threat.Footnote ⁷

The crucial point from this brief discussion of various accounts of liability is that they all permit the intentional harming or killing of politically powerful figures if they are culpable for posing an objectively unjust threat against an innocent person that is not excused or justified.Footnote ⁸ In principle, then, each account of liability allows for just assassinations if the target is morally culpable for posing an objectively unjust threat so long as other just war criteria are met. For the sake of argument, I adopt the last and most stringent of the accounts of liability because anyone who can justifiably be attacked according to this account can generally also be attacked according to the other accounts.

The definition of objectively unjust serious threat I adopt is one that from an impersonal point of view violates or imposes unreasonably high risks of threatening someone's right to bodily security. One can pose an objectively unjust threat directly or indirectly. A hostage taker poses a threat directly by holding a gun to the hostage's back. The subject of this paper, however, concerns those who generally pose threats indirectly. Someone ordering at least one other person to harm or threaten a third person is an indirect threat. A mob boss who orders the murder of a third person is posing a threat indirectly. As this example shows, someone who never kills, shoots, or bombs can be liable by commanding or persuading others to execute objectively unjust threats. As McMahan writes, ‘moral responsibility for an unjustified threat that one does not oneself pose is sufficient for liability to harm as a means of protecting the person wrongly threatened’ (McMahan Reference McMahan2009, 207, emphasis added). McMahan calls people ‘Culpable Causes’ who are morally and causally responsible for posing an objectively unjust threat (despite his responsibility and not culpability-based account of liability) (McMahan Reference McMahan2009, 206). I will adopt this term for those who meet the culpability criterion for liability.

That individuals can be morally and causally responsible for posing an unjust threat, even if they do not personally pose a threat, coincides with commonsense intuitions and many formulations of criminal law. Judges at international criminal tribunals hold high ranking political and military officials legally accountable for certain types of human rights violations even though many never personally murdered, raped, or tortured. For instance, one can be held criminally accountable by planning, instigating, aiding or abetting, or command responsibility, among others. The principle is present in domestic criminal law as well. A gang leader who orders a murder can be held legally responsible for the killing even if he were sitting comfortably in an office miles from the murder scene at the instant his thug kills.

But sometimes one can be excused for posing an objectively unjust threat and sometimes these excuses can annul liability according to those accounts of liability that require an individual to have some degree of moral responsibility (Rodin Reference Rodin2002, chap. 4; McMahan Reference McMahan2009, chap. 3–4). A full excuse can annul responsibility and culpability for an act that is objectively wrong (McMahan Reference McMahan2009, 110). If some moral responsibility is a necessary condition for liability, and some excuse sufficiently diminishes one's moral responsibility, one can no longer be liable. Excuses entail a mitigation of responsibility due to duress, epistemic limitation, or a ‘diminution of the capacity for responsible action’ (insanity or youth, for instance) (McMahan Reference McMahan2009, 116). These types of excuses generally do not apply to politicians. Unlike soldiers who are sometimes coerced to fight under threat of physical harm, politicians are generally not under such duress. Nor are they generally excused because of paucity of information or a lack of capacity to reason morally. They are typically the best-informed individuals of any society on intelligence matters, and generally have many other career opportunities. Of course there may be rare exceptions to these generalizations, but they are just that, atypical. Thus, generally politically powerful individuals cannot use the typical set of excuses to annul their liability.

Another way people may annul their liability is by a justification (McMahan Reference McMahan2009, 44). A justification is a permission plus a moral reason to act (McMahan Reference McMahan2009, 44). Rarely, there is a morally justified reason to override someone's right. McMahan calls this ‘infringing’ rather than ‘violating’ a right in order to differentiate the moral permissibility of each case (McMahan Reference McMahan2009, 10). The doctrine of double effect (DDE) is an example of when some innocents’ rights may be justifiably infringed. DDE holds that an attacker is justified in intentionally aiming at a military target that will foreseeably but unintentionally infringe the rights of a number of innocent civilians that is proportionate to the advantage of the military attack if the attack is necessary to achieve a just victory.

Considering justifications is important here because one type of justification that may exonerate political leaders, and hence annul their liability, is a situation of ‘dirty hands’ (Machiavelli Reference Machiavelli1513; Berlin Reference Berlin1971; Walzer Reference Walzer1973; Coady Reference Coady2008). The basic idea of the problem of dirty hands is that all the options available to powerful individuals will result in serious harms or deaths to innocents. They are responsible for the choices they make, but because the set of choices available to them does not include one that would not pose a threat, they cannot be culpable and hence not liable for the threats they impose on others. Unlike individuals who can (it is often thought) easily avoid posing objectively unjust threats, politically powerful leaders can face the problem of dirty hands.

I do not dispute the empirical claim of Walzer and others, and I concede that the problem of dirty hands justifies the option that minimizes unjust threats because they are unavoidable and one cannot be culpable or responsible for harms that are out of one's control. But even if we accept that the problem of dirty hands can exist and accept that justification can negate liability, it does not justify those actions that would result in more deaths than is absolutely necessary given legitimate constraints, such as consent of Congress and constitutional restrictions on power, that any individual would face in similar circumstances. For instance, if a politician assumes office with troops already engaged in combat abroad, and if either leaving the troops deployed or withdrawing them would directly or indirectly result in deaths of innocents, the politician should generally choose the option that minimizes threats to innocents. Of course there are other considerations that could count morally in whether a threat is objectively unjust, as this example shows. For instance, it may be permissible to give some priority to those with whom one has a special relationship: compatriots, friends, family, etc. This is far too complex an issue to discuss in detail here. Some special relationships may justify not minimizing unjust harms. My only claim is that the problem of dirty hands is a justification that annuls liability whenever threats to innocents are minimized, because the threats are unavoidable and the leader cannot be responsible or culpable for the threat, and that not minimizing objectively unjust harms can make one liable even in a situation of dirty hands.

Even if one does not believe that justification can negate liability, there is another reason that that the problem of dirty hands rules out killing the person facing the problem of dirty hands. When the individual in power minimizes harms to innocents, even if a great many are harmed, killing the powerful person would not prevent any threat, ex hypothesis. There is no less harmful option, so the criterion that the assassination would annul a threat to one or more people cannot be met. The problem of dirty hands thus precludes assassinating people who pose threats that cannot be avoided based on legitimate constraints for two different reasons. Accepting this limits in another way when politically powerful individuals can be intentionally targeted for attack.

The problem of dirty hands raises the question of whether assassination is permissible or required if it only mitigates but does not erase entirely an objectively unjust threat. Imagine, for instance, that the person who would replace Assad would still be willing to kill unjustly to maintain power, but that Assad's replacement would be willing to kill only half or even only one less innocent than Assad. Would it still be permissible or required to assassinate Assad, assuming the other conditions for a just assassination are met? So long as the assassination would likely result in the annulment of a serious objectively unjust threat against one or more individuals, it would be permissible on my view, assuming the other criteria are met. But because the degree of uncertainty is typically relatively high regarding how many innocent people's lives would be saved by a just assassination, in practice the person posing an objectively unjust threat would generally have to threaten seriously many people.

Civilian immunity and civilian liability

One might agree with the argument so far, but object to the possibility of extending liability to civilians because of the principle of distinction and the general prohibition under international law of intentionally targeting civilians who have not taken up arms [Geneva Convention, Additional Protocol 1 1977, Article 51 (3)]. I admit that my argument somewhat undermines the traditional principle of discrimination, which holds that all soldiers at war, but not civilians, are liable to attack. I also concede that my argument undermines the categorical prohibition of international humanitarian law of intentionally targeting unarmed civilians. My argument applies to civilians and soldiers alike.

But undermining the principle of discrimination given the liability criterion is not a problem for a few reasons. First, my argument does nothing to undermine the longstanding prohibition of intentionally targeting average civilians because the vast majority of civilians – even civilians who pay taxes to a government that perpetrates unjust threats against innocents in their name – are not liable according to any of the accounts of liability. It would be disproportionate to lethally target the vast majority of civilians because they have not done anything sufficiently wrong to make themselves liable to serious bodily injury or death, and often such individuals are to some degree excused because taxes are collected coercively. Second, the principle of distinction and the Geneva Conventions focus on jus in bello and are premised on a stark dividing line between jus in bello and jus ad bellum as is typical in the just war tradition. My argument allows targeting individuals, including civilians who start and direct wars without a just cause, as well as those who have a just cause but who order it fought unjustly.

This is not a radical view. Leading just war theorists agree that intentionally targeting civilians is sometimes permissible. Just war theorists as diverse as Walzer and McMahan allow intentionally targeting a small fraction of civilians who are directly engaged in war making activities or who are morally responsible for causing an objectively unjust threat (Walzer Reference Walzer1977, 145–146; McMahan Reference McMahan2009, 208, 210, 221–231). Walzer allows targeting civilians who are engaged in activities that are directly related to war and do not have dual uses, as I noted above with the tank and food factory examples Walzer gives. McMahan also allows the intentional targeting of a small class of civilians because they have made choices and committed actions that make themselves liable. McMahan calls civilians who have done nothing to make themselves liable to attack ‘just civilians’ and civilians who have done something to make themselves liable to attack ‘unjust civilians’ (McMahan Reference McMahan2009, 212). As McMahan argues, it does not matter morally whether someone is military or civilian. People can be targeted not because of what group they belong to but because of actions they decide to take. Perhaps in rare circumstances they may additionally be targeted because of actions they refrain from taking when this violates a duty that would result in objectively unjust grave risks to innocents. Basing liability on group membership, even if that membership is based on whether someone is part of a military, is close to a genocidal or terrorist mindset because it takes the morally relevant factor as group membership rather than the choices that someone makes (McMahan Reference McMahan2009, 209).

My argument may undermine some parts of international law as well. Although Roland Otto argues that assassinations are ‘generally regarded as being prohibited under international law’ (Otto Reference Otto2012, 19), some aspects of the laws of war allow targeted killing. ‘[C]ombatants or civilians directly participating in hostilities’ can be intentionally targeted under the laws of war (Melzer Reference Melzer2008, 56). The laws of war furthermore permit a country at war to intentionally target politicians from the enemy state as individuals if they control military units because such politicians are considered combatants (Dinstein Reference Dinstein2010, 107).

Perhaps the critic concerned about weakening the principle of distinction is convinced by this moral and legal discussion that sometimes civilians can be intentionally targeted for just assassinations, but replies that it is of little practical significance because civilians generally do not pose objectively unjust threats and therefore are not liable. One of the most prominent contemporary just war theorists defends this view. McMahan writes that ‘it is rare for any civilian to bear a significant degree of responsibility for any unjust war, or for any of the particular acts of war of which the war is composed’ (McMahan Reference McMahan2009, 225). On a probabilistic assessment of all civilians within a country, this is true. Yet when viewed by who decides whether to fight a war, and who sets strategic goals in a war, McMahan's claim is false. Take the United States as an example. The President and Congress, who are all civilians, have the responsibility for starting, funding, and strategically directing wars. Other countries are similar. Elected civilian leaders not only often bear a significant degree of moral responsibility for starting and directing wars, but they bear primary and overwhelming moral responsibility for both. If Bush decided not to wage war against Iraq in 2003 and Afghanistan in 2001, those wars would likely not have happened. If Obama wanted to withdraw all the troops from Afghanistan when he came to office, he could probably have done so. The same is true for other countries’ leaders. In fact, I am not aware of any democratic country that does not grant civilian leaders war-making powers, whether that power is vested in one person or many. Thus it is of great practical significance that civilian leaders can be targeted for just assassinations because civilian leaders often bear the highest degrees of moral responsibility for imposing objectively unjust threats on innocents.

Another objection is that it is problematic that my argument allows intentionally targeting any number of powerful individuals who have made themselves liable to harm without any degree of excuse or justification. This may seem extreme. There must be some limit on the number of people who can be targeted permissibly, a critic might claim. Yet targeting any number of enemy soldiers as necessary and proportionate is widely accepted within traditional just war theory and even within its revised versions of recent years (Rodin Reference Rodin2002, 53; McMahan Reference McMahan2009, 19–20, 22–23, 204). What matters is liability to harm and other just war precepts, not the absolute numbers of people who are killed. While it is better to kill as few liable individuals as possible to achieve a just aim, it is permissible to kill as many as is necessary to alleviate the objectively unjust threat, so long as the numbers killed are proportionate and necessary.

Imagine for instance that a dictator has populated the line of succession with ruthless individuals who have a history of committing objectively unjust acts. The 11th person is in the succession line, however, is an opposition figure who has been stripped of any real power but who is humane, pro-democracy, and who has no history of posing objectively unjust threats. Would it be permissible to all at once attempt to assassinate the dictator as well as the next nine people? The precepts of just war theory would allow their intentional killing only if they are culpable individually or collectively for posing unjust serious threats (and meet the other criteria). Although any number of liable people can be killed that are necessary and proportionate to the gains achieved, just assassinations aim to drastically limit the number of people harmed or killed compared with a just war or other harmful policy.

Proportionality

In this section, I argue that various proportionality requirements of jus ad bellum and jus in bello require just assassinations instead of just wars and other options that would harm a great number of partially or fully innocent people when the same just aims can be achieved by a just assassination. There are three groups that would likely suffer disproportionate harm if another policy that was less targeted were used in place of a just assassination. These groups are partially excused unjust combatants and civilians, just combatants, and innocent civilians on the side of the just and unjust attackers. I discuss in turn each of the groups and the various proportionality requirements.

There are proportionality requirements of jus ad bellum and jus in bello. And some scholars, like McMahan, believe they are connected. McMahan argues that generally only if a war has a just cause can a war be fought proportionally (McMahan Reference McMahan2009). I will not take a position on whether jus in bello and jus ad bellum are connected, but the argument presented here should be convincing whether or not one adopts McMahan's position. One of the central ad bellum proportionality requirements is that resort to war is impermissible if it would foreseeably but unintentionally cause more harms to innocents than refraining from war. The central jus in bello proportionality requirement holds that every action in a war should be proportionate to the action's contribution to the overall goal of the war (Hurka Reference Hurka2005; Mellow Reference Mellow2006). These comparative assessments of likely harms to innocents from doing nothing vs. intervening is what McMahan calls wide proportionality (McMahan Reference McMahan2009, 20–21). Wide proportionality and narrow proportionality refer to two different groups of people. Wide proportionality refers to fully innocent people, whereas narrow proportionality refers to partially or fully liable individuals (McMahan Reference McMahan2009, 20–21). If someone were responsible for doing something that only makes him liable to be harmed but not killed, it would be narrowly disproportionate to kill her. For instance, it would be narrowly disproportionate if A killed B if B were just threatening to stomp on A's toe if she refused to give up his spot in a grocery store line assuming that B's threat went no further. B has to some extent made himself liable to defensive action because of his violent threat, but neither A nor anyone else can justifiably use lethal force to protect A. Just war theorists have traditionally considered wide proportionality only (McMahan Reference McMahan2009, 22).

Yet if one accepts that some and perhaps most soldiers are to some degree excused and hence less liable than a fully responsible individual, as many do (Mapel Reference Mapel1998; Rodin Reference Rodin2002, chap. 4; McMahan Reference McMahan2009, chap. 3), the consensus view that generally any number of soldiers can be killed in pursuit of a just cause is undermined because everyone who is to some degree excused should count to some degree in narrow proportionality considerations. Partially excused unjust soldiers and civilians are one category of individuals that must be included in proportionality calculations. If many unjust soldiers and civilians are to some degree excused, narrow proportionality provides one reason why just assassinations are preferable to just wars. Considering only those individuals that have made themselves partially liable, a just assassination is more proportionate than a just war or other policy that would result in the severe harms or deaths of these individuals.

Second, because just combatants are generally innocent unless they violate jus in bello rules, they too should be included in wide proportionality considerations (McMahan Reference McMahan2009, 27). If this is correct, it is obvious why just assassinations are preferable to just wars when considering just combatants: more just combatants would die or be put at unnecessary risk in any war than carrying out a just assassination. It would therefore be disproportionate to carry out a just war, considering just combatants.

The third category is innocent civilians on the just and unjust sides. It is most obvious why it would be disproportionate to fight a just war or impose other policies that would harm more innocent civilians if a just assassination were possible. Even relatively limited wars inevitably result in numerous deaths of innocent civilians that should be avoided whenever possible. A just assassination is one way to avoid such deaths.

In sum, a just assassination is required instead of a just war or any other polices that would harm more partially or fully innocent people because any policy except the just assassination would impose unfair risks of harm or actual harms on three groups of people. These groups include civilians on the side with the just and unjust cause, just combatants, and partially excused unjust combatants and partially excused unjust civilians. Whenever a just assassination would suffice to achieve a just aim, I discuss next why any policies that would harm more innocents would be unnecessary and hence unjust.

Necessity

Just assassinations are sometimes the only option that can meet the principle of necessity (necessity for short). Although some such as Hurka suggest that necessity can apply to jus ad bellum as a placeholder for last resort (Hurka Reference Hurka2008, 129), necessity typically applies to jus in bello considerations. It prohibits attacking any human or material target with a military use if its destruction is not required to achieve a just cause. According to McMahan, ‘if harming a person is unnecessary for the achievement of a relevant type of goal, that person cannot be liable to be harmed’ (McMahan Reference McMahan2009, 9). The point about how liability relates to necessity is important. Even if someone is liable, they cannot be intentionally targeted unless doing so is a requirement to achieve a just aim. Seth Lazar writes that a necessary action is ‘one that best contributes to [a just] victory while minimizing non-liable suffering’ (Lazar Reference Lazar2010, 208). Daniel Statman writes that necessity requires choosing the action ‘less harmful than all these alternative acts’ (Statman Reference Statman2011, 436). Necessity holds that it is permissible to kill only the minimum number of liable individuals required to achieve a just aim.

It requires asking questions such as the following. Is it necessary to target these tanks to achieve (a just) victory? Must we target the bomb-making factory? Must we destroy the anti-aircraft guns? Sometimes killing a high-ranking official is the only person that is strictly necessary to alleviate an objectively unjust threat because of how hierarchical and how strictly organized command and control structures of political and military organizations are. Again consider Syria. If instead of viciously attacking Syrians in a desperate attempt to remain in power Assad had announced that he would transition to democratic institutions and hold elections within 2 months of when the Arab Spring began, few if any Syrians would have likely been unjustly killed. By the same token, if he were killed and replaced by someone willing to enact those reforms, that would likely be the only steps necessary to achieve the just cause of protecting innocents from objectively unjust harm. Even if a just war could be fought according to reigning just war principles, no war would have to be fought. Therefore, according to the principle of necessity, no war would be justifiable if a just assassination were all that were required to end the unjust threat.

Despite the expansion of democracy over the 20th century, political power remains highly concentrated in almost every state throughout the world as well as within many non-state organizations. The result of this concentration of power is that sometimes assassinating a leader may be all that is required to end a threat.

Last resort

There are a number of ways to interpret the jus ad bellum principle of last resort. I argue that both of its two main interpretations require just assassinations instead of a just war or other more harmful policies. The basic idea of last resort holds that because generally wars are the policies that harm the most innocents, even when there is a just cause, other policies that would be less harmful to innocents should be tried first. A literal interpretation of last resort holds that all other policies must be exhausted before war can be just. In this straightforward understanding of last resort, when there is a just cause and according to the other principles of just war theory, assassination must be attempted before any war could be just. The problem with this interpretation is twofold. First, one can never technically reach the end of one's options. Something more could always be tried, or existing policies could be given more time to work. The second problem is that such a strict interpretation of last resort could violate other just war principles such as proportionality. Not all other options to war would always harm fewer innocents than war. Broad economic sanctions are one such example that could be disproportionate and more harmful than certain types of limited warfare.

A more plausible understanding of last resort ‘forbids war if its benefits, although significant, could have been achieved by less destructive means such as diplomacy’ (Hurka Reference Hurka2008, 127). Proportionality is closely connected to last resort. As Thomas Hurka argues, there must be ‘a separate proportionality assessment for each alternative to war, subtracting its relevant harms from benefits, and count the war as a last resort only if its net effect is better than that of any alternative’ (Hurka Reference Hurka2008, 129). Typically war is juxtaposed to diplomacy and other polices that are often considered less harmful and less destructive. Yet diplomacy is not the only means that is less destructive than war. Just assassinations are another less destructive means that should be tried in situations where they are likely to succeed, before resorting to a just war. In other words, the last resort precept requires just assassination instead of war whenever the assassination would achieve the just aims of the war.

This discussion of proportionality, necessity, and last resort all show that just assassinations are required instead of starting what would widely be considered a just war or imposing other ‘peaceful’ policies that would result in serious harms to greater numbers of innocents than the just assassination.

Non-military means of posing objectively unjust threats

So far, I have concentrated on military means of posing objectively unjust threats. I now turn to directly defend the claim that individuals can and have posed objectively unjust threats by non-military means. One example of an objectively unjust threat is the non-targeted sanctions imposed on Iraq after the first Gulf War until 2003. The United States lobbied for and the UN Security Council implemented non-targeted sanctions against Iraq that quietly killed hundreds of thousands of innocent Iraqis (Gordon Reference Gordon1999, Reference Gordon2006, Reference Gordon2010). Most of those killed were children. Children in general cannot be liable according to the culpability criterion, because they lack an adequate capacity to reason morally, and these Iraqi children in particular did nothing to make themselves liable. The sanctions were avoidable and objectively unjust.

One could object that Iraq sanctions were not objectively unjust on the grounds that they were proportionate because they kept Saddam Hussein from acquiring and using weapons of mass destruction. Hussein had a record of using chemical weapons to kill Kurds in the potentially genocidal Anfal campaign of 1987–88 (Power Reference Power2002, chap. 8). Such an objection is difficult to sustain because the burden of proof is on the objector that Hussein would have very likely killed more than half a million innocent people had he acquired WMD. The objector must furthermore show two things. First, he must show that Hussein could not likely have killed the same number of people with conventional weapons. Second, the objector must show that there no other options (including assassination) that would have killed and harmed fewer innocents. No one to my knowledge has argued these cases convincingly.

Hussein's record of previous posing objectively unjust threats did not justify posing objectively unjust threats against innocent Iraqis. His actions might justify other measures such as taking seriously any verbal threats that Hussein might have made, which could make him more easily meet the liability criterion than other leaders who might bluster but not actually implement their threats. It is thus very likely that the realized threats against innocent Iraqis were objectively unjust (even though those who imposed them did not intend them this), and therefore made the people imposing them liable. Again, this does not mean that those who imposed the sanctions should have been assassinated because the sanctions could have been ended by less harmful means and because assassinating the architects of the sanctions might not meet other just assassination precepts.

Individuals who purposively starve people to death may be justifiably targeted for assassination as well, because this act is objectively unjust and avoidable. Individuals like Mao who knowingly impose foreseeable and avoidably policies that result in the starvation deaths of numerous innocents clearly make themselves liable (Waal Reference Waal1997, 18–19; Dikötter Reference Dikötter2010, Reference Dikötter2011). The same might be true of high ranking Ethiopian politicians who admitted to using food as a ‘major element in’ their ‘strategy against the secessionists’, which resulted in a famine that killed over 1 million people in the mid-1980s (Meredith Reference Meredith2005, 343). Politicians can be culpable for enacting objectively unjust threats in non-military and non-violent ways as well as violent ones, as these examples show. I turn now to some remaining objections to my argument.

Objections

One significant objection to any policy that advocates assassination is questioning the empirical claim that just killing one or a handful of people can change policy. Governments and many non-state groups are complex systems that are often intentionally constructed so that any individual's death would likely have little affect on policy. There are several responses to this objection. First, if an assassination were not likely to annul an objectively unjust threat, it would be morally impermissible because it did not meet the reasonable chance of success criterion. If killing a leader were not necessary to achieve a just cause, an assassination would be prohibited (although other policies including war might still be permissible). Second, the assessment should not be made regarding just the most powerful politician. Just war theory precepts do not impose a limit to the number of people who can be legitimately targeted if they are liable and it is necessary and proportionate to kill them. If it were necessary to kill 50 people in order to change a policy, this would be permissible assuming they were all liable. Third, this claim is empirically false, at least for some leaders. Major changes in policy occurred when individuals such as Hitler, Mao, Mussolini, Pol Pot, and Stalin both took and fell from power. Leaders matter.

Another set of objections is that permitting any assassinations would perniciously undermine the system or society of states (ICISS 2001, 48). Scholars such from the English School might make such an objection (Bull Reference Bull1977). This is a serious concern because for all its problems, Steven Pinker and Joshua Goldstein argue that current system of states has likely reduced violence in general and war specifically more than any other architecture of governance in human history (Goldstein Reference Goldstein2011; Pinker Reference Pinker2011). Permitting just assassinations might inadvertently undermine the existing international arrangement in which assassination is generally prohibited (even if the rule is sometimes broken). Finally, some might argue that those who disagree with the limitations just war theory imposes on the use of assassinations may co-opt the use of assassinations for their own nefarious ends. The result of this could be a more chaotic and unjust international society.

There are several responses to this multipronged objection. First, this empirical concern lacks supporting evidence. Leaders have already unjustly assassinated individuals without a breakdown in the system of states. It is unclear why just assassinations would likely do so if unjust assassinations have not undermined the system. Especially because information about liability and how to avoid making oneself liable should be made publically available, and because just assassinations should obviate some wars which could be far more disruptive to the international system, this concern seems overstated. Second, the logic of the objection is suspect. Other institutions and policies have recently challenged fundamental aspects of the international system without concomitant breakdowns of international society. For instance, both the International Criminal Court (ICC) and the reformulation of state sovereignty by commissioners of the ICISS who developed the responsibility to protect (R2P) limit state sovereignty in important ways. The ICC and R2P have been widely accepted by state leaders themselves. Furthermore, recent humanitarian interventions such as those in Kosovo (which did not have UN Security Council approval) and Libya (which did have UN Security Council approval) have not problematically undermined the current system of states. Indeed some from the English School such as Wheeler argue that legitimate humanitarian intervention strengthens rather than undermines the international system (Wheeler Reference Wheeler2000, chap. Conclusion). There is an important difference between problematically undermining the current state system and morally important reforms to the international system that state leaders themselves can come to accept in order to better protect innocents. Third, the potential of disrupting the international order must be weighed against the horrific polices of widespread human rights abuses that the system now allows all too frequently, and the potential alternate polices that often visit unnecessary harms and destruction on innocents while promoting a just cause. The potential benefits of just assassinations are great and could far outweigh such concerns. Not assassinating leaders such as Hitler, Stalin, and Mao resulted in some of the most calamitous outcomes in human history. Arguably in the case of Hitler, not assassinating him actually undermined the state system in Europe. Fourth, consider the objection that if the prohibition on assassination were weakened, someone like Assad might apply his own principles to when and why assassinations are justifiable, or, more likely, assassinate people when it suited his interests without considering moral justifications. There is no good reason to think that overturning what might be generally accepted as a settled norm (Frost Reference Frost1996, 105–112 and passim) on the prohibition of assassination would result in the increased use of unjust assassinations. Notice that this same problem already exists for humanitarian intervention and we have not seen a catastrophic breakdown of the international system. This is despite that even some actors such as Russia's 2008 invasion of Georgia have abused new norms such as R2P for their own ends. Actors could and do already provide reasons for their unjust wars and unjust assassinations that are inconsistent with the just war theory tradition. Changing a norm from something being completely impermissible to make it sometimes permissible under certain circumstances does not make it always permissible for whatever reason. Additionally, in so far as international law in general and international criminal law specifically has a deterrent effect (Hafner-Burton and Tsutsui Reference Hafner-Burton and Tsutsui2007; Keith et al. Reference Keith, C. Neal and C. Poe2009; Hill Reference Hill2010; Kim and Sikkink Reference Kim and Sikkink2010), the general prohibition of assassinations may provide some checks against the abuse of assassination especially if the aggression amendment to the ICC enters into force and its judges interpret to include unjust assassinations. Finally, even if making assassinations morally or legally permissible made their abuse more likely, it still might be preferable to the status quo because even with some abuse, permitting just assassinations might avoid some wars or severely harmful non-violent policies that wronged large numbers of innocents.

A third objection that Erica Chenoweth and Maria Stephan might raise is that because non-violent campaigns are about twice as likely as violent ones to meet the organizers’ goals of regime change and expelling foreign military occupation, the just causes that would permit just assassinations can generally be achieved by other means that would harm fewer innocents (Chenoweth and Stephan Reference Chenoweth and Stephan2008, Reference Chenoweth and Stephan2011). This inference does not follow. They focus on traditional campaigns of violence against security forces and regime supporters, not assassination. Their findings may undermine a just war strategy against some regimes posing objectively unjust threats. But it does not undermine the argument for just assassinations. Another reason it does not undermine the argument here is because there is no study comparing the proportionality of non-violent campaigns and just assassinations. Assassinations may results in harms to fewer innocents. Finally, the outcomes Chenoweth and Stephan study do not exactly coincide with the just causes discussed above.

A fourth objection is that just assassinations can problematically undermine collective self-determination and state sovereignty. Imagine a people elect a politician that who commits acts that meet the just assassination criteria while carrying out the will of the people. There are several responses to this argument. First, my argument is not limited to state leaders. My argument applies to non-state actors such as rebel leaders just as much as it applies to state leaders (Finkelstein et al. Reference Finkelstein, Ohlin and Altman2012). According to my argument, state leaders could permissibly assassinate rebel leaders if the killing met the just assassination criteria. Second, even if an international institution authorizes a just assassination, nationals from the state in question could be empowered to carry out the assassination, thereby avoiding the critique that state sovereignty is violated. Finally, whether one adopts a Walzerian collectivist argument about when violating state sovereignty is permissible or one adopts an account based on violations of human rights such as Luban's and Altman's, the critic worried about violations of state sovereignty could adopt whichever position she prefers on when it is permissible to violate state sovereignty to carry out a just assassination (Walzer Reference Walzer1977, Reference Walzer1980; Luban Reference Luban1980a, Reference Luban1980b; Wellman Reference Wellman2012). Another way of putting this it that the same just war theory precepts that permit violating state sovereignty to fight a just war or impose some other harmful policy also permit a just assassination.

A fifth objection is that because the outcomes of assassinations are inevitably uncertain, assassinations should be prohibited lest the assassins unintentionally cause a contemporary equivalent of the assassination of Archduke Ferdinand that was a precipitating cause of WWI. This is a serious but not a decisive objection. It is not decisive because there are good reasons to believe that sometimes violence against one or more individuals is the optimal policy given available options. The just war theory tradition provides guidelines in order minimize the chance of exactly the sort of catastrophic outcome of the assassination of Ferdinand. It allows and indeed requires taking reasonable risks under some circumstances in order to achieve just aims. The uncertainty of worse results occurring from not acting must be weighed against the results of acting. Furthermore, there are good reasons to favor assassinations over other policies including just wars because the risks to innocents are generally far lower with assassinations.