Political Violence: The Problem of Dirty Hands

Abstract

This paper argues that the reason why political leadership often involves dirty hands is because of its relationship with violence. To make the case, it maintains that violent means create and assert a form of dominating power that is in tension with the proper ends of political action. This power casts a wide shadow, frequently dominating large numbers of non-targets and empowering unscrupulous agents. On the other side of the balance, characteristically political justifications for violence are ‘supra-moral,’ meaning that they are motivated by the value of a conception of morality taken as a whole (or, indeed, morality as such) rather than by any particular moral value. The weight that ought to be given to such ends is indeterminate in a way that makes uncancelled remainders arising from the evil of violence likely in many cases.

We must do; and one of two things must we do…

Herman Melville, Billy Budd

…the bookkeeper doesn’t exist who could strike such a moral balance.

Michael Walzer, ‘Emergency Ethics’

1 Introduction

In this article, I defend a neo-Weberian account of dirty hands, upholding some of Max Weber’s conclusions about politics, violence, and ethics, while accounting for them in novel ways.

In ‘Politics as a Vocation,’ Weber juxtaposed the ethics of responsible political leadership with an ‘unworldly ethic of love’ sourced in Christian teaching which enjoins the agent to ‘“resist not him that is evil with violence”.’ By contrast, Weber argued, the politician’s responsibility is such that they ‘must abide by the opposite commandment: “You shall use force to resist evil, for otherwise you will be responsible for its running amok”’(Weber 2004: 82). ‘Force’ is indispensable to politics and, by taking up the calling, the politician ‘enter[s] into relations with the satanic powers that lurk in every act of violence’ (Weber 2004: 90).¹ Weber’s view is that, while such means might sometimes be used to good effect and, indeed, might sometimes be necessary, they are also prone to produce ‘evil side effects’. The problem of dirty hands arises from the fact that ‘no ethic in the world can say when, and to what extent, the ethically good end can “justify” the ethically dangerous means and its side effects’ (Weber 2004: 84).² The frequent reliance of political actors on violence means that uncancelled remainders will be relatively common.

Like Weber, I argue that the dirty hands problem is a more frequent (if not quite quotidian) feature of politics rather than an extraordinary matter occurring only in ‘supreme’ emergencies. Secondly, I argue that moral contamination is closely associated with violence per se; it doesn’t only occur when agents resort to more extreme forms of violence such as torture and terrorism.³ Thirdly, while the argument that violence tends to contaminate suggests that dirty hands might occur quite often, I maintain that the problem is nevertheless closely bound up with politics in particular rather than with moral life in general.

I set out my case in the following parts. Part 2 argues that prominent explanations of dirty hands can be sorted into four categories: intra-moral, inter-moral, supra-moral, and extra-moral. These explain the occurrence of dirty hands with reference to conflicts, respectively, within morality; between moralities; between moral duties and the need to defend morality as a whole; and between moral and non-moral sources of normativity. Distinguishing them provides a frame of reference within which to situate the neo-Weberian account I offer in subsequent parts and helps specify the way in which it justifies dirty hands decisions. In part 3, I build on a novel account of the nature of violence to argue that it is violence and its relationship with power that commonly renders political action morally problematic, particularly when used at scale by or against the state. Weber’s identification of violent means with dirty hands makes best sense, I maintain, if we interpret it in light of what Jonathan Leader Maynard calls ‘empirically constitutive political realities’ (Leader Maynard 2022). These are empirical features of political life that have significant normative consequences. So, without necessarily postulating a form of political normativity distinct from morality as such, the neo-Weberian view suggests an important way in which the normative demands of political action depart in important respects from those usual in other practical domains. In part 4, however, I argue that the most characteristically political justifications for resorting to violence are supra-moral: the reason for action is to contest (to assert, challenge, or defend) a system of values as it may be instantiated socially and legally in a particular political order. Part 5 concludes by indicating ways in which means and ends as I interpret them are likely to be incommensurable—especially where the reasons for resorting to violence are supra-moral—and, hence, why definitively cancelling out the contamination of violent means is often impossible.⁴

2 Four Types of Dirty Hands Theory

The expectation that normative conflicts will give rise to dirty hands can be accounted for in four different ways. I call these intra-moral, inter-moral, supra-moral, and extra-moral respectively. All four occur in the dirty hands literature. The same author will sometimes lean towards one account, sometimes towards another.

2.1 Intra-Moral Dirty Hands

The idea of intra-moral dirty hands takes morality to be a singular code or system of values and rules and to be, if not necessarily comprehensive, at least normatively dominant. That is to say, all normatively significant decisions ought to be acceptable at the bar of morality at a minimum. But whereas a theory of this sort recognizes the existence of only one morality, one way of viewing the problem of dirty hands sees it as arising from the fact that morality is heterogeneous. While its deontological constituents prohibit certain kinds of action, for instance, it also contains a consequentialist commitment to maximise the good and, in doing so, to prevent harm. To arrive at the view that dirty hands are possible, a theory of this sort must suppose first, that these different elements can sometimes make opposing claims; second, that the reasons arising from claims on one side will sometimes defeat the action-guiding salience of those on the other; and third, that the defeated reasons will nevertheless remain standing as normative claims in some further sense. Psychologically, these non-action-guiding oughts will typically assert themselves through remorseful feelings.⁵ Or it might be that, following Walzer’s 1973 account, they should be reasserted legally and might even justify punishment.

In some ways, this conception of dirty hands has proven hardest to understand and most vulnerable to attack from philosophers sceptical about the idea of uncancelled moral remainders.⁶ On the face of things, it also seems less likely than rival accounts to which I will turn to specify why this is specifically a problem for ‘political action,’ as is sometimes suggested.⁷ Conflicts between moral duty and responsibility for consequences might arise more often or in more dramatic ways in politics, but they aren’t distinctively political. Medical ethics seems likely to pose problems of this sort, for instance, and so too might the ethics of running businesses and managing employees. Politicians perhaps more often face consequences at scale, but the idea that dirty hands is distinctively political is more often taken to say something about the qualitative nature of politics than merely about quantities.⁸

2.2 Inter-Moral Dirty Hands

Inter-moral dirty hands occur when an agent is torn between the claims of multiple moralities. Belief in such a possibility rests on a series of assumptions: first, that multiple moralities exist; second, that they are self-contained, rival (as opposed to complementary) ways of mapping out the same (or overlapping sets of) ethical questions; third, that sometimes the same individual is nevertheless answerable to more than one morality; and finally, that even if principled choices between two of them are possible in cases where they conflict, both moralities retain their claim on someone forced to choose. Were these assumptions true, then it would be more immediately clear than in intra-moral theory why choices could result in dirty hands since it’s not unnatural to suppose that the moral claims of one discrete morality would be incapable of defeating the moral claims of another.

Something like this idea is explored in Greek tragedy. In the most influential example, the eponymous Antigone is forced to choose between the claims of an ancient ethos of dedication to family which Sophocles identifies with the ‘divine’ law, and the ‘human’ law of the city set down by the Theban king, Creon. When she defies Creon’s decree and buries her brother after his death fighting against Thebes, Antigone renders herself guilty before the state: to do right by one ethic, she must do wrong by another. On Hegel’s reading, the protagonists’ opposing worldviews will ultimately be reconciled dialectically under a higher, more universalistic understanding of ethics. But this doesn’t save them from the one-sidedness of their ethical commitments which are both inescapable and, so far as individual action is concerned, irreconcilable (Taylor 1975: 175; Sophocles 1984).

Isaiah Berlin reads Machiavelli as a proponent of inter-moral conflict who thinks that political leaders are torn between ‘two incompatible ideals of life, and therefore two moralities. One is the morality of the pagan world […] Against this moral universe […] stands in the first and foremost place Christian morality’ (Berlin 2013: 289). These present respectively ‘exhaustive alternatives between two conflicting systems of value’ (Berlin 2013: 303).

Another modern variant of inter-moral dirty hands can be found in Marxist thought. Georg Lukács argues that the characteristic ethical challenge of revolutionary action arises when political actors are compelled to violate the prohibitions of a currently hegemonic morality in the name of a future morality, that of socialism and ultimately communism. Consequently:

ethical self-awareness makes it quite clear that there are situations—tragic situations—in which it is impossible to act without burdening oneself with guilt. But at the same time it teaches us that, even faced with the choice of two ways of incurring guilt, we should still find that there is a standard attaching to correct and incorrect action (Lukács 2014: 10).

That standard is offered by a duty of ‘sacrifice’: the ‘imperative of the world-historical situation’ demands a sacrifice ‘not only of [the revolutionary’s] life, but also of his purity, his morals, his very soul’ (Lukács 2014: 11).⁹

But if an action is impermissible in relation to the standards of one, now dominant, morality, it might yet be justifiable in terms of a rival, future morality in two different ways. First, an act regarded as evil under bourgeois ethics might be a virtuous one under socialism: e.g. a property owner who defended their possessions might be regarded as innocent under the former, but as a wrongdoer liable to remedial violence under the latter. So the same act will have multiple complexions, each determined by a different morality.¹⁰

Second, instead of being directly mandated by the morality of a future socialist society, revolutionary violence might be required only instrumentally, violating today’s morality as the necessary means of bringing about change.¹¹ This being the case, a necessary act of violence might be permissible by neither bourgeois nor socialist morality so that the revolutionary dirties her hands by violating both. But even while she can’t claim to be acting in the name of some principle occurring within a particular conception of morality, she might act for the sake of securing morality—or a conception of morality—as a whole: the morality of the future. If so, then the justification is supra-moral, which takes us to the third type of conflict.

2.3 Supra-Moral Dirty Hands

After his classic 1973 article, ‘Political Action: The Problem of Dirty Hands,’ the focus of Walzer’s theory shifted from more routine dirty hands cases to cases of ‘supreme emergency.’ This term refers to the circumstances in which Britain’s leaders decided to bomb German cities during World War II when faced with defeat by Nazism (Walzer: 251–68; 2004: 33–50).

Different descriptions of the evil Britain sought to forestall suggest different ways of capturing the underlying normative conflict with which its leaders grappled. One thought, for example, might be that the sheer numbers of innocent people under threat of violent death and enslavement by a regime like the Third Reich could warrant immoral means of defence if necessary.¹² But, if so, such grave but still finite, quantifiable evils wouldn’t require stepping beyond the first, intra-ethical type of theory.¹³ Walzer therefore suggests that supreme emergency isn’t defined by the sheer number of people at risk, turning instead to something best captured in communitarian terms. The thought, secondly, that the evil defining a supreme emergency is the defeat of a political community by an aggressor and its loss of independence has been criticised for leaving the door open to any state faced with defeat by an aggressor to raise the red flag.¹⁴ But Walzer also alludes to a third possibility which is more distinctive and perhaps more plausible. At stake in supreme emergency is morality itself or, at least, a particular conception of morality as it is embodied and carried forward in the values and practices of a particular community: ‘A supreme emergency exists,’ as Walzer writes, ‘when our deepest values and our collective survival are in imminent danger…’ (Walzer 2004: 33). Such a threat is not merely a matter of losing ‘territorial extension or governmental structure,’ but confronts the community with the possibility of terminating ‘its ongoingness;’ this jeopardises not only ‘a set of particular lives’ but ‘a way of life’ (Walzer 2004: 43; 2000: 252 − 55).

In such cases, morality is threatened from two directions at once: on one side, the Nazis threaten morality itself by attempting invasion of states in which it thrives and replacing it with national socialist values. Provided we are content to assume that ‘morality’ in any reasonable sense must exclude whatever Nazi ethics might look like, this characterization of the emergency seems reasonable. On the other side, morality is also under threat from those who seek to defend it. If those defending Britain did so by means ‘no less immoral’ than those that set Nazism out as something beyond the pale of reasonable conceptions of morality, then there was a danger that morality itself might be destroyed from within: such actions ‘put morality itself at risk’ (Walzer 2004: 33).

A justification for dirtying one’s hands for the sake of morality as such is what I characterise as supra-moral. Characterising it this way helps defend Walzer’s supreme emergency account from criticisms such as Tony Coady’s, that it ultimately resolves the problem into a decision occurring ‘within morality’ to pursue ‘what it is right to do, all-things-considered’ (Coady 2008: 82, 80, emphasis in the original). It is supra-moral in that the justificatory reasons for action don’t come from within a particular conception of morality. Instead, the justification is, in a sense, above morality because the end of action is morality as such.

Different ways of construing the term ‘morality’ are possible and different accounts will envisage an existential threat to morality in correspondingly different ways. On a communitarian view, for instance, morality might be thought to exist concretely in the values, habitual practices, and ways of life of one or more communities; destroy those communities, cancel their practices, and replace them with a nihilistic ideology and you thereby destroy morality in fact. Morality couldn’t meaningfully exist, on such a view, in the abstract. A Kantian view might place more emphasis both on the state as a public source of determinate norms and on law as a positivised expression of the most important principles of moral right (Ripstein 2009).¹⁵

2.4 Extra-Moral Dirty Hands

For the fourth possibility, let’s turn back to Walzer’s 1973 account. In that piece, Walzer’s reading of Machiavelli describes the good prince who learns how to be bad as one whose ‘political judgements are indeed consequentialist in character, but not his moral judgements’ (Walzer 1973: 175).¹⁶ On the face of it, this suggests two things: first, ‘his moral judgements’ are non-consequentialist; second, his consequentialist judgements are non-moral. Consequentialist reasoning is not, then, moral reasoning and, I suppose, our responsibility for consequences is not, then, moral responsibility.¹⁷ Moreover, the non-moral reasoning of consequentialism is specifically political. Likewise, when characterising Weber’s approach to dirty hands, Walzer identifies morality with the Christian doctrine of love and describes the politician as one who ‘views this conflict when it comes with a tough realism, never pretends that it might be solved by compromise, chooses politics once again, and turns away from love’ (Walzer 1973: 177, emphasis added).¹⁸ And in his discussion of Camus, Walzer’s comparison between dirty hands and civil disobedience contrasts politics and morality again: ‘In most cases of civil disobedience the laws of the state are broken for moral reasons […] In most cases of dirty hands moral rules are broken for reasons of state…’ (Walzer 1973: 179, emphasis added). ‘Reasons of state’ seems to be a synonym here for ‘political reasons.’

What this suggests is a fourth way of framing dirty hands: rather than arising from conflicts within morality, between moralities, or in defence of morality as such, extra-moral dirty hands occur when the normativity of morality comes into conflict with norms arising from another, non-moral but nevertheless compelling source, and where it is not clear in advance that moral norms always trump non-moral norms.¹⁹ Morality, on this view, is neither comprehensive nor dominant.

For reasons I will turn to in part 4, the view of dirty hands that I propose sits most comfortably in the supra-moral category. But, as part 3 will show, it also owes something to the idea of extra-moral justification.

3 Dirty Hands and Violence

With regard to the idea of extra-moral dirty hands, my argument is that dirty hands don’t arise from a clash between morality and distinctly political norms but from the political actor’s need to engage with certain empirical phenomena that frequently generate peculiar normative problems for political action (Leader Maynard 2022).²⁰ The most important phenomenon is the one Max Weber emphasises: the practice of violence.

I quoted at the outset Weber’s suggestion that the ‘morally dubious or at least dangerous’ nature of violent means had to do with generating ‘evil side effects.’ For a just war theorist, this phrase will bring to mind foreseeable but unintended harms to civilians during war (Walzer 2000: 151 − 56; Neu 2013), but I don’t think this will help us capture what Weber alludes to. Instead, I argue, we need to turn our attention to something else: the evil of domination. Violence dirties political agents’ hands when it projects a destructive power that not only dominates its intended targets but also generates an excess domination threatening many others.

3.1 The Shadows of Violence

I have argued elsewhere that violence is co-constituted by a dual will: on the one hand, agents of violence aim at destructive harm to their targets; on the other, they aim at destructive power over them (Finlay 2017).

By ‘destructive power,’ I mean an agent’s ability to inflict destructive harm on a target at will and, by the same token, the vulnerability of the target to that will. This power relation is established in paradigm cases of violence by adopting a technology and a technique of employing it that are together designed to defeat, evade, or eliminate the victim’s opportunities for escape or defence, thereby permitting the infliction of destructive harm. Technologies might be as primitive as a fist or as sophisticated as an unmanned aerial vehicle. Violent means like these commonly achieve destructive power by combining elements such as velocity, secrecy, distance, and surprise.

Hence, violence may be defined in the following way:

Violence is defined by the presence of Violent Agency consisting of the intentional infliction of (1) destructive harm by human agents on a target using a technique chosen with the further intention (2) of eliminating or evading the target’s means of escaping it or defending against it. In paradigm cases of violence by single-minded attackers, (2) will be realised as far as is necessary to secure (1) or, failing that, as far as possible to maximise the chance of doing so (Finlay 2017: 73).

In typical acts of violence, establishing destructive power is a necessary means of achieving destructive harm, though, as I’ll suggest in 3.3, their roles can be reversed in special cases.

Moreover, destructive power is distinguishable from the coercive power that it might be used to establish. I might aim a gun at someone, employing distance or secrecy plus velocity to render them defenceless, and shoot them dead without making any demands. Or I might leverage the same destructive power over them to compel them to obey me, i.e., to coerce them. In both cases, I may be said to have power over them, but of different kinds. Coercive power is a secondary form of power which may be derived from the primary power to destroy.

When Weber writes of the ‘power behind which violence lies concealed,’ I take him to mean secondary, coercive power.²¹ In his analysis, the ultimate guarantee of the coercive power at the heart of politics is the means of violence.²² My claim here, by contrast, is that behind (or within) violence itself, another, more primitive form of power ‘lies concealed.’ Insofar as this ‘destructive power’ describes cases where an agent has the power to interfere with someone in ways prejudicial to that person’s basic interests and without being constrained from arbitrarily exploiting that power, we can follow republican political philosophers and call it a form of ‘domination’. (I will come back to the question of how arbitrariness might be defined in 3.2.)

One way in which excess domination occurs is when the attempt to mount a violent campaign against a certain set of targets entails the domination of non-targets too. Not only, in other words, are non-targets faced with a higher-than-normal probability of severe harm, but the reason for this is that they are subject to the reality (not the mere probability) of being subject to destructive power. The enemy threatening them has it in their power—whether they choose to exercise it or not—to inflict destructive harm on them. This is true whether any non-targets actually come to be harmed or not: those parties are in fact rendered vulnerable to the possibility of violent death or such lesser harms as the agents of violence choose to inflict. This is a common feature of warfare but also characteristic of a wider array of violent practices.

Sometimes acts of violence are relatively simple one-on-one matters. Often, however, they involve more complex systems of dominating power relations. Once we move beyond the simplest cases, we are likely to find that the use of violence, first, exposes more people to the arbitrary decision of the agent of violence and, secondly, does so in a way that is less likely to be deniable as part of the agent’s intention.

To illustrate, consider the following cases:

• Pistol An assassin locates her victim, approaches, and fires a pistol at close range only once she has confirmed their identity beyond question.

• Rifle A sniper tracks her victim’s movements and positions herself at a vantage point from which to shoot her appointed victim when he appears at the exit of his place of work.

• Missiles Missiles and artillery are positioned on Israel’s territory in such a way as to be able to project destructive force into any part of Gaza. Once a suspected target has been identified, operators can fire a missile or shell.

• Drones A fleet of drones is used by US forces to maintain a permanent presence loitering over an Afghan province where suspected terrorist militants are hiding. Once a likely terrorist is identified, operators can fire a missile at them.

In all four types of case, we can assume, the agent of violence seeks destructive power over the intended targets. Arguably the assassin in Pistol could deny inflicting excess domination—i.e., domination of non-targets. In Rifle, however, excess domination is highly likely—just as long as any other person than the target crosses the sniper’s firing line. The sniper intends to kill the target; but, to do so, she intentionally puts herself into a position where she can execute harm or not according to her own will against anyone passing a particular location. In case of a non-target doing so, she is likely to decide against exploiting her destructive power; in case of the target doing so, she may decide in favour.

The sniper might want to argue that the excess domination is a foreseeable but unintended side-effect. She might say that she intends to dominate the target, i.e., to position herself so as to be able to harm him (or not), but that, by occupying her vantage point, she also foreseeably but unintentionally positions herself to kill any other person (or not) who appears in the doorway before the target exits. But I think this description would be disingenuous. It underplays the fact that the assassin has, in fact, put herself into a position of power over anyone passing the relevant spot and not just the target. This power is exercised both in shooting the target and choosing (intentionally) not to shoot any non-targets who appear first. She is equally in a position to shoot either, target or not.

The decision to dominate non-targets (in vastly greater numbers) is even harder to deny in the third and fourth cases. In Missiles and Drones, weapons are positioned to be able to project destructive harms anywhere in the respective territories and thus dominate all persons and objects in each territory taken as a whole. In these types of case, operators can fire against anyone (or anything) betraying the signatures of a legitimate target.

This might occur in a range of different ways:

• (1) Where a genuine militant, militant vehicle, or group thereof moves into a visible position.

• (2) Where a person who occupied a civilian role changes role in a visible way and transitions to a militant or military role.

• (3) Where a position of unknown or indeterminate status or which appeared to be occupied entirely by civilian objects is identified as, in fact, a military target.

• (4) Where civilian persons or objects appear to be any of the above cases and are misidentified as legitimate targets.

Moreover, they are in a position as a result to inflict destructive harm on the following:

• (5) Civilians and civilian objects that don’t appear to be anything other than what they are.

Armed forces in Missiles and Drones should only intend destructive harms in cases (1)-(3); they should foresee (and seek to minimise) the likelihood of striking in cases of (4). But they set up the weapons systems in such a way as to empower agents to attack all five types of case: in order to be able to target (1)-(3), they are equipped to target (4) and (5). People falling into all five categories are equally (and intentionally) dominated.

So this is one way in which adopting the means of violence gives rise to excess domination. If it can be described as a foreseeable ‘side-effect,’ it can’t plausibly be characterised as an unintended one. But there is a yet wider penumbra to the dominating shadow of violence. For Weber, the focus for political ethics should not be on an, as it were, ideal case in which force is used only by suitably motivated, knowledgeable, and competent agents as directed. Instead, the ethics of responsibility recognizes that leaders rely on ordinary human beings and must harness ordinary passions. The responsible politician,

has absolutely no right to assume humankind’s goodness and perfection. He does not feel that he is in a position to shift the consequences of his actions, where they are foreseeable, onto others. He will say, “These consequences are to be ascribed to my actions” (Weber 2004: 84).

The ill consequences for which politicians take responsibility arise from passions such as vengefulness and base hatred on the parts of their followers as well as from their followers’ rapacity after they come to power by force (Weber 2004: 89–90).

On my account, the point to register here isn’t merely that bad people will do bad things due to bad motives; it is that the responsible politician who is compelled to authorise a resort to violence must recognize that, in doing so, she will have empowered bad people, equipping them with the means to act on bad motives. Equipping ordinary human beings with destructive power has the side-effect of enabling them to leverage that power coercively according to their will. The inevitability that organized, large-scale violence will generate opportunities of this sort deepens and widens further the shadow cast by violent means.

3.2 Demonic Means and Godlike Power

So the sinister complexion of violence isn’t fully comprehended by looking only at the harms it is designed to inflict. It requires a standpoint from which to encompass the primary, dominating power constitutive of violent means as well as their aptitude for generating the secondary, coercive power that Weber identifies with politics. But in what way, specifically, does the employment of violence leave traces of moral dirt?

The sort of power that a political leader becomes implicated in when she takes up the state’s ‘monopoly of legitimate physical violence’ has a deep affinity with the notion of ‘evil’ that some philosophers identify with dirty hands (Weber 2004: 33). But when they do so, they tend to narrow the focus of dirty hands theory to the most morally troubling subset of violent methods (practices like torture) rather than encompassing violent means more generally.

Matthew Kramer, for instance, characterises dirty hands as ‘a moral conflict that involves the prospect of evil conduct’. He describes as ‘evil’ that ‘conduct [which] is impelled by sadistic malice or heartlessness or extreme recklessness’ and which is ‘connected to severe harm in the absence of any significant extenuating circumstances.’ Naturally, it is ‘heartlessness’ rather than sadism or recklessness that has the most relevance to dirty hands decisions on Kramer’s view:²³ ‘Heartlessness,’ specifically, captures ‘a frame of mind in which someone’s lack of inhibitions about causing or witnessing other people’s adversities is due to his regarding their woes as instrumentally gratifying (rather than inherently gratifying)’ (Kramer 2018: 189).

Evil in this sense is exemplified by the ‘steely contract killer’ and the ‘cold-hearted torturer.’ As Kramer puts it, ‘[w]hen a heartless frame of mind is connected to the infliction of severe harms such as murder and torture, the actions which underly it are evil.’ Even if necessity might demand it, inflicting ‘gross harms’ while in a ‘heartless frame of mind,’ Kramer writes, is ‘incompatible with an elementary regard for other people as rational agents and as creatures whose basic interests include the avoidance of intense pain and the undergoing of positive experiences’ (Kramer 2018: 190).

I take Kramer’s aim here to be an account of ‘evil’ that captures the nature of extraordinary violence. Torture, for example, ‘is commonly offered as one of the few unproblematic examples of a type of act that is morally impermissible without exception or qualification’ (Sussman 2005: 2). And murder is wrongful by definition (albeit that there might be cases where not murdering is ‘an even more serious wrong’ than murdering) (Kramer 2018: 188). Indeed, the historical example Kramer uses to illustrate dirty hands is Truman’s decision to drop atomic bombs on Japan in 1945 (Kramer 2018: 194 − 95). I presume therefore that the category of ‘evil actions’ that Kramer intends will consequently exclude many cases of permissible killing, e.g., by a soldier fighting discriminately in a just war.

But, if so, then I doubt that the implicit moral distinction between extraordinary ‘evil’ means and ordinary, permissible violence will track ‘heartlessness.’ Indifference to the suffering of one’s target and, indeed, taking pleasure in the instrumentality of inflicting harm on them are as likely to occur in just war or legitimate policing as in cases of murder and interrogational torture. Imagine a police sniper shooting a marauding terrorist: surely the sniper too may take professional satisfaction (i.e. a form of ‘pleasure’) if their action succeeds in preventing further killing. The same is likely to be true of soldiers when they contribute successfully to a just, defensive war through legally and morally permitted acts of military violence. If that is commonly the case, then heartlessness seems likely to be shared by violence both in its extraordinary forms and in forms that are widely seen as straightforwardly justified.

I’ll set aside the question of how morally extraordinary violence might be set apart from other, seemingly similar, acts of violence until 3.3. For now, I want to emphasise the continuities between them. Kramer emphasises the grave harms that evil actions seek to inflict, harms defined against a basic human interest in the avoidance of pain and in the pursuit of pleasurable experiences. But he also highlights the heartless arrogance of the perpetrator by alluding to power. The assassin is someone who ‘wields [a] godlike power of life and death’ (Kramer 2018: 190). A nearly identical description appears in a recent analysis of drone strikes by Christian Enemark. Following Robert Sparrow, he writes that, ‘[i]f, instead of war, drone strikes resemble “a godlike power to call down destruction from the skies,” the rules for restraining such strikes would need to be derived from a different concept of violence’ (Enemark 2014: 368)²⁴. Like Kramer, Enemark seeks to distinguish a subset of violent methods by highlighting something that distinguishes an especially abhorrent or novel kind from violence in general.

My argument, by contrast, is that it is precisely this ‘godlike power’—rather than the harm it is used to inflict—that captures the distinctive ‘evil’ in which agents engage when employing violence. To Kramer’s ‘basic interests’ in avoiding pain and pursuing positive experiences I would add that there is a common and basic interest in not being subject to the arbitrary power of others. Here ‘arbitrary’ might be defined in various ways, one of them—following Kramer—by reference to heartlessness: someone who is heartless, on Kramer’s account, is not compelled to take your basic interests into account and disregards or discounts them in favour of some other value.

The notion of a basic human interest in being free from arbitrary power to interfere—and perhaps above all to do so destructively—is explained by Philip Pettit in terms of its importance for ‘subjective self-image and intersubjective status’: ‘The non-interference you enjoy at the hands of others is not enjoyed by their grace and you do not live at their mercy. You are a somebody in relation to them, not a nobody. You are a person in your own legal and social right.’ Absent this assurance, individuals suffer a disastrous loss of standing in the regard others and, consequently, in their self-respect: the enjoyment of non-domination in relation to another agent—at least when that agent is a person—goes with being able to look the other in the eye’ (Pettit 1997: 71).

The way in which what I’m calling the primary power inherent in violence specifically dominates and thereby compromises moral standing is captured in an extraordinary passage in Simone Weil’s essay, ‘The Iliad or the Poem of Force.’ In contrast to cases where force kills its victim (‘force in its grossest and most summary form’), she writes,

How much more varied in its processes, how much more surprising in its effects is the other force, the force that does not kill, i.e. that does not kill just yet. It will surely kill, it will possibly kill, or perhaps it merely hangs, poised and ready, over the head of the creature it can kill, at any moment, which is to say at every moment. In whatever aspect, its effect is the same: it turns a man into a stone. From its first property (the ability to turn a human being into a thing by the simple method of killing him) flows another, quite prodigious too in its own way, the ability to turn a human being into a thing while he is still alive. He is alive; he has a soul; and yet—he is a thing (Weil 2005: 184 − 85).

Like Pettit, Weil underscores the psychological, indeed spiritual, harm this sort of domination causes its victim:

An extraordinary entity this—a thing that has a soul. And as for the soul, what an extraordinary house it finds itself in! Who can say what it costs it, moment by moment, to accommodate itself to this residence, how much writhing and bending, folding and pleating are required of it? It was not made to live inside a thing; if it does so, under pressure of necessity, there is not a single element of its nature to which violence is not done (Weil 2005: 185).

Both Weil and Pettit thus capture from different perspectives what might be understood as the ‘godlike power’ attributed to different kinds of killer and, indeed, with the ‘satanic powers’ (and ‘demons’) that Weber associates with violence (Weber 2004: 86, 90). Doing so directs attention to life under the dominating shadow of the drone and, in perhaps less obvious ways, the sniper, the bomber, the artillery operators, the advancing infantry. In the face of any such threat, appearing to act like a target might get you killed; standing next to someone resembling a target might do so too. Those living under such regime will necessarily find themselves adapting their behaviours so as to try to appease the powers they are subject to: living ‘in fear of that other,’ they are forced to realise that the ‘non-interference’ they need in order to survive is ‘enjoyed by their grace’ and, because they are ‘at their mercy,’ they must avoid doing anything that might trigger a strike (Pettit 1997: 71).

This is to say nothing of unscrupulous, sadistic, negligent, or reckless behaviour on the parts of those controlling the weapons. But even if we postulate reliable, morally conscientious agents of violence, my argument is that wielding the means of violence in the organised, purposeful ways required by political leadership is inseparable from something like the attitude of ‘heartlessness’ by which Kramer seeks to distinguish ‘evil’ from merely bad acts. Heartlessness is a facet of violence because violence requires—is co-constituted by—a will to subjugate and to expose others to the possibility of destructive harm.

It might be objected that not all the actions of just warriors and legitimate police officers will be wholly arbitrary. If not, then some republicans might say that they don’t dominate. But, in the context of political action, violence has greatest utility when non-arbitrariness is hardest to guarantee.

Leading republican accounts of non-domination are divided between procedural and democratic accounts of non-arbitrariness. According to the former, ‘[p]ower is not arbitrary […] to the extent that it is reliably controlled by effective rules, procedures, or goals that are common knowledge to all persons or groups concerned.’ As Frank Lovett argues, ‘[r]oughly speaking, the procedural view equates republican freedom with the traditional idea of the rule of law, provided we are willing to extend the latter idea considerably’. By contrast, on the second view, ‘[p]ower is not arbitrary […] to the extent that it is directly or indirectly controlled by the concerned persons or groups themselves’ (Lovett 2022).

Either account could be cited to defend some more or less routine operations by police against, say, gangs of armed robbers. Properly conducted, they can occur as part of the rule of law and be held democratically accountable. By contrast, however, in cases where force is used for political purposes, it is less likely to satisfy procedural and democratic requirements. Take, for instance, the use of force to uphold the authority of the state against a domestic insurgency. Or consider, similarly, the defence of a state against foreign invasion. Or even, to turn things around, consider the decision to use force by the leader of a political revolt against illegitimate rule. In all three cases, the nature and proper sources of right—of procedure, law, and democratic will—are among the matters that violence is used to contest: each side uses force to establish the proper sources of authority from which to judge arbitrariness.

In any case, whether they occur in armed political contestations or routine police actions, violent means subject those trapped under their shadow to a peculiarly intense form of power which dominates at least in the sense that their agents arrogate to themselves the ability to inflict destructive harm on them at will. So perhaps this captures something of the moral taint that arises even in cases of legitimate violence. Both targets and, in many cases, non-targets of legitimate violence are dominated by a ‘godlike power’ which, even if it betokens harms to targets to which they are liable, is defined by heartlessness in a specific sense. And with this domination come the attendant moral and psychological evils of domination. The behaviour of those people living under the shadow of violence is reshaped pre-emptively around whatever appears likely to propitiate the vengeful pseudo deities that cast it.

3.3 ‘Death’s Whimpering prey’: Violence, Torture, Terror

The argument that it is violence as such that connects politics with evil involves pulling the lens back from the narrow focus on extraordinary means of violence that often dominates discussion. Yet it doesn’t preclude recognising that its most extraordinary forms are even more likely, all else being equal, to contaminate political action. My analysis of violence suggests that the reason why practices such as torture and terrorist bombing are even more likely to dirty actors’ hands is that they magnify what is present even in ordinary cases of violence. Their extraordinary status isn’t, therefore, due to being radically different from other forms of violence in some way.

This is important for dirty hands theory for two reasons. First, it helps highlight even more starkly the morally troubling nature of violence itself. Second, it indicates a further way in which committing to the use of violence in general implicates political actors in moral evil and dirties their hands. This is because the deep affinity between ordinary and extraordinary forms of violence means that admitting the former makes it a great deal harder for political actors to prohibit the latter.

I suggested earlier that, in paradigm cases of violence, the will to dominate is typically subordinate to the will to harm: the sniper establishes destructive power over their target in order to kill. However, cases of torture and terrorism frequently reverse this relationship. Both use destructive harms to highlight and thereby deepen domination by making their victims acutely aware of their tormentors’ destructive power.²⁵ As illustration, consider the account of sniper attacks in Aleppo during the Syrian civil war recorded by the surgeon, David Nott:

on some days there was a weird consistency to the injuries we saw coming in—the patients all seemed to have been shot in the same part of the body. One day we would receive patients who had all been shot in the left groin area; on other days six or seven would arrive shot in the right groin. The same thing was happening with patients shot in the upper limbs and chest—the injuries all seemed to be on the same side, in clusters. [W]e rarely saw the head shots that would have resulted in an instant kill (Nott 2019: 217).²⁶

Clearly, Nott reports, the snipers could kill but chose not to. He recalls hearing from another doctor that the snipers were playing a game with rewards for hitting different parts of their victims’ bodies (Nott 2019: 217). But, if this was in some sense a ‘game,’ it is likely to have had a very material purpose. The snipers were put in place to terrorise the population. Killing people outright would certainly have had that effect. But they chose to spell things out more dramatically. This was a display, as Weil said, of ‘the force that does not kill, i.e. that does not kill just yet.’ By shooting people in a patterned way, they demonstrated outright domination—an ability to inflict harm at will, to kill or not to kill on a whim. In effect, the snipers inscribed in the minds of the survivors (those who survived the gunshots, those who witnessed them) two faces of violence at once: the ruthless will to harm; but at least as important, the ruthless will to dominate.

Acts of terrorism and of torture share with the lethal violence more typical of warfare the same components: the will to dominate destructively; the will to harm destructively. Where they differ is in the way one will is subordinated instrumentally to another: to dominate in order to harm; to harm in order to dominate.

Terrorists extend a penumbra of destructive power over a much wider set of people than those they actually injure (Walzer 2004: 34). They use public acts of destruction to thrust the face of primary power (domination and vulnerability) to the foreground. Similarly, the emphasis on domination as an essential component of violence helps specify what should trouble us most about torture. Torture binds the victim and suspends the threat of destructive harm over their head. It might best be described a form of frozen violence, where, as Adriana Cavarero puts it, ‘violence is taking its time about doing its work’ (Cavarero 2009: 31). Violence isn’t something that the torturer uses only once the victim has been rendered defenceless; as the definition in 3.1 indicates, violence here is both the rendering of a person defenceless and the infliction of harm by means of the resulting power. But, for the torturer just as for the terrorist, the means-ends relation between the two faces of violence becomes ambiguous: power certainly paves the way for harm; but harm is also the means of asserting and deepening power.

As Jean Améry writes of his experience at the hands of SS torturers, ‘[t]hey wholeheartedly devoted themselves to their cause—power, domination of the body and spirit, extreme uninhibited self-expansion.’ Indeed, Améry captures the experience in terms redolent of those invoked in 3.2: ‘there were moments when I showed a kind of abject reverence for the torturous sovereignty that they exercised over me. For isn’t one who is authorized to reduce a human being to a body and death’s whimpering prey a god? Or at least a demigod?’ (Améry 2022: 26–27).

A third way, then, in which the dominating shadow of violence extends beyond the immediate aims of the political actor is when, by opening the door to ordinary violence, we struggle to shut it again when faced with exigencies that appear to demand its most extraordinary forms. Attempts to do so by arguing that they are essentially different in kind from ordinary violence tend to flounder. Torture, terrorism, and war share the same essential features: they wield primary, destructive power; they inflict pain, injury, and death deliberately by means of that power; and the combination of these two elements yields further power through coercion.²⁷ A commitment to the practice of violence is therefore apt to slither down a slippery slope towards the exceptional admission of torture and terrorism into the repertoire of politics.

4 Dirty Hands and Political Action

I want to turn now to the claim that the affinity between violence and the evil of dominating, primary power helps explain why political action and leadership are particularly prone to dirty hands. There are two thoughts to support this.

The first can be expressed in Weberian terms: it is the sociological claim that, even if violence has been regarded as legitimate in other forms at other historical times, it is presently the case that violence is closely identified with the state just as politics is. As Weber maintains in ‘Politics as a Vocation,’ the modern state is the monopolist of the legitimate means of violence and the preeminent locus of politics in modern societies. A revisionist or ‘cosmopolitan’ just war theorist might argue that anyone can justifiably use violence, given just cause, citing individual human rights (e.g. Fabre 2008). But, in present political circumstances, cases where a private person can justify violence are apt to be interpreted in light of the state’s pre-eminence. Someone self-defending against an assailant is permitted to do so only insofar as the police can’t save them; the protester or resister who resorts to violence does so permissibly only insofar as the legitimacy of a particular state, its configuration, or its government is doubtful (including cases where the state fails in its protective duties).²⁸ The point about the state is that its claim is universal. This doesn’t mean that each state aims at global monopoly; it means that the State is implicitly global in the claim it makes as the paradigm form of legitimate political ordering.²⁹

There is, however, a deeper point to make in reply to the same objection from private persons. Weber’s argument about the state isn’t only a sociological one; it is also a conceptual point about politics and violence in general. It’s not simply that (a) politics is associated with the state and (b) the state is associated with a monopoly of violence; therefore (c) politics is identified with violence. The position that Weber seems to be committed to holds that politics is identified with the means of violence; and then the state is identified with political leadership due to its monopoly of violence. As he puts it in Economy and Society, ‘[w]here organizations make use of force […] they must to this extent be called political organizations’ (Weber 2019: 136; 2004: 32 − 3). If this is the right way to read Weber, then he should be taken as asserting a closer, more enduring relationship: violence is the characteristic means of politics regardless of whether politics is sited in a monopolistic state or dispersed across plural associations. This being the case, then the question is, why is violence characteristic of politics per se?

To help identify a deeper link between politics and violence, it is useful to turn to Walter Benjamin. In his ‘Critique of Violence,’ Benjamin distinguishes two functions that violence performs in relation to political power: on the one hand, it can perform a ‘law-preserving’ function; on the other, it can be ‘law-positing’ (Benjamin 2021: 45–49). The distinction is quite suggestive once we combine it with the treatment of violence and power above in part 3.

On my account, violence is constituted by one kind of power—primary, destructive power—and, as such, it has an inherent potential to establish secondary, coercive power relations. Given this fact, if private persons forcefully challenge the state, then their violence, if it succeeds, can potentially posit new power relations, displacing the established relations of authority. Insofar as these new relations posit ends that differ from those of the incumbent authorities—new values, principles, and laws—we can characterise the force that establishes them as ‘law-positing’. Similarly, the violence by which the state defends established power relations, ends, and laws is ‘law-preserving’. Where I might go further than I think Benjamin does is in suggesting that the relationship here between violence and political—law-preserving or law-positing—potential isn’t a contingent one. Violence is potent in this way because of its dual relationship with power: the power to destroy; the power to coerce. To monopolise the (legitimate) means of violence successfully is to prevent others from positing law because it deprives them of the power-positing potential of violence itself; to challenge such a monopoly is to diminish the sphere of the legal dispensation it upholds and to open up potentially new avenues for law-making power.

This takes me to the second reason for thinking of political action as being prone to dirty hands problems through its relationship with violence. It has to do with the ends for which violent means are used politically.

Whereas in private life the aim of action is typically to secure various goods or values in a retail sense, the aim of the state (and political action more generally) is not only to do the same but also to secure core moral commitments expressed in law (value wholesale). To illustrate, we can distinguish between the retail act of defending a particular person from a particular attacker and the wholesale act of upholding a law prohibiting murder. When a private person uses force in self-defence against an isolated murderous threat posed by another private person, it is primarily a case of the first kind; when national armed forces prevent invasion by a foreign power that will eliminate the government, annul the law, and cancel the immunity of anyone suspected of political dissent from being murdered, it is clearly a case of the latter. The state, of course, is tasked with protecting individuals from both types of threat. But defending against the second one is political in a further, distinctive sense less likely to be seen in isolated cases of self-defence.³⁰ Again, actions for the sake of defending (or asserting) a conception of morality and law might also be undertaken against the will of the state and by private persons. Armed resistance or revolutionary action might both aim at such a result. If so, the actions of those private parties take on a public, political character.³¹

To the extent that violence aims to contest (assert, challenge, or defend) the system of values expressed in a conception of morality and law, its claim to justification will have an affinity with the supra-moral justifications introduced in 2.3. As I characterised them before, however, I identified supra-moral dirty hands cases with ‘supreme emergencies,’ where an entire system of morality and law was under threat from a wholly nihilistic force and, hence, where the survival of morality itself is in jeopardy. Such cases, to my mind, answer best to Walzer’s idea of supreme emergency. By contrast, a challenge to one particular system by agents of a rival set of moral values mightn’t constitute an existential threat in the same way as a challenge from a threat like Nazism. Yet defending against it might, nevertheless, have a supra-moral complexion: the end for which violence is used is an order of moral and legal values as a whole rather than any finite quantity of particular goods that might be defended under its laws.³²

This is not, of course, to say that violence should be used to create and shape political orders directly or that doing so justifies the involvement of political action in violence and dirty hands. Given its tendency to produce relations of power-over, violence might seem to lend itself to this.³³ But both liberalism and republicanism are centrally animated with the aim of liberating political spaces from the sway of force, spaces within which ethical commitments can be negotiated and established non-violently. In that perspective, the responsible political leader’s use of violence should be focused not directly on political creation but on defeating violence when it threatens political spaces for nonviolent action or stymies their creation.³⁴ But, even so, the justification for violence in the face of such threats remains supra-moral insofar as the aim is to uphold the system of values embodied in a polity’s institutions by defending its preconditions.

5 Conclusion: Dirty Hands and Calculability

Giving violence the attention it deserves helps correct an unevenness of focus in discussions of dirty hands, one with Machiavellian origins. Machiavelli highlighted injustice, lying, and promise-breaking in general as forms of badness that a prince must be prepared to learn and apply. Yet his discussion of violence focuses exclusively on the most extraordinary forms—not merely ‘force’ but ‘cruelty’—which implicitly normalises less extreme forms of violence.³⁵ Contemporary dirty hands theory has inherited a similarly low tolerance threshold for things like mendacity along with a high tolerance threshold for violence.

I don’t deny the possibility that injustice, lying, and promise-breaking might dirty the hands of those who resort to them, but I agree with Weber that violence is exemplary. Emphasising violence as such rather than narrowing attention to its most extraordinary forms leaves the door open to acknowledging how things other than violence might dirty a political agent’s hands. It is beyond the scope of this article to defend such a claim, but it might be that what unites these different forms of wrongdoing is that they arrogate power in various problematic, dominating ways likely to pollute the properly non-dominating orientation of political action.

To conclude, why, then, are there uncancelled remainders when politicians commit to the use of violence in politics? Remainders occur because of the relation between two factors.

On the one hand, as pacifist critics argue, the evil of violent means is apt to pollute any ends it is used to secure (Gregg 2018). The analysis of violence and domination shows why this sort of autotelic pollution is likely. As I argue in part 3, violent means are deeply problematic for any politics committed to eliminating domination since violence is impossible without asserting dominating power and it is prone to generate excess domination as a side-effect. This side-effect isn’t down-stream from the intended effects of using violence in the way collateral harms are in war; it is part of the means itself and therefore cannot be disowned as ‘merely foreseeable but unintended’.

On the other hand, the supra-moral form that justification takes in classic cases where violence is used for distinctively political purposes (as opposed to purely defensive usages against threats to particular goods) is inherently prone to indeterminacy. It is not possible, that is, to determine an acceptable moral price from within a conception of morality when the end for which costs are paid is that conception of morality itself. To use a phrase from Wendy Brown’s recent exploration of Weber’s ‘vocation’ lectures, how could we measure ‘the value of values’ (Brown 2023: 52) and weigh the result against costs measured in terms of any particular value? It is therefore unlikely that the evil identifiable with violent means can be cancelled in a determinate way by the good that political action seeks to secure. If Timothy Snyder is right to suggest that ‘the realm of values’ is ‘where possible futures are enabled,’ then the price paid now is gambled against hope in a possible future whose value cannot be reduced to terms directly commensurable with that price.³⁶