Violence and moral appraisal

What do we usually mean when we describe an act as violence? To begin with, in this sort of usage, violence is a moral term. There are other, non-moral usages: we describe certain kinds of non-human event as ‘violent’ – thunderstorms, for instance – as well as sometimes the purely physical characteristics of a human action [as Harris (Reference Harris1980, 14) suggests, one can stir a cup of tea violently]. But while these are valid, they do not rival the moral usage. Rather, the word serves different purposes in different contexts. I shall restrict my attention for the most part to the morally evaluative possibilities. Unless otherwise stated, where I speak of an act being ‘violent’, I mean it to be understood as ‘an act of violence’.Footnote ⁷

In his reflections on method in the history of moral language, Skinner distinguishes three different things we need to understand to be able to use a moral term like ‘violence’ correctly. First is the evaluative attitude that it can be used to express or invoke. In this regard, it is useful to compare ‘violence’ with the words ‘courage’ and ‘murder’. ‘Courage’ or ‘courageous’ cannot correctly be used as terms of disparagement or disapprobation. We could use them sarcastically, of course, but doing so is parasitical on their positive connotations. The word ‘murder’, by contrast, cannot correctly be used in a positive or even neutral way: one cannot, strictly speaking, murder justly; nor can a murder properly be a matter of moral indifference. By contrast, violence falls somewhere in between. Still a morally evaluative word, it holds up two possibilities: first, those means correctly described as ‘violent’ are wrong ‘prima facie’, as Holmes (Reference Holmes1992, 37) writes, or presumed wrong until proven otherwise. But second, whereas it is at least jarring if not downright nonsense to speak of ‘morally justified murder’, it is not logically self-contradictory to speak of ‘morally justified violence’, which is widely presumed to be a meaningful moral possibility.Footnote ⁸ So, for purposes of moral evaluation, the word ‘violent’ expresses (and demands) an attitude of presumptive disapprobation: that is, a case correctly described as ‘violent’ is something we should disapprove of unless sufficient reason is offered to justify an exception. Usage thus implies that the practice therefore requires a special treatment in ethics, an assumption I will defend.

The second thing Skinner specifies is the ‘sense’ of the word, given by the criteria defining the occasions calling for its use. These, in turn, specify a category of things that are understood to merit the evaluative attitude that the word expresses, determining, third, its range of ‘reference’. The rhetorical term for using a word like ‘violent’ in this way in relation to a particular case whose significance is in contention is ‘redescription’ (Skinner Reference Skinner2002, 161–62). One way of specifying the sense of the word ‘violence’ is what Harris calls the ‘Strict’ definition. In common usage it is based on a cluster of features typically found in standard cases, some of them referring to the kind of agency violence involves and others to its external, descriptive features along the following lines:

The Strict Conception has proved difficult to defend, however, in the face of a series of attempts at revisionary redescription. Whereas ordinary redescription sees interlocutors simply disputing the status of a contested thing (‘is this violent?’), revisionary redescription involves arguing about contentious cases on or beyond the margins of the usual category [in ‘territory lying on an uncomfortable borderline between violence and non-violence’, as Coady (Reference Coady2008, 41) puts it] in order to dispute the definition itself (asking ‘is violence this?’) and thus proposing more or less radical definitional amendments (cf. Finlay Reference Finlay2009; Garver Reference Garver2009).

Revisionary theories

Probably the most influential challenge to the Strict Conception is Galtung’s (Reference Galtung1969; cf. Coady Reference Coady2008, 24–29).Footnote ¹⁰ At its heart is the idea as that as long as members of a society suffer any needless privation, they are not ‘at peace’. And if ‘peace’ may be said to exist where ‘violence’ has been excluded, then ‘violence’ must equal privation: ‘violence’, Galtung writes, ‘is present when human beings are being influenced so that their actual somatic and mental realizations are below their potential realizations’ (Reference Galtung1969, 168). It can therefore take many different forms: in addition to its Strict forms, violence encompasses the disfigurement people suffer when excluded from educational opportunities, deprived of resources, or exploited for their labour. It occurs when any individual, collective, institutional, or social forces may be said to have caused harm. The most dramatic result is the extension of the category to include ‘structural violence’, occurring where harms arise without the action of any ‘subject (person)’: ‘The violence is built into the structure and shows up as unequal power and consequently as unequal life chances’ (Galtung Reference Galtung1969, 170–71; cf. Winter Reference Winter2012; on structural violence in security studies, see Schnabel Reference Schnabel2008, 88–95, and in feminist IR, Jones Reference Jones2010, 133–34, and True Reference True2012).

While Galtung radically alters the intension of the word and the category, its negative moral connotations remain the same. But he resists the idea that revision should lead to a more permissive view on physical force as a political means, that is, one that justifies using it to resist the wider variety of types of violence (a possibility recognized by Lee Reference Lee1996, 68; cf. Hardt and Negri Reference Hardt and Negri2009, 7, 16). By contrast, Žižek’s revisionary approach challenges the restrictiveness of the ethics of violence directly. Žižek distinguishes between three kinds of violence: ‘physical violence (mass murder, terror)’ and ‘ideological violence (racism, incitement, sexual discrimination)’ are both species of ‘subjective’ violence while the third is ‘objective’ or ‘systemic’ violence (Žižek Reference Žižek2008, 8, 9; cf. Fanon Reference Fanon1980, 63–64). Žižek’s agenda is similar in one respect to Galtung’s: ‘the task’, he says, ‘is precisely to change the topic, to move from the desperate humanitarian SOS call to stop violence to the analysis of that other SOS, the complex interaction of the three modes of violence’ and ‘to resist the fascination of subjective violence’ which is ‘just the most visible of the three’ (Žižek Reference Žižek2008, 10; cf. Hardt and Negri Reference Hardt and Negri2009, 7; Winter Reference Winter2012). But in one other respect he pushes further. In an argument recalling the Red Army Faction’s use of the notion of structural violence to justify targeting bankers and other socially influential figures (Varon Reference Varon2004, 238), Žižek argues that complicity in objective violence may render one liable to subjectively violent harm (Žižek Reference Žižek2008, 14, 33, 184, n. 1).

In perhaps the sharpest revisionary argument, Harris claims that there is such a thing as ‘non-violent violence’ consisting of the ‘negative acts’ of people failing to respond to injustice and help those in desperate need (Harris Reference Harris1980; see also Salmi Reference Salmi2009 [1993]). To bring its boundaries into question, he argues from cases that seem intuitively to belong in the category of violence but that Violence _s excludes. Examples include slow poisoning where there are no descriptively ‘violent’ characteristics in either the agent’s action or its effects on a victim – I will discuss his examples in more detail in third part. Since they seem to have the same moral character as standard, descriptively violent cases like shooting, stabbing, and so on, these non-standard cases must, he argues, belong in the same category. But once this adjustment is accepted and we abandon the assumption that ‘acts of violence’ must also be loud, sudden, or dramatic, Harris pushes further, maintaining that there can be no reason to exclude further harms for which individuals might be held responsible even when they did not exercise direct, intentional, active agency in bringing them about. If someone fails to prevent harms comparable with those caused by Violence _s – such as the shortening of lives by deprivation – then this too should be seen as violence.

Removing the Strict Conception’s descriptive and agential criteria in the way Harris proposes gives rise to a category of violence narrower than Galtung’s since it excludes structural harms for which no one bears personal responsibility but a great deal wider than that which arises from ordinary usage. Like Galtung’s argument and various others offered, for instance, by Garver (Reference Garver2009), Salmi (Reference Salmi2009), and Lee (Reference Lee1996), Harris contributes to the ideological critique of the ethics of violence, in which the distinctiveness of violence in a narrow sense as a form of harmful agency – paradigmatically, the resort to arms – is challenged and, with it, the special status it has been given by the ethics of violence.Footnote ¹¹

Violent Agency (I): double intent

Violent Agency is defined first by an attempt (1) to inflict harm by a technique designed (2) eliminate or evade the target’s means of escaping it or defending against it. It therefore occurs where there is a double intent: on the one hand to inflict harm; and, on the other, to narrow the window of opportunity within which its victim can respond or the range of means available for doing so.

Consider the way many of the means of violence are designed. One person throws a punch at another. She intends harm of some sort: pain, perhaps bruising, broken bones, unconsciousness, or worse. Like many other technologies used for violence, as theorists of strategy often emphasize, thrusting a fist forward at speed operates as a ‘force multiplier’, increasing its impact as it comes into contact with the victim (e.g. Stone Reference Stone2013, 106–07; also Stone Reference Statman2007). But the method also, crucially, has a second advantage: it shortens the time within which the victim has a chance to evade the blow. (I will refer to this as the ‘response window’ below.) This is even more remarkable when instruments are used.

Just as thrusting a dagger employs a mechanism similar to punching, albeit with greater likelihood of harming severely, firearms mimic the action of stabbing but with even higher speed and force and over significantly greater distances. The ability to inflict harm from a range at which it is impossible for (perhaps unsuspecting) victims to defend themselves greatly enhances the ability of an attacker to narrow their response window. The suddenness of a bomb exploding – whether fired remotely by artillery, dropped from the sky, concealed beneath the seat of a car, strapped to someone’s body, or stowed in a baggage hold – works along similar lines. Speed, distance, secrecy, and surprise may be combined in different ways to narrow the response window and inflict harm without impediment. The more effectively a technology does so, the better suited it is to employment as a means of violence.

It is useful to distinguish, here, between the means of violence and its deployment. In each standard case, the response window is narrowed not only by a chosen technology (the fist, the knife, the gun) along with its characteristic method of employment (striking, stabbing, shooting) but also by the way it is brought into engagement.Footnote ¹² The figure of the gunslinger in the American Western illustrates the point: he defeats opponents not only by having a gun and knowing how to shoot accurately; he also knows how to do so suddenly and without warning. Or in a duel, it is his ability to draw quicker than his opponent that makes his a successful act of violence. To succeed, he must narrow the opponent’s response window a fraction more than the other narrows his. Like the sniper, the drone operator, or the suicide bomber, he brings the means of harming into engagement in such a way as to maximize its capacity to evade or eliminate his opponent’s means of escape or defence. I will use the term ‘technique of violence’ to refer to a particular means and its manner of deployment taken as a whole. Some techniques realize the second intention either through the choice of victim or the timing of an attempt (for instance, by attacking someone when they are already immobilized).

While I take the twin intentions to be necessary features of violence, the degree of motivation driving an agent to pursue them will vary. In a case where the Violent Agent is absolutely determined to inflict harm, they will seek to realize both intentions to whatever extent is possible up to the point necessary to ensure success. In other cases, however, countervailing motives might limit the energy with which either intention is pursued. One might, for instance, attack someone half-heartedly without freeing oneself from the accusation of having committed or attempted an act of violence (e.g. by throwing a punch that is too slow to be assured of making contact with its target). Likewise, it is possible to pursue the twin intentions while treating moral constraints such as those set out in International Humanitarian Law as limits on what it is ‘possible’ to do (e.g. by forswearing the use of prohibited weapons even where they might be necessary for the greatest possible assurance of success and instead selecting others that, although they might still succeed, have a lower chance of doing so). These various factors will condition the chances of realizing the agents’ aims without calling into question the fact that they were actually intended. And, of course, Violent Agency may attempt to harm but fail due to the inadequacy of available means.

The potential of the idea of Violent Agency to reconcile seemingly contrary intuitions is seen in its ability to unite the standard cases within the same category as the non-standard ones that Harris marshals in his challenge to the Strict Conception (also Stone Reference Stone2013, 106–07). Since they lack the sensationally ‘violent’ experiences usual in cases of Violence _s , Harris uses the latter to drive a wedge between intuitions (that these are ‘acts of violence’) and the theory (that violence=Violence _s ). One example describes children in Belfast during the troubles hanging a length of cheese wire between two lampposts to injure the soldiers on passing vehicles. As Harris says, this was ‘[n]o violent act’ since the wire could be prepared at leisure, but it was ‘clearly an act of violence’ (Reference Harris1980, 16). Other tricky cases Harris cites include the use of poisoning, especially if the toxins operate on people without causing physically violent reactions. And, in fact, one of the Strict Conception’s ablest defenders, Coady, concedes that at least slow poisoning must be excluded from the category on the terms of a strict definition (Reference Coady2008, 41; also Stone Reference Stone2013, 107, n. 13). Another example of Harris’s (Reference Harris1980, 17) involves slowly inserting a stiletto between someone’s ribs as an ‘act of violence’ that lacks descriptively violent characteristics.

Because they display Violent Agency in perhaps even more dramatic ways than the standard cases, the non-standard examples are untroubling for the Double-Intent Account. The Belfast children chose their method because the wire would be imperceptible to its victims right up to the moment when it was too late to do anything about it. Closing the response window in this way was essential to their plan, just as it is in cases of poisoning. Imagine I bear you a grudge for which I want revenge and I lace your drink with a deadly toxin, one that gently but irreversibly puts your vital organs into a lethal sleep. You know nothing of my hostility. I hand you your drink and you sip it as we chat and I wait till you have drunk enough to guarantee your death. I might then declare it to you, that you have been poisoned and it is too late to do anything about it. Or I might never tell you, letting you die in ignorance of your fate and its cause. Either way the secret poison shuts your response window even more tightly than some standard techniques of violence: you might at least see a fist raised or hear a gun cocked in time to duck or run for cover. And whereas a booby trap bomb announces itself and injures simultaneously, secret poison ensures that death is irreversibly in train before you know you have a problem: its response window is measured in negative units (cf. Scarry 1985, 79).

Weaponized nerve agents unite something of the bomb and the secret poison. Sarin can be emitted rapidly across an area analogous to the blast radius of a shell and if it moves more slowly than explosive combustion, it makes up for it by the fact that no one knows it’s there until it’s too late: it ‘has no smell or taste and is colourless, so the first people may know of its use is when victims start to fall’ (Sample Reference Sample2013). Harris does not specify the circumstances in which it is possible to stab someone slowly but presumably the victim would have to have been subdued. If so, then it is with the act of subjugation that an assailant shuts the response window after which speed is no longer needed. Or, if the assailant exploits a victim who is already helpless it is a case of opportunistic violence. I will say more about the intimate relationship between violence and subjugation in the last part of this section.

So in all these non-standard cases, a technique of violence is designed specifically to achieve the effect of suddenness by ensuring that the interval between the victim (or a third party) becoming aware of a threat and the harm occurring is too narrow to allow them to do anything about it. In this respect, they are the same as the standard cases. At the same time, the element which unites these two groups also distinguishes cases of Violence _i from other ways in which harm may be inflicted, for example, as the foreseeable side-effects of primary actions that are not themselves Violent _i or as a result of recklessness. This is potentially a more controversial claim which I will defend in fourth section (second part). It distinguishes them from structural harms without culpable agents. And finally, it identifies a differentia supporting the intuition that negative acts such as negligence or the failure to assist where needed are not the same as acts of violence. As wrongful as they may be in many cases and as culpable as those responsible for them might be, they are different in ways that (I will argue) are sometimes morally significant (cf. Linklater Reference Linklater2011, 51–61).

Later I will defend this dimension of my account against some apparent counter-examples. But first I need to address one way in which it might be seen as over-inclusive which I can do while specifying the second feature of Violent Agency.

Violent Agency (II): destructive harming

Violent Agency occurs paradigmatically in the range of actions commonly denominated ‘acts of violence’, including both the standard and non-standard types that the Strict Conception struggled to unite within a single category. But these are not the only forms of interaction that its first feature – double intent to harm – appears in and shapes. It is also a feature of some practices often thought to fall somewhere (perhaps only just) outside the category of violence such as theft, slander, and exploitation.

Imagine someone discovers vulnerabilities in your bank’s security system and hacks into your accounts to steal your savings. Clearly they intend (1) to harm you. Moreover, the means chosen are probably intended (2) to deprive you of any means of resistance. Perhaps (2) is even clearer if your house is burgled: robbers strike when you are away and unable to defend your property. Compare these, then, with slander: someone harbouring a secret grudge attempts to diminish you by spreading unfounded rumours behind your back. As with hacking and burglary, your enemy seeks to harm you, choosing a technique that maximizes your exposure by depriving you of the means of defence.

If these express the double intent characteristic of Violent Agency, then are we forced to expand the category of violence to include them too? Were it necessary to do so wholesale, then the Double-Intent Account of Violence might have unattractively revisionary consequences. But I do not think it is. One reason for this is that the type of harms Violence _s inflicts is less controversial than the manner in which they are inflicted. Harris’s aim, for example, was to argue that causing potentially life- and health-threatening harms is morally the same whether the causal mechanisms are ‘fire and sword’ or global inequality and neglect [just as it was for Engels and Galtung, as Winter (Reference Winter2012) argues]. Along similar lines, then, one way of upholding a narrow conception would be to define as ‘violence’ cases of Violent Agency that aim at the sorts of harms usually associated with the term, that is harms to the person inflicted through bodily injury, psychological trauma, or deliberate damage to their property, etc. The chief point of the argument, then, would be that double intent is what distinguishes the infliction of such harms by ‘acts of violence’ from non-violent causation, for example, through failure to assist those in need or recklessness. Harms of other kinds are simply beside the point.

For present purposes, such a response might just suffice. But I think there is something more deeply distinctive about violence in a narrow sense accounted for by the second feature of Violent Agency: destructive harming. Distinguishing it from appropriative harming specifies more precisely how ‘acts of violence’ intend a different kind of harm from acts like theft.

At first glance, the distinction between the two categories seems to track the following contrast:

1. a. Appropriative Harming: Agent deprives Victim of x, which Victim would otherwise have continued to possess; Agent seeks thereby to benefit herself to the tune of x.

2. b. Destructive Harming: Agent deprives Victim of x, which Victim would otherwise have continued to possess; whereas Agent does not gain x or benefit from x as a good, she does benefit from A’s loss of x. Let us call this derivative or indirect benefit ‘y’.

So whereas robberies follow (a) by transferring good x from Victim to Agent, killing someone with a knife has a structure similar to (b). If you steal a wallet, you gain a wallet; but if you take a life, you do not gain a life. This analysis may prove too facile, however, if we do not push it further. To see why, imagine a thief stealing not a wallet but a part of his victim’s body. If he drugged someone in order to steal one of their kidneys, for instance, his action’s structure would resemble (a) but it would surely also be a case of violence. It therefore presents problems for a division between violence and other modes of harming based on the distinction above, indicating a need for further refinement.

Let x be the loss suffered by Victim, whatever the form it takes, and z be the gain enjoyed by Assailant as a result of inflicting it. My suggestion is that the difference between appropriative and destructive harming hinges not upon whether x is destroyed or transferred to the assailant but more deeply on the degree to which z is commensurate (and therefore also commensurable) with x. The more it is, the more likely we are to categorize the form of harming as appropriative (things like theft); the more jarringly incommensurate (or even incommensurable) they are, the more likely it is that we will see it as destructive (things like homicide and assault). Where x is destroyed by the assailant in order to secure some indirect benefit (revenge, security, military advantage, or whatever it may be) it creates disparities of this sort: this is why disparity and destructiveness are united in so many cases.

So, returning to the organ robbery, the thief gains a saleable asset – worth, say, some thousands of pounds on a black market in donor organs – whereas the victim suffers serious damage to their body, its integrity, and their health (not to mention pain and discomfort). Even if a monetary figure could be imagined that would restore to the victim a degree of contentment similar to what they enjoyed prior to the crime, there is no reason why it should necessarily match the assailant’s profit. But there is also a qualitative difference between the commercial asset gained and the loss of well-being suffered. It is trickier if we imagine the thief to be the transplant beneficiary herself since in that case she aims eventually to gain something that her victim loses. But even then, the value of x is greater than the organ as such; it also includes the bodily integrity that the attacker violates by cutting into it and thus forcibly removing the organ. And in any case, arguably even insofar as the perpetrator seeks ultimately to regain a good that their victim will lose (health; longevity), it is not acquired in the act of violence itself, which only affords the means to achieve it. To that extent, z≠x even if z will eventually be used to achieve the equivalent of (part of) x.

So just like other bodily violations, organ robbery involves double intent and destructive harming. The category of violence might therefore seem to be defined conjunctively by this pair of characteristics. By way of contrast, let’s return to hacking. I remove funds from your account and spend them. In that case, z=x; they are essentially the same. If this seems likely to be a fairly uniform feature of many kinds of theft, then we can say that whereas they express a double intent, they generally omit the destructive harming that is characteristic of Violence _i . It seems possible, then, to draw a reasonably clear division within the category of harms resulting from double intent distinguishing those that are Acts of Violence _i from others such as theft even if, as I suggest, they are close relatives.

Before I turn to exploitation and slander, I need to add a caveat. While x and z might be materially identical in many (perhaps most) robberies, they will not always be entirely identical. Say Robber steals £1000 from Victim. Robber is already wealthy but Victim is poor with only £1000 to his name. He is unlikely to be able to acquire more funds in the immediately foreseeable future and there is no insurance to compensate him. In these circumstances, x and z are identical in one sense (they are quantitatively commensurate) but in another they are not (they are qualitatively incommensurable). Robber’s enhanced power to acquire luxury goods is not equivalent, we may say, to the means of escaping imminent starvation or significant privation for Victim. It might be, then, that robberies sometimes may be interpreted as cases of appropriative harming on one level but also cases of destructive harming on another. If so, then we might want to argue that whether or not a particular case is also properly an act of Violence _i depends partly on contextual factors. At the very least, it may justify using term violence figuratively to capture fully its moral significance.

Exploitation might also be seen as involving double intent and yet be distinguishable from Violence _i but it is likely to vary according to type and degree. Consider two types:

1. 1. Surplus Value: Whereas most people will accept some form of paid employment, at least as a necessary evil, exploitation occurs where more value is taken from workers than they are compensated for.

2. 2. Servitude: Exploitation involves taking advantage of someone’s relative social vulnerability to force them to undertake labour they would not otherwise have been willing to perform (or had independently good reasons to perform).

On this description, Surplus Value takes something from the victim (x) that directly constitutes a benefit to the agent. As a case of appropriative harming it is therefore relatively straightforward to distinguish it from paradigm cases of Violence _i . But what about Servitude? At its most intense, it amounts to slavery. Should slavery too be interpreted as a purely appropriative mode of harming or is it destructive and, hence, a candidate case of Violence _i ?

Slavery lies right on the border between the two categories. On the one hand, it resembles an appropriative harm in classic cases. The ancient Greek practice, for instance, is sometimes interpreted (e.g. Arendt Reference Arendt1958, Ch. 5) as an institution intended to liberate citizens from the burden of labour, supporting their enjoyment of a free and dignified life from which slaves are simultaneously excluded. Insofar as in this picture labour is, in a fairly literal sense, taken from its victims and appropriated by its beneficiaries, then we might say that x=z. But this oversimplifies things.

To begin with, as well as being deprived of the benefits of their labour, slaves are harmed in being forced to undertake toil that they would not otherwise have chosen or had reason to choose. Moreover, the victim is also deprived of freedom in a sense that is distinguishable from the substantive loss of labour as such (see Pettit Reference Pettit1997). To describe freedom itself as the substance of both x and z can really only be figurative, not literal: the citizen does not literally gain the slave’s freedom any more than someone killing an assailant in self-defence literally gains the life of his attacker. On both scores, then, x turns out not to be the same as z. On the Double-Intent Account, therefore, slavery should be interpreted as a form of opportunistic violence in which the social vulnerability of the victim is exploited as a means of ensuring non-resistance for purposes of inflicting a harm that is essentially destructive.

I will say more about the connections between Violence _i and domination/vulnerability in the last part of this section. But first, one type of case that remains problematic is that of slander since the agent inflicting harm in this way does not seem to carry off x as an asset. If it can involve both double intent and destructive harming, then it would at least sometimes belong within the category of violence on the Double-Intent Account. Imagine, for example, someone lying to the head of a primary school that its janitor was rumoured to have a history of downloading child pornography. If the slanderer knew that this would cause him to lose his job, it is hard to see how it differs in any significant way from other ways of intentionally (1) inflicting destructive harm (2) by means chosen to deprive the victim of a means of defence. If it seems like a counter-intuitive extension of the category of violence, however, the fact that this sort of malicious harming is widely known as ‘character assassination’ reflects the deeper intuition that it has all the characteristics of violence, even if it seems like a ‘soft’ variant.

On the other hand, while slander might be exploited as a means of violence, it perhaps frequently bears greater affinity with modes of appropriative harming like theft. Consider the nature of the good – x – that slander sometimes seems to destroy, namely, reputation and relative esteem in a social group to which one belongs. Keeping this in mind, we may say that where I bring someone down in reputation through backbiting, I do not simply deprive them of some good; nor, furthermore, do I destroy an asset without gaining part of it. Provided I am a member of the social group within which reputation and status is awarded, my victim’s loss is also a relative gain for me and others. If slander is directed towards altering status between rivals, then it has something in common with other cases of appropriative harming: I deprive you of x (elevation in relative status) in order to gain z (elevation in relative status); hence, x is at least commensurable with z.

Violence, vulnerability, and domination

Before responding in the next section to some further objections, I want to say something about the important connections this account identifies between violence and the cluster of concerns that theorists analyse through the terms security, vulnerability, and domination. This will contrast it with rival approaches towards the analysis of violence in IR as well as helping specify more fully the significance of necessity and the possible.

Feminist IR theorists sometimes speak of a ‘continuum of violence’ to characterize the ways in which women suffer exposure to wrongful harm in contemporary global politics. True, for example, writes of

Assimilating various kinds of oppression, harm, and injustice under the heading of ‘violence’ in this extended sense lends rhetorical force to an analysis emphasizing their interrelationships. The same move, however, arguably runs the risk of sacrificing an ability to distinguish between different things. I think it better to reserve the term violence for those types of action categorized as such by the Double-Intent Account and to distinguish them from the background conditions of inequality that render Violence _i ‘possible’ (cf. Tickner Reference Tickner1992, 57–58; Young Reference Young2011, 61). Doing so helps bring out more clearly the important functional relationships between, on the one hand, the occurrence of violence in a narrow sense and, on the other, the vulnerabilities arising from political economy (True Reference True2012), the material and normative composition of international society (Deudney Reference Deudney2007; Clark Reference Clark2013), and background conditions of oppression within particular societies (Young Reference Young2011). Insofar as all create vulnerabilities exposing people to arbitrary interference from those to whom the same structures grant relative power, they are forms of what republican political theorists call domination.

Domination is crucial to violence on the account I offer. One person may be said to dominate another when they are enabled by a position of social power to interfere with them arbitrarily and without obstruction (Pettit Reference Pettit1997; Lovett Reference Lovett2010). As Locke characterizes it, the most intense form of domination occurs where a ‘master’ possesses the ability to choose whether his ‘slave’ lives or dies:

This passage portrays precisely the kind of relationship that paradigm cases of Violence _i are intended to establish between attackers and victims.

All the techniques discussed in the first part of this section seek to place their victims in a position where it is (close to) impossible for them (or others) to resist. In a word, therefore, we may say that violence on this account typically combines a force multiplier with a dominance multiplier. But whereas republican theory is centrally concerned with more enduring social relations of domination within and between states (Pettit Reference Pettit1997; Deudney Reference Deudney2007; Lovett Reference Lovett2010), those created as part of an act of Violence _i are usually transient and ad hoc, lasting just long enough to inflict the harm. In various ways, both standard and non-standard cases of Violence _i exploit speed, distance, secrecy, and/or surprise to create more or less fleeting moments in which the vulnerability of its victims to the attacker’s will and, hence, to the harm they wish to inflict is intensified (on defencelessness, Deudney Reference Deudney2007, 27; see Cavarero Reference Cavarero2009, Ch. 8).

Violent_i techniques harness domination in three different but often complementary ways. In ‘opportunistic’ forms (I) perpetrators target people already rendered defenceless and exploit their vulnerability as part of a technique of Violence _i . As an extreme case, for instance, assaulting someone paralyzed by illness would clearly be a case of violence, however calmly and slowly the harm was inflicted (cf. Cavarero Reference Cavarero2009, 31). But more generally, vulnerabilities created by wide differentials of wealth and power or permitted by political or legal conditions that fail to protect individuals, classes, or states all contribute to greater or lesser degrees to opportunistic violence (Tickner Reference Tickner1992; True Reference True2012; Clark Reference Clark2013).

Domination is experienced by its victims (whether states or persons) as an asymmetric vulnerability to interference. As such, the more intense it is, the more it excludes in advance any means to resist interference or evade harm, thus reducing the effort necessary for a Violent Agent to achieve their second intent. Established domination is therefore likely to make Violence _i possible across a wider range of cases by those it empowers. Kraut, for instance, writes that Greek slaves ‘could not take legal action in the courts, and were therefore vulnerable to beatings and maltreatment from their masters’ (Reference Kraut2002, 280). To kill a slave, the slave-owner might still need to employ one of the standard and familiar techniques of Violence _i , but his assurance that the slave will be unable to self-defend is greatly strengthened their social relationship as defined by background norms. Contemporary violence against women too is likely to exploit opportunities opened up by material inequality and by forms of domination underwritten by the normative composition of gender relations (Tickner Reference Tickner1992, 58; True Reference True2012).

Violent Agents can thus exploit the established relations of domination that republican and feminist theorists analyse by incorporating them within techniques of violence. They can also operate (II) by advance preparation. If a serial murderer drugged their victims before killing them, then we should see it as the part of their technique for closing the response window. Someone might object that this suggests that the act of, say, slowing inserting a knife in someone’s side in such a case (discussed in Harris Reference Harris1980; Stone 2013) would not itself be an act of Violence _i , only part of one. But this merely reflects the familiar idea that similar acts differ in meaning depending on context: whether the cut of a knife is a case of Violence _i depends on whether it is part of a surgical intervention in a medical emergency or an attempt at grievous bodily harm. Issues of intention, consent, and domination are vital to the way we interpret the act narrowly defined.

Cavarero (Reference Cavarero2009, 31) singles out torture as exemplifying the way defenceless victims may be ‘produced’ but the Double-Intent Account suggests that this occurs in all of the most familiar techniques of Violence _i . These harness a dominating relationship in the third way: (III) punching, shooting, bombing, and so on, all aim at creating a momentary, radically intensified relationship of domination and inflicting the intended harm (more or less) simultaneously. This may occur with or without prior vulnerabilities. In war, for instance, as Clark argues, normative regulations generate ‘categories of the vulnerable’ through the ‘language of proper targets, and gradations in the liability to be killed’ (Reference Clark2013, 43). These privilege opposing combatants by rendering mutual Violence _i impunible while depriving them of peacetime protections.

Whereas the dominating relationship is sudden and transient across a wide range of techniques, methods like laying siege to a target population or blockading them, by contrast, seek to maximize domination over a longer duration. Where essential goods are denied, the intention might simultaneously be to produce a defenceless subject and inflict harm (by means of starvation, for instance) as well as rendering them maximally vulnerable in preparation for armed attack (cf. Weizman on Wallfare in Reference Weizman2012, 80–81, Ch. 3; also Mbembe Reference Mbembe2003, 28–30). Sieges and blockades may be described as acts of violence on this account insofar as they are intended to inflict destructive harm by means intended to deprive targets of any means of resistance to it, rather than by the presence or absence of ‘armed force’ as such. Likewise the use of sanctions: arguably these may or may not aim precisely at destructive harm to their targets. If so, the distinction between violent means and those that are not violent therefore draws a different line from the one marked by the UN Charter’s differentiation between means of enforcement excluding ‘armed force’ (art. 41 including sanctions) and those requiring it (art. 42 such as ‘blockade’).Footnote ¹³

Asymmetries of power/vulnerability therefore permit agents to project harm towards potential victims, increasing the chance that they will suffer violence. But vulnerability is not violence as such, on this account, which occurs only once an agent exploits, intensifies, or creates radical asymmetry through Violent Agency. This is expressed in acts designed to deepen the victim’s vulnerability drastically by increasing the degree to which the agent dominates them to whatever extent remains necessary to harm them destructively (and to whatever extent it is possible for the agent to do so).

Over-inclusion

First it might be asked whether Violent Agency (as I define it) does not occur in any attempt conceived in earnest (what Locke called a ‘settled design’) to harm another person. Is it not true that anyone truly resolved on harming would try to eliminate their victim’s means of resistance? And if so, would not Violence _i become equal to all serious attempts at harm?

In general, while Violence _i may be an efficient and, consequently, a common way to realize a settled design to harm, it is not the only one. Consider two alternative courses of action that someone – call him Godfather – might take against a man who had previously wronged him, say, by killing Godfather’s beloved relative. Godfather wants ‘justice’ and has the option, first, of using his private investigator to gather evidence that will help secure a criminal conviction and lead to the man’s execution for his crime. Or, second, he could exact revenge directly by planting a bomb in the man’s car. Let us imagine that in the circumstances both means of harming had the same probability of success (whether due to the strength of the evidence or the difficulty of organizing a reliable assassination). The man might be lucky and escape either, but one vital difference between them is that, whereas proceeding against the man through a court of law affords him an opportunity to defend himself, planting a bomb does not. This is why, although either could issue from a clear, settled intent to harm, the latter is a case of Violence _i while the former is not (even if, for contingent reasons, the odds are the same either way). By contrast, were the Godfather’s investigator instructed to manufacture false evidence and destroy material helpful to the man’s defence, then it would no longer be part of a non-violent attempt to secure justice. It would be an act of Violence _i writ large, one that instrumentalized the powers of the court to inflict harm while intentionally depriving the victim of his means of defence.

Speech is another medium through which one might intentionally harm another without excluding opportunities for defence though I will indicate below some ways in which it too might be sometimes be instrumentalized within a technique of Violence _i .

Collateral harms

Some theorists argue that describing an act as ‘violence’ should not necessarily require that its agent intend harm. For Bufacchi, for instance, the category should include actions threatening foreseeable but unintended harms. Since the account I propose defines violence by a double intention, it is necessary to defend it from this argument.

As illustration, Bufacchi borrows H. L. A. Hart’s example of Fenians:

This is violence, Bufacchi argues, ‘not because the [bomb plotters] intended to kill anyone (they clearly did not), but because the fatalities were foreseeable, therefore the deaths were avoidable’ (Reference Bufacchi2007, 75). Indeed few, presumably, would find it satisfactory to exclude such an obvious example of violence. This seems problematic for an account that takes a specific way of intending harm to be definitional of violence. I will first offer an interpretation of Fenians that addresses the difficulty and then seek further clarification from a slightly trickier hypothetical.

Like Bufacchi, I presume that violence encompasses attempts to damage property and other things external to the body insofar as doing so can constitute destructive harm affecting people: as well as destruction of personal property, it includes sabotage of public utilities, spaces, or monuments (see Mandela Reference Mandela2002, 151; in the context of war, see Stone 2013, 105–06). The first line of response to Fenians is therefore to point out that bombing the prison constitutes an act of violence itself regardless of its unintended consequences. What we might call the Fenians’ ‘primary act’ therefore satisfies the criteria of Violent Agency. If the bombing also killed people collaterally, then we can say on this basis that they died ‘as a result of an act of violence’. But while the victims suffered harmful consequences resulting from violence, it does not necessarily follow that the Fenians committed an act of violence against them. Eliding the two descriptions marks a shift from literal to metonymic description.

To throw into sharper relief the salience of the primary – that is, intended – result in altering the way we describe an act considered as a whole, consider another hypothetical:

As with Fenians, the owners suffer harm foreseeably and by the deliberate use of classically ‘violent’ means. But are they victims of an act of violence? The primary act seems intuitively not to merit that characterization, nor does it fit the definition of violence on either the Strict or Double-Intent Accounts, so we cannot say that they ‘suffered from an act of violence’ in the way the Fenians’ victims did. But neither, I think, would we say that the rescuers committed an act of violence against them. If so, then foresight of the harm does not appear sufficient to make this an act of violence, which tends to confirm that intention is a prerequisite for that description.

I want to anticipate a second line of response to Fenians that some might think appropriate according to which Violent Agency should also encompass at least some cases where reckless behaviour threatens harm in such a way as to close the window of opportunity for defence. Think, for instance, of the casualties in Tripoli caused by rebels firing their guns into the air in celebration of victory in 2011 [a practice sometimes called ‘celebullets’ (Zeiton Reference Zeiton2011)]. Should we describe these gestures as ‘acts of violence’?

While these actions were vanishingly close to acts of violence, I think they were nevertheless different in a significant way. The gunmen used a technology originally designed to threaten harm and to narrow their targets’ response window and, although unintended, their manner of using it risked fulfilling that original design. However, I do not think it a trivial matter that the rebels acted in a way that was not intended to cause harm. For sure, an elementary sense of physics (and of responsible behaviour) might have indicated that the risks to bystanders were unacceptably high and their actions were probably reprehensible. But they did not involve Violent Agency and are therefore more accurately described as acts of ‘callous recklessness’ rather than of ‘violence’.

In this respect, the Libyan rebel is comparable to a drunken driver. Both know the potential of their instrument for serious harm; both, we will assume (ought to), realize that the way they use it imposes an unacceptably high risk on others; both (ought to) recognize that they are using it in a way that serves no purpose of sufficient moral value to countervail against this risk; both, therefore, act irresponsibly and may rightly be seen as culpable if they do in fact harm others as a result (and even, in a lesser degree, if they do not). Now, I think at least some of those who would have wished to describe the rebels as literally being engaged in violence would nevertheless agree that if we said the same about the drunken driver we would be speaking figuratively (if not downright inaccurately). But if both actions involve essentially the same sort of recklessness, then why distinguish between them in this way?

This cannot be due to differing degrees of risk because either might be worse depending on circumstances. I therefore suspect that it has to do, once again, with metonymy. The rebel’s gun is an instrument originally designed to serve Violent Agency; and in the celebrations, its use for other purposes nevertheless poses risks similar to those it was designed for (even if perhaps to a different degree). It seems natural, therefore, to characterize the rebel’s actions as ‘violent’ figuratively and, because of their close resemblance to Violence _i , it is easy to mistake metonymy for literal redescription. The relationship between the drunken driver’s actions and a true case of Violence _i , however, is less direct, so redescribing it as such is more obviously figurative.

On the Double-Intent Account, therefore, when civilians suffer harms as a side-effect of permissible attacks in war, we should not say, strictly speaking, that the agents perpetrated acts of violence against the civilians but we can certainly say that the victims suffered as a result of acts of violence. [This is true, incidentally, of any definition of violence that makes the intention to cause harm definitional of the act (including variants of the Strict Conception).] In this respect, it tracks the logic of the Doctrine of Double Effect (DDE) quite closely. But I’ll indicate in the final section one way in which the Double-Intent Account might also add further weight to the burdens of responsibility that the DDE imposes on combatants in war.

Harming by consent

To illustrate the consent objection, imagine two friends agreeing to take turns hitting each other; each also agrees not to resist the other.Footnote ¹⁴ This is a clear case of violence but since neither has to evade the other’s defences, the objection goes, the Double-Intent Account must exclude it. But I do not think it does. Consider, by way of a response, what might motivate such an unusual agreement:

1. a. Paired Sadists: Each wants to hurt but without the desire to be hurt; or

2. b. Paired Masochists: Each wants to be hurt but without the desire to hurt.

In Paired Sadists, the relationship is a contracting one in which each agrees to receive an unwanted injury in exchange for the benefit of inflicting one. It is a peculiar case, therefore, of opportunistic Violence _i : each exploits something about the circumstances of the other to lower their guard, namely, their compulsive desire to inflict pain. Consenting to endure a punch is thus a means by which each realizes the second intention. The case only seems problematic for the Double-Intent Account if we assume the contract is external to the violence. Once we view it as a constituent part of the violence, it is clear that Sadists intend both (1) destructive harm and pursue it by means intended (2) to eliminate or evade the other’s defences.

Whereas Paired Sadists belongs squarely within the category of Violence _i , Paired Masochists is more ambiguous. On the one hand, if both want to be hurt – for instance, if each somehow experienced pain as a source of exhilaration – we might question whether either is really harmed at all. And if neither inflicts harm, then neither commits Violence _i . If this judgement seems perverse, then compare voluntary euthanasia. Where physicians end the lives of patients seeking to avoid a protracted, more painful death, one could surely deny that they thereby commit acts of violence. One reason for this, presumably, is a doubt that killing someone, in such cases, really inflicts harm rather than, as some argue, a ‘benefit’ (summarised in Young Reference Young2015). Paired Masochists might then be interpreted along similar lines, mutatis mutandis. If pain is experienced by each as a benefit, then, by inflicting it, the other acts more like a physician tending to the other’s needs than a violent assailant in the moral sense of the term.

But perhaps, on the other hand, the Masochists do not precisely enjoy pain but suffer from crippling feelings of guilt that can only be assuaged by inflicting genuine suffering (Feinberg Reference Feinberg1984, 115). In which case, we might be more inclined to a second interpretation. Both Masochists and suicides of various kinds might be seen as engaging in Violence _i if we distinguish more sharply between what they destroy (x) and what they hope to gain (z). Take a genuine case of voluntary euthanasia sought by someone with terminal illness first. We might want to say that this is not an act of violence because of (a) consent and (b) the putative benefit to the patient. But I suspect this to be a distraction arising from focussing on the physician as agent. In the best case, presumably, the physician is merely an accessory whereas the principal agent is the patient. The question then is not whether the physician commits an act of violence, but whether the patient does. If we follow the analysis from third section, it is possible to argue that the patient intends destructive harm insofar as he hopes to benefit from one thing, the relief from continued pain (z), by depriving himself of another, continued life (x). As for the second part of double intent, as Ludwig Wittgenstein once remarked, ‘anyone who has visualised what is in practice involved in the act of suicide knows that suicide is always a rushing of one’s own defences’ (quoted in Monk Reference Monk1991, 187). If true, it reflects the idea that even where the agent is also the patient, there will be a need to engage in conflict with oneself, suppressing intentionally one’s instinctive tendency to block a threat. So, with both destructive harming and double intent in place, it is possible to see suicide even in the best case as ‘Violence _i ’ (which presumes nothing about its permissibility).

Turning back to Paired Masochists, whether compelled by guilt or by a pleasure derived from pain, their arrangement can be interpreted in terms similar to that between the patient and the physician. Even if they find pain or injury desirable, they may still need to storm their own defensive reflexes in order to bring them about. Partnering helps overcome the problem: while each must still suppress an instinct to raise an arm in defence, both enjoy the assistance of another who is willing to throw the punch at sufficient speed to inflict the desired injury. So, as with euthanasia, the masochists may be seen as engaging in Violence _i insofar as each recipient of a punch is also its author (with the other acting as accessory). As such, each intends (1) a harm to himself by an elaborate means (2) of overcoming his own defensive reflexes; z, the gain each enjoys (relief from guilt; exhilaration) differs from x what each deprives himself of (freedom from injury and pain).

Defence and futility objections

To illustrate the defence and futility objections, consider two hypothetical cases. In the first, Guerrillas attack a military compound and try to kill as many soldiers as possible, realizing that they will themselves be killed in doing so. Or, similarly, imagine Soldiers being sent over the top of a trench in World War I, and running with guns blazing into enemy fire. They know they have little chance of harming the enemy and every chance of being killed. In both cases, since the agents are all killed, it might seem that the targets’ ‘defences’, on one possible understanding of that term, remain intact, rendering the second intent moot (call this the defence objection); in the second, the absence of any reasonable expectation of harming the targets destructively seems to render the first intent moot (the futility objection).Footnote ¹⁵

Take the defence objection first: the word ‘defence’ is sometimes used ambiguously between two aspects of what someone does in an attempt at ‘self-defence’. On the one hand, they defend themselves against the attack, by which I mean they block it, prevent it, or deflect its intended effects. On the other, we might say that they defend themselves against the attacker. Sometimes the best means of escaping an intended harm is to inflict defensive harm on the agent threatening it; hence, defence in the first sense might require defence in the second. On the other hand, where the victim of an attack harms their assailant while realizing that there is no way that doing so can defend against the attack, it can be argued (and is often assumed) that the harms are not, strictly speaking, defensive (Rodin Reference Rodin2002). Even if we thought them justifiable, it would have to be for reasons other than defence, strictly speaking, like retribution or to prevent future attacks on others (see Statman Reference Stone2008). In the Double-Intent Account of Violence, ‘defence’ is intended in the first sense. This means that in cases such as Soldiers and Guerrillas, Violence _i may be said to have occurred even if the agents had no chance of surviving. As long as they can reliably be said to have intended both destructive harm to their targets and the elimination or evasion of their defences against the attack, then the cases satisfy the requirements of Violence _i in spite of the expectation of harm to the attacker.

For a variation on the defence objection, consider another hypothetical: Terrorists launch an attack intended to provoke their enemies into an indiscriminate counter-attack, radicalizing the affected population and intensifying support for the terrorists. This might be seen as a case where an act of violence is intended to widen the enemy’s response window rather than narrowing it. But again, this misapplies the theory. In fact, Terrorists aim at two assaults against two targets and they employ a technique that will realize the double intention in each case. The first targets are those harmed and rendered defenceless by the initial terrorist attack; the second are those harmed and rendered defenceless by a counter-attack engineered by the Terrorists.

As regards the futility objection, success is not part of the definition of Violence _i ; nor, for that matter, is ‘reasonable prospect of success’. All that is required is intent, and this is satisfied fully where the agent believes they have any chance of realizing the double aim at all. In the first example, soldiers combine the method of storming an enemy with the use of firearms to evade enemy defences as far as possible. That this might not be as far as necessary does not render it any the less an act of Violence _i . What it means is that the soldiers might engage in an unsuccessful act of Violence _i .