There have been two great, rival conceptions of power in modern social and political thought. The first conceives of power as a relation between agents and outcomes. Thomas Hobbes’s (2012: 10.11)
1
famous 1651 dictum in Leviathan, that power consists in an agent’s capacity or ‘means, to obtain some future apparent Good’, is one of the historical fonts of this tradition. Peter Morriss’s (2002: 32) more recent statement that power consists in an agent’s capacity ‘to effect outcomes (states of affairs)’ is paradigmatic.
According to the second conception, by contrast, power consists in an asymmetrical relation between agents. It is a pleasant irony – but also, I shall later suggest, telling – that the second conception of power also seems to find expression, 11 years before Leviathan, in another well-known passage by the same author in his first political work, the Elements of Law of 1640:
because the power of one man resisteth and hindreth the effects of the power of another, Power simply is noe more, but the excesse of the Power of one above that of another. For equall powers opposed destroye one another. and [sic] such theire opposition is called Contention (Hobbes, 1640: 8.4).
2
The classic modern formulation is Robert Dahl’s (1957: 202–203) proclamation that power ‘is a relation among people’ such that ‘A has power over B to the extent that he can get B to do something that B would not otherwise do’. Michel Foucault’s (1994: 233) later, even more influential treatment – despite his rather different, less agential conception – falls into the second tradition as well. In contrast to the term ‘capacity’, Foucault asserted, power (pouvoir)
brings into play relations between individuals (or between groups): . . . if we speak of the power of laws, institutions, or ideologies, if we speak of structures or mechanisms of power, it is only insofar as we suppose that certain [individuals] exercise power over others.
3
If for the first tradition power consists in an agent’s power to effect certain outcomes, for the second it consists in an agent’s power over other agents.
4
Proponents of the former tradition frequently argue power-over is merely a species of power-to; proponents of the latter, by contrast, often hold that all relevant social power consists in power-over, so that social power-to reduces to power-over. My agenda here is sevenfold. I will show, first, that the agential power agents possess and exercise – including agential social power – essentially consists in a relation between agent and outcome (power-to) and, second, that not all social power reduces to power-over. Third, I provide a diagnostic explanation for the widespread view that all social power reduces to power-over, namely, the pervasive conflation – which we can spot early on in Max Weber’s deeply influential treatment – of power-over with a rather distinct notion: power-despite-resistance. Social power is frequently equated with power-over, in other words, because it is invariably assumed – again with Weber – that social power intrinsically presupposes the capacity to overcome resistance. Fourth, I demonstrate that this widely held assumption is false: although overcoming resistance can be one aspect of social power, not all social power presupposes the capacity to do so. Instead, what makes power social is simply the fact that it consists in a relation between agents within a social-structural context. The Weberian resistance thesis ignores the fact, highlighted in Hannah Arendt’s work, that one type of social relation consists in assistance, not resistance – and, concomitantly, that one important form of social power is the capacity to effect outcomes with others’ assistance. Fifth, I accordingly outline three different types of power-with: via joint intentional action, strategic coordination and non-strategic coordination. My fundamental goal is, sixth, to provide an analysis of agential social power in general and to articulate the grammar of the concept – and this, by integrating the Weberian and Arendtian moments within a single conceptual framework. My thesis is that agential social power is a bounded four-variable term: within a given structural context c, an agent v’s agential social power consists in her capacity to effect outcomes O, with the assistance of agents X, despite the resistance of agents Y (where either X or Y may be a null set). More succinctly: agential social power is power-to-effect-O-with-X-despite-Y. Finally, one of the pay-offs of this conceptual framework is to explain a common but seemingly puzzling sociopolitical phenomenon: the sense of empowerment that individuals often derive from joining mass movements in which they are wholly subordinate to the dominating power of elite leaders. The framework shows precisely why this sense of empowerment need not be illusory: sometimes, by subordinating themselves to a leader who unites those subject to his power, these subjects gain considerable power-with thanks to the collective power of the group of which they are members. This is what I call empowering subjection. Moreover, not only can power-over be an indispensable basis for power-with, power-with can also be an indispensable basis for – indeed, take the form of – power-over.
Before proceeding, two preliminary comments. First, my focus here is on agential as opposed to structural power. The notion of agential power as I understand it is intrinsically linked to two aspects of exercising agency. First, agency is manifested in intentional actions, that is, actions constituted by an intention-in-action (Searle, 2010: 33). Thus, agential power is the power agents have to effect outcomes by way of their intentional actions, including intentional acts of forbearing. Second, intentional action is responsive to one’s intentional states, that is, subjective mental states with representational content. The exercise of agential power presupposes an appropriate link between effected outcomes and the agent’s subjective intentional states. Consider a victim who would not have been robbed had he not pulled out his wallet to give alms. The victim’s action is one of the robbery’s antecedent causes, but he did not exercise power (over the thief, for example) to effect it: the outcome is unwelcome to him (Ball, 1975b: 205; Morriss, 2002: 29). To count as an exercise of agential power, the agent must have a favourable attitude towards the outcomes.
5
By structural power, in contrast, I mean the supposedly ‘passive’ power (Morriss, 2002) agents have, in virtue of which the outcomes they prefer – or that serve their interests – obtain, thanks to the social structures in which they find themselves, but not by way of their intentional actions. For example, a man whose interests are served thanks to his position in his society’s structure of gender norms and status functions, but not by way of his intentional actions, would have power in this passive, structural sense.
6
Whether structural power in this sense is a legitimate power-concept is a matter of sharp disagreement: some theorists, such as Morriss and Pierre Bourdieu (2001), defend it,
7
but others reject it, reducing all social power to agential power.
8
I side with the former but, in this article, suspend judgement: in what follows, ‘power’ refers to agential power.
9
Second, my primary focus here is on having and exercising power, in contrast to the distinct notion of being subject to (or under) another agent’s power. It is one thing for one agent to have power over another (the ‘patient’); it is another thing for the patient to be subject to the agent’s power. As I construe the notion of subjection, that a patient is subject to an agent’s power entails that the latter has power over the former, but that an agent has power over the patient does not conversely entail the patient is subject to the agent’s power. This is because for an agent to subject a patient to her agential social power, she must either actually exercise the power she has over him, or be known (to the patient) to be disposed to exercise it (in a way, for example, that prompts him pre-emptively to respond to her anticipated reactions; Friedrich, 1937: 16–18, 1963: 199–215), or, finally, at least occupy a position within the social-structural context such that people with such power and in that position would normally be (intersubjectively) expected to have a sufficient reason to exercise that power.
10
For example, most agents have the power to kill strangers on the street, but in most well-ordered societies with effective social norms and sanctions against murder, agents would not normally be expected to do so; in this sense, the strangers/patients are not subject to other agents’ power to act murderously unless it is either credibly threatened or undertaken or the disposition to do so manifest.
11
My primary focus here is on power from the agent perspective, that is, whether she has or exercises power, and not from the patient or structural perspective, that is, whether he is subject to it. The following analysis is therefore agent-centred in two distinct senses: it focusses on agential, rather than structural, power, and it focusses on the agential power agents have or exercise, rather than the agential power to which patients are subject. (The treatment of empowering subjection is the exception, where I show that being subject to another’s power-over may be the basis for one’s agential power-with.)
Power-over and Power-despite-Resistance
Social power intrinsically involves a relation between agents. This much follows from the meaning of ‘social’. The characteristic thesis of proponents of the second tradition is that all specifically social power is a kind of power-over: the power of an agent (or group of agents) over others.
12
The concept of power-over, as I construe it, refers to a capacity to effect specific states, dispositions or actions in other agents (patients) in conformity with the power-holder’s will or interests. Thus, the second tradition’s characteristic thesis is that social power inherently involves the capacity to effect such states, dispositions or actions in others. I shall momentarily demonstrate that this thesis is false, but first I want to provide a diagnosis for why it is so widely accepted.
Weber’s well-known notion of Herrschaft – typically translated as domination or rule – is a kind of power-over. Weber distinguished ‘two diametrically contrasting types of domination’. The first is ‘domination by virtue of a constellation of interests [Herrschaft kraft interessenkonstellation]’. The second is ‘domination by virtue of authority [Herrschaft kraft Autorität], that is, the binding force of command [Befehlsgewalt] and the duty to obey [Gehorsamspflicht]’. The first type imposes an agent’s will on others’ behaviour – that is, gets others to do what the agent wants – by way of determining their external incentive structure in light of their ‘interests’ taken as given. Weber took the paradigmatic case of this type of domination to be a monopolist in ‘possession of goods or marketable skills’ who acts ‘upon the conduct of those dominated, who remain, however, formally free and are motivated simply by the pursuit of their own interests’. The second type – ‘domination by authority’ (autoritäre Herrschaft) or the ‘authoritarian force of command’ (autoritäre Befehlsgewalt) – consists in ‘the probability [or possibility: Chance] that a command [Befehl] with a given specific content will be obeyed [Gehorsam] by a given group of persons’. This type of domination – for which Weber reserved the term Herrschaft strictly speaking – imposes an agent’s will on others’ behaviour by way of influencing them internally to accept the agent’s command ‘as a “valid” norm’ for their actions, as if they ‘had made the content of the command the maxim of their conduct for its very own sake’, either through the influence of ‘empathy’, ‘inspiration’, ‘persuasion by rational argument’ or ‘some combination’ thereof. Emblematic of this type of domination is of course ‘patriarchal, magisterial, or princely’ authority (Weber, 1922: 28, 604–607, 1978: 53, 943–946, translations modified).
Weber did not, however, take Herrschaft to cover the entire range of social power. He took domination in either of these two senses to be a species of the more general notion of social power or Macht, which he famously defined as ‘the probability [or possibility: Chance] that one [actor] within a social relationship will [be in a position to] carry out [or impose: durchzusetzen] his own will even despite resistance [auch gegen Widerstreben]’ (Weber, 1922: 28, cf. 603, 1978: 53, cf. 941). This characterization of social power is at the heart of why many theorists reduce agential social power to power-over: because they assume, with Weber, that the social power to effect outcomes intrinsically presupposes the capacity to overcome resistance by other agents. Echoing Weber, for example, Brian Barry (1988: 341) has asserted that social power ‘entails that you have the ability to overcome resistance or opposition’ by others (see also Laswell and Kaplan, 1950).
There are two problems here. The first is that, even if it were true that social power presupposes a capacity to overcome resistance, it would follow that social power is intrinsically power-over only if the power to overcome resistance were inherently a mode of power over resisters. Many have indeed equated the concepts of power-over and power-despite-resistance in this way. Amy Allen (2016: section 1), for example, cites Weber’s definition of Macht above, which characterizes it as a kind of power-despite-resistance, as the ‘classic formulation’ of the concept of power-over. Indeed, Weber himself, despite initially defining Macht as the capacity to carry out one’s ‘own will even despite resistance’, later in the same work went on to gloss it as ‘the possibility of imposing one’s own will upon the behaviour of other persons’
13
(a thought clearly reflected in Dahl’s formulation). Weber seems to have assumed, in other words, that the power to overcome resistance – to which his initial definition refers – is always a mode of power over those whose resistance could be overcome. Therefore, to say that Herrschaft is a species of Macht is simply to say that Herrschaft is a specific mode of power over others. Indeed, Weber (1922: 604, 1978: 943) insisted that the term Herrschaft should not be used to cover all modes of power over others: doing so would render the term ‘scientifically useless’, presumably in part because it would collapse Herrschaft into the more general notion of Macht. It is precisely this equation of power-despite-resistance with power-over that furnishes the basis for seeing Hobbes’s (1640) formulation in Elements of Law, according to which ‘the power of one man resisteth and hindreth the effects of the power of another’, as paradigmatic of the second tradition.
Yet collapsing power-despite-resistance into power-over is an illicit conflation. The capacity to overcome resistance is distinct from, and does not entail, power over others: one may overcome others’ resistance without helping to determine their internal states, dispositions or actions. Take a simple example: under simple majority-rule voting, a majority of voters who vote for change can overcome the resistance of the minority who vote for the status quo, without thereby wielding any power over the states, dispositions or actions of the resistant minority, that is, without preventing the minority from preferring the status quo, being disposed to vote against change and in fact voting against it. It is true that the outcome of the vote may eventuate exercising power over the losing voters – if, for example, the vote concerns a decision that will bind the losing voters themselves. But this is a contingent eventuality, an instance of power distinct from the one in question. We can see this by turning to a vote to decide what people other than the voters are to do. Consider, for example, a vote by MPs in the colonial British Parliament for laws that bind subjects of a far-off colony. Here, there need be no sense in which the winning MPs effect the outcome they prefer by exercising power over the minority of MPs whose resistance was overcome. The fact that power-despite-resistance is distinct from power-over is precisely why it is possible for even some partisans of the first tradition, who argue that power-over is a species of power-to, to hold that all social power presupposes the ability to overcome resistance (Braham, 2008; Goldman, 1974: 231).
The second problem is that in any case it is not true – as is widely supposed – that all social power is power-despite-resistance, that is, that social power presupposes the capacity to overcome resistance. We can see this by noting that the power one has over others within a social-structural context is intrinsically social power. But not only does power-despite-resistance not entail power-over, power-over does not entail power-despite-resistance. Exercising power over others does not entail a capacity to overcome their resistance, because sometimes one can exercise power over others only with their acquiescence or even cooperation. Consider a teacher who would have no way of effecting the pedagogical outcomes she intends were her students to resist – or even fail to cooperate with – her efforts. She could still exercise power over them, but her power over them would be entirely welcome (Wartenberg, 1990). She exercises power over her students insofar as she effects states, dispositions and actions in them that they would not otherwise have or undertake. This is social power, because it is power-over in virtue of the social-structural context, but it is not power-despite-resistance.
This already provides a basis for replying to Alvin Goldman’s (1972: 261–262) objection to conceptions of power-over cashed out in terms of the capacity to effect states, dispositions or actions in other agents. Goldman argues that such a capacity cannot amount to power-over because one may effect an action in others simply by politely asking them to undertake it. Goldman presumes, of course, that getting others to do something via a polite request does not count as exercising power over them. Why presume this? For one of two reasons. First, one might think that, whereas polite requests are ineffective if the patient resists, power-over presupposes the capacity to overcome resistance. Second, one might think that motivating others to act through their own (autonomous) consent or assessment of reasons is not a mode of exercising power over them. The student–teacher example of welcome power-over gainsays the plausibility of both thoughts: we must reject the resistance thesis and recognize that power-over is not negated by rational motivation. An agent who is not recognized by others as having the standing to make polite requests or to advance reasons in communication is deprived of a significant kind of social power available to more privileged individuals in the social hierarchy. One mode of power is what Jürgen Habermas (1996) has called communicative power (cf. Forst, 2015).
14
The social power to effect outcomes is social in the sense that it involves a relation between agents within a social-structural context. But not all such relations are relations of resistance: they may also be relations of assistance. Consider again a case of voting, but this time under unanimity rule. Unless everyone votes in favour of a proposed change, the status quo is retained: each has a veto over changes. Now imagine everyone has unanimously voted for change. In this vote configuration, each voter is fully decisive for the outcome: had anyone voted differently, the outcome would be different. Take the first voter: ex post, she has efficaciously exercised the voting power she had ex ante to effect the outcome, and she could have effected the contrary outcome had she so wished. But ex post, she did not in this vote configuration overcome anyone’s resistance to change: everyone voted in favour. Indeed, ex ante she could not even in principle overcome anyone’s resistance to change: since each individual has a veto over changes, there is no voting scenario in which she could effect change by overcoming someone else’s resistance. In fact, under one-shot unanimous voting, no one has the capacity ex ante to overcome anyone’s resistance to effect change. Those who claim that all social power presupposes the power to overcome resistance must conclude that under unanimous voting, ex ante no one has any social power to effect change. And they must ipso facto conclude that when, ex post, change has been effected, no voter has efficaciously exercised any social power – they had none to exercise! – despite the fact that each is fully decisive for the outcome. This is an absurd implication. Barry (1980b: 343, 2002: 167), for one, was willing to bite this bullet: he asserted that in cases like this, agents exercise ‘outcome-power’ not social power. But this is just a misleading way of acknowledging that in social contexts such as this, agents exercise power even though it is not power-despite-resistance. They exercise power with each other’s assistance. Assisting is a social phenomenon.
Power-with
The management theorist Mary Parker Follett (1942: 78–79) was perhaps the first to mark the distinction in English, in 1925, between the expressions ‘power-over’ and ‘power-with’. Follett’s distinction was subsequently taken up by the philosopher Dorothy Emmet (1953–1954: 9) in her 1953 presidential address to the Aristotelian Society and, a decade and a half later, Hannah Arendt (1969: 35–44) made a similar conceptual distinction, albeit without reference to either Follett or Emmet, and without using Follett’s labels. Power-over she associated with ‘the old notion of absolute power that accompanied the rise of the sovereign European nation-state’, which, echoing ancient Greek conceptions ‘of government as the rule of man over man’, found its paradigmatic expression in Bodin and Hobbes. Power-with she associated with the ancient Athenian conception of the polity as an isonomy, and the Roman conception of government as a civitas, ‘where the rule of law, resting on the power of the people, would put an end to the rule of man over man’. Arendt unreservedly threw in her lot with the latter model, stipulating that power ‘corresponds to the human ability not just to act but to act in concert’ with others. The former, command-obedience model she linked not to power, but violence. Power in her telling is made possible by the mutual ‘support’ individuals give to each other in common endeavours, and this support is nothing ‘but the continuation of the consent’ lying at the origin of even hierarchically institutionalized modes of joint action.
Arendt’s reduction of all social power to power-with is, to be sure, as one-sided as its reduction to power-over: acquiescence is not the same as consent, and even if a hierarchy had some basis in consent, this would not preclude consent from sustaining, and hierarchy from being constituted by, relations of power of some over others – including the coercive power to handle dissent (Habermas, 1977; Wartenberg, 1990: 43–50). Arendt’s official conceptual apparatus does not help us to grapple with the way in which power-with can be the basis for – indeed, take the form of – power-over. If a set of individuals are able to act in concert, they may do so in ways that interfere with the choices of others, or that shape their character, states or dispositions. Indeed, a group’s collective power may even be exercised over its own members – precisely those individuals who have power-with the other members. What Arendt’s (1969: 42) work does make clear, however, is that the reduction of social power to power-over is also inadequate: it ignores a kind of power that, as she put it, ‘stands in need of numbers’, and which breaks down without the ‘support’ – or, as I will put more broadly, assistance – of others. Recall the case of the teacher: if she requires the cooperation of her students to effect her pedagogical goals, then her power-over her students is also intrinsically a power-with them.
I take power-with to be the power to effect outcomes in virtue of others’ assistance. ‘Assistance’ is appropriate here because its etymology, from the Latin assistere or ‘taking one’s stand’, evokes a broad range of intentional actions, including omissions or forbearance, that might enhance others’ capacity to effect outcomes – even without the intention to do so. Assistance can be unintentional and even unwitting. Arendt herself had in mind a much narrower conception of the relevant kind of assistance: by acting ‘in concert’ with others, she meant joint action on the basis of a shared intention and shared goals. There are, however, numerous ways in which one may effect outcomes with others’ assistance – even without shared intentions or goals. I shall classify these under three broad categories: joint intentional action, strategic coordination and non-strategic coordination.
15
My purpose in doing so is, first, to achieve greater classificatory and conceptual clarity and, second, thereby to illuminate further the relation between power-with and power-over.
Joint Intentional Action
Joint intentional action can take at least four relevantly distinct forms. First, a group of individuals with shared goals may assist each other by engaging in joint corporate action, as a collective agent (List and Pettit, 2011), in virtue of a shared intention to coordinate their individual actions on the basis of common decision-making institutions and mutual obligations.
16
Members of a political party furnish a typical example. Second, individuals with shared goals may assist each other by engaging in joint action on the basis of a shared intention, but without centralized decision-making institutions. Arendt seems to have had these rather thick notions of joint action in mind when she wrote of acting in concert with others.
However, for the purpose of analysing the power to effect outcomes with others’ assistance, we should cast our net more broadly. A third form of power-with is grounded in the following possibility: individuals may engage in joint action – and thereby assist each other – on the basis of a shared intention even when they do not have shared goals or values. Michael Bratman cites two people going for a walk together as a simple example of joint intentional activity. The joint action need not be based on shared values or common goals: perhaps I value socializing, and my goal is to cultivate friendship and to enjoy the company, whereas you value healthy living (and fear walking alone), and your goal is to promote healthy living (without fear) by exercising with others in fresh air. What makes our walking together a joint activity with a shared intention, on Bratman’s account, is that it satisfies several conditions concerned with participants’ intentions, beliefs, relations to each other, common knowledge and mutual responsiveness.
17
First, each of us intends that we walk together, by way of each person’s intention that we walk together (i.e. our intentions are both interlocking and reflexive, since the intentional content refers to both the other’s and one’s own intention), and by way of each responding to the other’s sub-plans and actions in implementing one’s plan to walk together in a mutually compatible way (our sub-plans and actions can succeed together). Second, each believes that if these intentions persist, then we will walk together by way of these intentions, and each believes that each person’s intention will persist if the other’s does. Third, our intentions are in fact interdependent in this way, in the sense that each person’s intention will in fact persist if the other’s does, because each can tell whether the other persists, and each responds to rational norms of planning (such as seeking coherence and consistency in intentions and beliefs). Fourth, all of this is common knowledge to each. These conditions together imply, according to Bratman, a shared intention, which, if combined with a fifth set of conditions, implies joint intentional action: namely, that our shared intention to walk together leads us to do so, by way of that intention, and by way of mutually responding to each other’s sub-plans and intentions in a mutually discernable (or ‘public’) way that tracks the jointly intended end of walking together. My power to effect the outcome in which I walk together with others is of course a power I have only thanks to the assistance of those who would walk with me. But the relevant outcome, which I have the power to effect with others, need not be restricted to the jointly intended activity itself: my power to prevent our neighbourhood from flooding may also depend on others’ disposition to cooperate and so assist me in the endeavour (where my goal or end is the non-flooding of the neighbourhood, not the preventive activity itself).
A fourth type of joint intentional action may even occur without genuine cooperation. Normally, the presence of deception or coercion blocks the jointness of intentional activity: if I get you to walk with me or to help prevent flooding by threatening you with a gun, I am not intending that we walk or prevent flooding together by way of your intention that we do so (I intend that you walk or prevent flooding with me by way of your intention to avoid getting shot). This is neither cooperation nor joint activity. But sometimes deception or coercion, although it blocks genuine cooperation, might be compatible with joint activity. For example, if you are a war captive under my command, the background coercion to which you are subject may nevertheless frame joint activities between us insofar as you can adopt an intention to undertake an activity with me by way of that intention, and I intend that you do so. If the flood is threatening our encampment, we may each intend that we prevent the flood together by way of our shared intention to do so. This is not genuine cooperation – you are still operating under my command thanks in part to a background threat of force – but it is a joint intentional activity.
Recognizing this fourth form of (non-cooperative) joint intentional action is significant because it shows that not only can power-with ground or take the form of power-over but that conversely a relation of power-with between agents can obtain in virtue of a relation of power-over between them. If I successfully get you to assist me to effect an outcome by threatening you, I wield power with you (to effect the outcome) by way of my (coercive) power over you. The fact that you are effectively coerced by me to effect the outcome does not negate you also having power-with-me to effect it: if you did not have any power to help me effect the outcome, there would be no point in me coercing you. This need not imply that you actually exercise power with me in effecting the outcome: recall that the exercise of agential power presupposes that the effected outcomes are welcome to the agent, that she has a favourable attitude towards them. But if you do have a favourable attitude towards those outcomes – you intend to help prevent the flooding, or prefer that our encampment not be flooded – then you too exercise power-with-me.
A similar point holds for cooperative forms of joint intentional action: if I successfully get you to assist me to prevent flooding by persuading you, without any deception or coercion, to join me in a shared activity – that is, by getting you to adopt the intention that we prevent flooding by way of our intentions that we do so together – I may very well do so in virtue of my intention that we prevent flooding together by way of our intentions to do so. We would thereby undertake a joint intentional activity, but here again I would wield power with you (to effect the outcome) by way of my (persuasive) power over you.
18
These examples of coercive and persuasive power-over are paradigmatic of how power-over can be the basis for the power-with of those subject to it, that is, of empowering subjection.
Strategic Coordination
The second broad category of assistance arises with merely strategic coordination. Strategic coordination presupposes purposive agents who respond to the potential goals and action-strategies of other purposive agents who similarly respond and for whom these facts are common knowledge. What distinguishes strategic coordination from joint intentional activity is the lack of a shared intention that they undertake the relevant activity. Consider Bratman’s distinction between two people who go for a walk together versus two strangers who just happen to be walking in the same, northward direction alongside each other. Each stranger intends that she walk north, and each is aware of the other’s presence and apparent intention to walk north, and each responds to the other’s actions and what she believes the other’s intention is – seeking to avoid collision, for example. All of this might also be common knowledge between the strangers. But this is not a walking together, because the strangers do not intend that they each walk together by way of that intention. Indeed, even if the strangers each intended that they each walk northward alongside each other, but not by way of that intention, their walking would not count as a joint intentional activity. Consider Bratman’s example of two rival gang leaders, each with guns pointed at each other, each of whom intends that they both walk together northward to somewhere more private (the first for the purpose of killing the second, the second for the purpose of killing the first: each thinks he’ll get the best of the other once alone). Here, the first intends that they both walk north, but via the second leader’s intention to avoid getting shot (and not via his intention that they walk together), and vice versa. This is still strategic interaction without a joint intention.
Strategic coordination can take at least three relevantly distinct forms. First, there may be strategic coordination around (some) shared goals and with complete information. (The condition of complete information obtains when agents have common knowledge about agent preferences and types, available strategies and the associated pay-offs, that is, about the structure of the game and players’ pay-off functions.) Searle’s (2010: 47–48) example of bleeding-heart libertarian classmates in business school provides a good illustration: each shares the goal of maximizing social welfare; each is aware they share this goal; yet they each believe the best way to help maximize social welfare is to pursue their own economic profit in competition with others (which each believes will, thanks to an invisible hand, maximize social welfare); and they each consequently go off, with the overall intention of advancing the shared goal of maximizing social welfare, to seek their own individual profit in response to what others do. They are engaged in strategic coordination around a shared goal, but not in joint intentional activity. And if they were right about the invisible hand, then they would have been assisting each other to effect their shared goal.
For another illustration, consider a referendum with 99 voters who cannot influence each other’s voting intentions, nor even communicate with each other to form any voting coalitions or to coordinate their votes explicitly. Imagine there are three alternatives a, b and c; the decision rule is plurality-rule voting; there are 49 Type 1 voters who rank the alternatives a > b > c, 35 Type 2 voters who rank them b > a > c and 15 Type 3 voters who rank them c > b > a, and all of this is common knowledge. As everyone knows, if everyone votes for their top-ranked alternative, then a will be selected, which is the second-ranked alternative of Type 2 voters and the lowest ranked alternative of Type 3 voters. Type 2 voters have no strategic incentive to vote for their second choice (which is a anyways) nor, obviously, for their third choice. But Type 3 do have an incentive to vote for their second choice b, in order to avoid their lowest ranked alternative a. Given that this is all common knowledge, Type 3 voters, who share a common set of goals in the referendum, have a strategic incentive tacitly to coordinate their votes on their second-ranked alternative to prevent their lowest ranked. When Type 3 voters act strategically in this way, they each exercise their voting power, with the assistance of other Type 3 voters (along with Type 2 voters), to prevent outcome a (and despite the resistance of Type 1 voters). They do not act jointly, with a shared intention, because they each merely intend to vote for b on the expectation that other Type 3 (and Type 2) voters will vote for b as well; they do not intend that other Type 3 voters vote for b. (Given planning agents’ responsiveness to norms of consistency and coherence, intending that others undertake some action with one involves being disposed to help them to undertake that action (Bratman, 2014), but here we have ruled this possibility out.)
Second, there may be strategic coordination around shared goals but with incomplete information. A variant of the last example serves as illustration. Imagine the voters do not have complete information about the distribution of voter types; they do not even know with whom or with how many they share their preference ordering; but everyone has enough information to know that alternative c will not be selected. The same analysis now ensues.
Third, there may even be strategic coordination without shared goals – if, for example, the strategic means that some actors pursue to realize their goals also happen to serve as means for realizing some other agents’ different goals. Imagine that our team’s sole goal in playing the soccer match is to defeat you, and that we will defeat you if and only if we give it our all; whereas your team’s sole goal is to play a beautiful game, but it is only in the face of us giving it our all that you will be spurred to the beautiful heights of soccer magic. We do not share goals, yet by each of us pursuing our own distinct goals in strategic interaction with the other, we help effect the realization of the other’s respective goals. You have the power to effect a beautiful game only with our (strategic) assistance: only with us taking our stand (assistere). Strategic coordination without shared goals is of course characteristic of market relations and helps explain why markets can enhance agents’ power to effect a broad range of outcomes.
As with joint intentional action, the exercise of power-with via strategic coordination may also operate in virtue of a relation of power-over between the coordinating actors. Market relations and exchanges facilitated by the state’s coercive enforcement of contracts and the rule of law provide an obvious example of empowering subjection, as did the United States in the second half of the twentieth century, who as a global hegemon underwrote a neoliberal interstate framework of strategic economic coordination.
Non-strategic Coordination
Finally, the third broad category of assistance occurs with non-strategic coordination. What makes the coordination non-strategic is that the actions undertaken by those who assist each other do not depend on their expectations about how the others might act. This occurs, for example, when all actors have dominant strategies, such as under majority-rule voting over two options, where voting sincerely is a dominant strategy. Alternatively, coordination might be non-strategic because it is unwitting: the assisted agents may not be aware of the role played by the others in determining outcomes, in which case they would not respond to the others’ potential goals and actions. There are two variants worth considering: one with shared goals among the relevant agents, the other without. First, consider two individuals inhabiting the same planet – members of the same global society, but unaware of each other’s presence – who unwittingly share the goal of preventing planetary climate disaster. (The goal is all-or-nothing: there is no point in merely reducing the pace of climate change without doing so to a level sufficient for preventing disaster.) Imagine that neither individual’s actions on their own could prevent disaster, but combined they would. Their actions would be coordinated in the sense of each contributing to effecting the same outcome, but there would be no strategic interaction: the coordination is neither explicit nor tacit, but unwitting. Second, consider two individuals who do not share any goals, are unaware of each other’s goals and activities, but whose activities are mutually symbiotic. Take, for example, two would-be farmers: the goal of the first is to produce honey, of the second to produce almonds. So the first is disposed to build a honeybee farm, the second an almond orchard. Now suppose the bees would not produce enough honey without feeding from the almond trees (there being not much else around for them to eat), and the almond trees depend on the second farmer’s bees for pollination. The two farmers do not share any goals, do not strategically interact, but the activity of each would assist the other.
19
This is an extremely thin sense in which these agents would assist each other, but it is nevertheless true that, in this world, each agent’s power to effect the outcome in question is a power she has only thanks to the disposition of the other (unwittingly) to assist the other (and vice versa).
Social Power as a Four-Variable Term
Agential social power need not include a capacity to effect outcomes despite others’ resistance: it may instead take the form of the capacity to effect outcomes with their assistance. This suggests social power is entirely consistent with the core concept of power as construed by the first tradition: the power to effect outcomes. If social power is the capacity to effect outcomes in the context of relations with other agents, this relation may consist in the capacity to overcome others’ resistance to effecting the outcomes, or it may consist in the capacity to effect outcomes with their assistance. Indeed, as the case of the students whose cooperation the teacher requires illustrates, even the power one wields over others may take the form of power-with rather than power-despite. The appropriate contrast to power-with is therefore not, as has been traditionally supposed, power-over: power-with can itself be a mode of power-over – as when a group with collective power has and exercises power-over its own members (or, indeed, over others) – and power-over may be the basis for the power-with relation – as when the state solves collective action problems and thereby enables individuals to act in concert. The key distinction for analysing agential social power is, rather, between power-with assistance versus power-despite resistance.
Whether an agent’s action (including forbearance) counts as assistance or resistance depends on the context. For example, in a voting structure with two possible outcomes, yes or no, in which a yes-outcome requires that an absolute majority of all eligible voters vote yes, abstaining from voting – forbearing – would unequivocally count as resisting the yes-outcome. This is because there are possible vote configurations in which the abstention would be decisive in securing a no-outcome. If the votes are just one shy of the quota for a yes-outcome, then the outcome would have been yes had a given abstainer voted yes (holding everyone else’s vote constant), that is, her abstention would be fully decisive for effecting the no-outcome. But there is no possible vote configuration in which an abstention would be decisive in effecting a yes-outcome.
The grammar of the concept of (agential) social power is embedded within the grammar of power more generally. On the analysis defended here, (agential) power is a bounded two-variable term consisting in a relation between agent and outcomes (within a specific environmental or structural context) as follows. Within a given environmental or structural context c, an agent v’s power with respect to a specific (contextually possible) outcome-type O consists in her capacity to effect that outcome type; her overall power within context c consists in her capacity to effect any (contextually possible) outcome type. To say that v has the agential power to effect outcomes of type O is to say that there is at least one possible occasion in context c such that, were v to undertake an intentional action available to her, it would be efficacious in helping to bring about an outcome of type O.
20
(The degree of her overall power in context c is, in turn, a function of the proportion of occasions in that context in which her intentional action would be efficacious in effecting the outcome on that occasion, weighted by the relative contextual significance of that occasion.
21
) This analysis covers both social and non-social power.
To analyse specifically social power, we can disaggregate and make explicit the two types of social relation within which the agent’s capacity is embedded, relations that remain implicit in the more general analysis of power. Social power is consequently a four-variable term consisting in a relation between agent, outcomes, assistants and resisters. Within a given social-structural context c, an agent v’s social power with respect to a specific (contextually possible) outcome-type O consists in her capacity to effect that outcome-type with the assistance of agents X, despite the resistance of agents Y (where either X or Y may be null sets). This yields three types of social power: the general case in which X and Y are both populated, and two special cases. Unilateral power arises when X is empty: this is social power-despite Y’s resistance (but without anyone’s assistance). Omnilateral power arises when Y is empty: this is social power-with X’s assistance (but with no capacity to overcome anyone’s resistance). (The case where both X and Y are empty includes non-social power.) Unilateral social power is illustrated by the case of dictator-rule voting, where there is one voter whose vote always determines the outcome regardless of how other voters vote. Whenever the dictator votes yes, she unilaterally effects a yes-outcome despite the no-voters (but without the assistance of the other yes-voters: their votes are impotent). And as we have already seen, omnilateral social power is illustrated by the power of voters to secure a yes-outcome and pass a proposed change under unanimity-rule voting.
Agential social power is the power-to-effect-O-with-X-despite-Y.
Conclusion
It is a central aim of many social and political theories – whether liberal-egalitarian, republican, democratic or socialist – to identify and counter many of the power inequalities found in our societies. By recognizing that social power in general neither presupposes the capacity to overcome resistance nor need take the form of power-over others, the framework furnished here lays the conceptual groundwork for identifying the multiple modes of power that such theories might otherwise fail to register.
Beyond imparting conceptual clarity, the recognition that the counterpoint to power-with is not power-over but power-despite makes at least two important contributions to social and political theory. First, the current framework enables us to recognize the ways in which relations of power-with can enable relations of power-over: the stability and persistence of hierarchical institutions constituted by relations of power-over among agents are often possible only thanks to agents acting in concert with each other to sustain them.
Second and conversely, hierarchical relations of command and obedience, and the coercive regulation of social interaction – paradigmatic examples of which are furnished by state institutions – can furnish the structural background against which joint intentional action, strategic coordination or even unwitting coordination may occur. This corresponds to what I have dubbed empowering subjection. We cannot fully understand the appeal of hierarchically organized sociopolitical movements to those occupying their bottom rungs without recognizing that relations of power-over often enable and constitute relations of power-with. Social scientists are sometimes puzzled by the appeal of certain nationalist, populist or religious movements to their most underprivileged, subordinated members. But that allure becomes much more comprehensible when we recognize that although being subordinated to the power-over of another could in one respect diminish one’s agential power – if, for example, the power-holder interferes with and thereby restricts one’s options – in another respect it can be empowering insofar as one becomes a member of a group with collective power, which can enhance one’s power-with others to effect desired outcomes, to overcome other groups’ resistance or even to dominate other groups. The same insight applies to patriarchal relations of gender subordination or domination, or to poor Whites’ acquiescence to their class subordination to rich Whites: insofar as poor Whites’ class subordination grounds the collective power of Whites over non-Whites, they are also empowered. Class-subordinated Whites’ feeling of racial empowerment is not wholly illusory.
Of course, the present framework, focussed as it is exclusively on agential power, only partially removes the obstacles to our understanding of power inequalities. To complete the task, we also need a theory of structural power. For another day.
