The Capitalist Cage: Structural Domination and Collective Agency in the Market

This article develops and defends a triadic account of structural domination, according to which structural domination (e.g. patriarchy, white supremacy, capitalism) is a triadic relation between dominator(s), dominated, and regulator(s) — the constitutive domination dyad plus those roles and norms expressively upholding it. The article elaborates on the relationship between structure and agency from the perspective of both oppressor and oppressed and discusses the deduction of the concept of the capitalist state from the concept of capitalism. On the basis of these de ﬁ nitions, it shows that structural domination under capitalism presupposes collective power but no joint agency or shared intentions on the part of the dominators. This article argues for the cogency of the idea of structural domination and for its application to capitalist economic structure. The article develops and defends a triadic account of structural domination, according to which structural domination (e.g. patriarchy, white supremacy, capitalism) is a three-place relation between dominator(s), dominated, and regulator(s) — the constitutive domination dyad plus those roles and norms expressively upholding it. The article elaborates on the relationship between structure and agency from the perspective of oppressor and oppressed and discusses the deduction of the concept of the capitalist state from the concept of capitalism. On the basis of these de ﬁ nitions, it shows that structural domination under capitalism presupposes collective power but no joint agency or shared intentions on the part of the dominators. The article has two parts. The ﬁ rst, conceptual, part argues that structural domination is a triadic relation. The second, normative, part applies that idea to capitalist domination. More precisely, Section 1 sketches two accounts of the wrong- or bad-making features of domination. Section 2 introduces the triadic account of structural domination. Sections 3 and 4 use that account to offer a de ﬁ nition of structural domination. Section 5 argues for the idea that the capital relation, the relation between capital and labour, is one of domination. Section under what conditions that

replaced. The same is true of social structures. Consider prisons. Simultaneous wholesale replacement of all of a prison's constituents-its staff, cells, bricks, or mortarneed not affect its structure. The question is what constitutes that structure.
This section argues that a complete description of specifically structural power relations involves a triadic relationship between those who possess power, those who are subject to power, and certain third parties-role occupants or norms-whom I will call regulators.

Structural power relations in general
Consider: Promise relation -Promisor promises Promisee to give her a copy of Crime and Punishment.
Promises are institutional facts 4 that confer on Promisee a power over Promisor-the power to demand or waive compliance with Promisor's antecedently undertaken obligation. Call this the constitutive power dyad. The institution of promising defines, in addition to offices (such as promisor/promisee), moves (under what conditions the utterance 'I promise' counts as a promise), penalties for non-compliance, and so on. Now contrast: Promise structure -Promisor promises Promisee to give her a copy of Crime and Punishment. Regulator is a bystander disposed to reward Promisor's compliance with her obligation and to penalise non-compliance.
Regulator's addition to the example completes the structure (or practice) 5 of promising: regulators define offices and legitimate moves, in addition to instituting and enforcing pay-offs corresponding to these moves. Regulators thereby impart structure on the promise relation by stabilising existing offices and by providing assurance of compliance to their officeholders-a function that requires at least the semblance of independence from these offices. That is, even when the regulators are existing officeholders, their actions do not exemplify their offices as such: Promisor, for example, is not exercising her office as promisor-is not promising-when she takes steps to ensure that she can perform promises in the future. That is the job of regulator.
A complete description of any structural power relation therefore involves a threeplace relation between the officeholders of the constitutive power dyad 6 plus the regulators (in this case: Promisor, Promisee, Regulator). In other words: (1) Structural power relations are triadic relations.
In terms of our original cathedral analogy, regulators relate to the constitutive power dyad in the way cathedral roofs relate to cathedral walls: without the roof, the walls are unstable and, in a sense, structurally incomplete. Bases need superstructures. 7 I now argue for: (2) Structural domination is a structural power relation.

Consider:
Interpersonal domination -Man finds Woman in a pit. Man can get Woman out at little cost or difficulty, but he demands an extortionate price to do so. Man would have done the same to a man, and a man could just as easily have found himself in the pit.
Suppose that this power relation is objectionable, such that Man dominates Woman. Still, Man does not dominate as man; 8 Man's domination of Woman is merely interpersonal. Contrast: Structural domination -Regulator pushes women, and only women, into pits. Regulator pushes Woman into a pit. This enables Man to demand an extortionate price from Woman in return for extracting her from the pit.
In this example, Regulator confers structure to the domination relation, in exactly the same way Regulator confers structure to the promise relation. 9 More precisely, Regulator makes the domination structural and gives it its discriminating character (as a form of sexist domination). Something similar is true of white supremacy and, I will argue, of capitalism.
In the examples used so far, the regulators help constitute the constitutive power dyad while remaining somehow external to it. External in what sense? Consider the relationship between cathedral wall and roof. Cathedral walls have powers, including the power to support each other. What confers that power on each wall, let us suppose, is the cathedral's roof. But what confers a power is not identical 10 with that power or with its possessor; to think otherwise is to commit the vehicle fallacy. So roofs can co-constitute the power of walls (e.g. the power to support other walls), but lack that power themselves; roofs regulate the constitutive power relation between walls. I claim that all instances of structural domination have the same triadic structure: they involve pit-like dyads constituted by third-party entities-the regulators-who complete the triad. Figure 1 elaborates the triadic relation graphically. To summarise the conceptual part of the argument: (1) Structural power relations are triadic relations.
(2) Structural domination is a structural power relation.

∴ (3) Structural domination is a triadic relation. (From [1] and [2])
Suppose this argument is sound, such that structural domination involves a relationship between dominator P, dominated Q, and regulator R. Note that P and Q might themselves be regulators, although not in their roles as dominator/dominated: when the playground bully exhorts her friends to bully vulnerable others, she is not bullying; rather, she is making (future) bullying possible. In so doing, she acts as regulator, not as bully.
It now follows that the existence of regulators constrains dominators to act in certain ways, including ways that constitute the domination of others (recall that roofs regulate, and, in so doing, constrain what walls can do). It does not follow, and it is false, that dominators are unfree qua dominators. That prison guards or slaveowners are structurally constrained to do what constitutes the domination of prisoners or slaves does not make prison guards and slaveowners unfree. 11 To think otherwise is to confuse the domination's conditions of possibility or ground (i.e. the regulators) with its subject (i.e. the dominators). By implication, advocates of the idea that structures can themselves dominate confuse grounds with subjects.
Drawing on the set of claims defended in this section, I now offer a definition of structural domination.

Structural = Regulated
The triadic account of structural domination implies the following definition of structural domination: Regulated Domination -A given instance of domination is structural just when it involves a triadic relationship between powerful agent(s) P, disempowered agent(s) Q, and regulator(s) R, such that (i) P dominates Q and (ii) R regulates that domination.
According to the triadic account of structural domination, this means regulated domination, that is, domination co-constituted by agents, roles, or norms external to the constitutive power dyad. Call this the regulation condition. 12 The regulation condition does not, by itself, tell us what domination does to the dominated. Consider common stock instances of the patriarchy: the wife-abusing husband, the lecherous male boss, the pornographer who stigmatises women. What makes these instances of structural domination?
A possibility that suggests itself is that women under the patriarchy are confronted with a dilemma between performing an independently valuable activity by the leave of men and not performing it at all. The wife abuser, for example, dominates his wife, in the sense that she must either perform unfree actions (e.g. have sex in order to avoid being beaten) or perform free actions (e.g. sex for pleasure) by the leave of her husband. Similarly, when pornographic norms restrict access to empowering sexual possibilities for women, such that women have to choose between unfree actions (e.g. sex on the model of male fantasy) and merely contingently free actions (e.g. sex by the leave of those who have internalised that model), they are dominated. Call this the double-bind condition. 13 Regulated Domination is a modal definition of structural domination, in two senses. In the first place, it says that structural domination is not just about the actual choices of structurally dominated agents or groups, but also about their subjunctive choices. That is, according to this definition, Q is dominated even in the nearby possible world in which she abstains from any interaction with P: the unmarried woman in a patriarchal society and the celibate woman in the pornography-saturated society are both dominated. In the second place, the definition explains a modal feature of structurally dominated choice, namely the clustering of subaltern optimising behaviour around the disjunction of unfree action and contingently free action. The thicker the bars, the thicker the clustering, and the fewer the opportunities for the undominated option of constitutively free, mutually affirming activity.
To conclude this section: I have offered a triadic definition of domination and suggested that this definition fares well in some of the cases that exercise feminists. I now ask what counts as a regulator in the definition of structural domination.

Defining the regulation function
A regulator is any agent, role, or internalised norm that contributes appropriately to the creation, reproduction, or perpetuation of the constitutive power dyad. 14 The question is what counts as 'appropriately'. I consider three possible specifications of the regulation function: causal, expressive, and moralised. a) Causal regulation: R is a regulator just when R contributes causally to P's domination of Q. b) Expressive regulation: R is a regulator just when R contributes expressively to P's domination of Q. c) Moralised regulation: R is a regulator just when R is morally responsible or otherwise blameworthy for P's domination of Q.
Specification (a) is too weak. Suppose I trip over a wire, causing the leader of the feminist revolution to fall on a bus, causing her to die, causing the patriarchy to survive. I have causally contributed to the reproduction or perpetuation of the patriarchy. But my tripping is an expression of maladroitness, not of the patriarchy: I am no regulator. So (a) is false. 15 On specification (b), my behaviour must somehow express the constitutive power dyad, indeed expressively uphold it. In other words, to count as a regulator, I must, in principle, be able to interpret my action as a move imbued with salient institutional meaning, even if I am not morally responsible for its (possibly oppression perpetuating) consequences. On specification (c), if I succeed in dissociating myself morally from P's domination of Q, I am a non-regulator. The feminist husband under patriarchy and the republican prince under monarchy are not regulators, according to (c).
Of these three specifications, (a) is too weak, while (c) seems too strong. For the purposes of this essay, I shall exclude (a) and remain non-committal between (b) and (c).
This concludes my discussion of the logic of structural domination. I now turn to the normative part of the argument. In what follows, I apply Regulated Domination to a specific kind of market structure: capitalist economic structure.

How capitalists dominate
This section argues that capitalism satisfies Regulated Domination. It is therefore an instance of structural domination, the domination of labour by capital. I begin by discussing the nature of the capital relation. I then explain how the addition of the state helps impart structure to that relation, that is, explains the constitution of domination under capitalist economic structure. Consider: Capital relation -Capital owns money, Labour does not. Labour needs means of consumption, which she cannot obtain without money. So Labour places her labour at Capital's disposal, in return for money. Capital appropriates Labour's labour, sells its product for money, returns half the money to Labour, and pockets the other half as profit.
The capital relation, in general, is value-constituted control over labour, or, more succinctly, a form of monetised servitude. 16 It follows that capital is not fundamentally a property or class relation, although it presupposes both private property and class division. Recall that roofs can co-constitute a power relation-say, of wall over wall-without being identical with or reducible to that power relation. By the same token, property can co-constitute the capital relation without being identical with or reducible to that relation. 17 If I am right, the capital relation is, fundamentally, a form of reified control over labour: it is reification that makes the capital relation conceptually and historically distinctive. 18 Now note that, although the capital relation presupposes the existence of private property in commodities-that is, a market for means of consumption-and money, it does not presuppose the existence of private property in the means of production. This is both a historical fact-the capital relation predates the capitalist mode of production -and a conceptual necessity-pre-capitalist forms of merchant and usury capital are instances of the capital relation, but they do not presuppose private ownership in the means of production. 19 It follows that the capital relation does not presuppose wagelabour: Capital can come to have reified control over Labour's labour through a credit market, or through mere market exchange. 20 Suppose I am right that the abstract capital relation need not involve propertyless workers or a labour market. I will defend that assumption in the next subsection. The question now is how the capital relation is reproduced, or, which is the same thing, how it is imbedded in social institutions. Contrast: Capitalist structure -Capital owns the cookshop; she gives propertyless Labour a cooking job. In labour market equilibrium, So Labour places her labour at Capital's disposal, in return for money. Capital appropriates Labour's labour, sells its product for money, returns half the money to Labour in the form of a wage, and pockets the other half as profit.
Unlike Capital relation, Capitalist structure posits the existence of a labour market, which presupposes the existence wage labour. Note that Capitalist structure relates to Capital relation in the same way that Promise structure relates to Promise relation: the structure-capitalism-is a concrete expression of the abstract power relation at the centre of the structure, the relation between capital and labour. If it can be shown that this relation is necessarily one of domination, then the structure that contains it-that is, the triadic relation of dominator, dominated, and regulator-will, necessarily, betoken structural domination. I now argue for this conclusion, namely: (4) Capitalism is an instance of structural domination.

Unfreedom in the market
The thought behind (4) is as follows. For as long as the profit motive reigns supreme as the mainspring of human productive activity, constitutively free production will remain impossible, and the exploitation of human by human will remain inevitable. Contrast a system in which state, market, and workplace are constitutively geared towards making free production available. This would make such production non-contingent on the extraction of surplus labour and profit. It would thereby disconnect social reproduction from the realisation of profit, making constitutively free production possible. And since constitutively free production includes production performed for the sake of putting goods to their most socially productive uses, price and value might survive in the free society, but only in their allocative functions. The next two subsections elaborate.
Economists distinguish between two roles of markets: allocative and distributive. The allocative role consists in communicating relative scarcity to appropriately situated agents. Leon Walras famously described markets as spatiotemporally extended auctions, in which the auctioneer announces the relative scarcity value of widgets, given total demand for all marketed goods, including widgets. In its allocative role, the market seems morally unobjectionable; it merely signals relative scarcities. It is different with the market's distributive role-a role built into the capital relation. Let me explain.
The capital relation, introduced above, presupposes that Labour lacks adequate non-market access to her means of subsistence. This lack confers on Capital control over a portion of the product of Labour's labour. What must be shown is that this makes Labour's action dominated, whether along neo-republican or neo-Aristotelian lines. Insofar as Capital's role is distributive, that is, consists in providing Labour with non-moral incentives, it satisfies both the neo-republican and the neo-Aristotelian definitions of domination. For, under the capital relation, Capital's relation to Labour is one of subjection to power, 21 for the purpose of making a profit. 22 The performance of free production, in other words, is contingent on sufficient profitability for the owners of the means of production.
It now follows that Capital dominates Labour, even if Labour's production is contingently free (e.g. Labour performs meaningful work for a decent employer). The double bind of capital thus consists in either producing unfreely, such as cooking-formoney or producing freely, such as cooking-for-the-sake-of-its-value, but only by the leave of capital. This is how capitalism, patriarchy, and white supremacy are structurally homologous: all three restrict the choice sets of the subaltern to a dilemma between unfree action and merely contingently free action. The worker with the decent job under capitalism, the woman with the feminist husband under patriarchy, and the non-white person with the anti-racist white companions under white supremacy are all dominated in that sense.
I conclude that Capital dominates Labour in Capital relation. 23 And since Capitalist structure is but a regulated instance of Capital relation, it follows that Capitalist structure necessarily involves domination. A typical expression of that structure is the capitalist labour market. Wage labour-Labour paid to obey Capital-is dominated labour.
I now briefly explain how markets can be made to serve constitutively free actions, as opposed to domination.

Freedom in the market
Joseph Carens defends a system of market-based, profit-maximising firms, whose profits get redistributed after production, such that the post-tax distribution is egalitarian, in the appropriate metric. 24 Carensian socialism enlists a set of moral incentives in production, which, in conjunction with redistribution, engender equality. Unlike some of Carens' defenders, 25 I do not take this to be a representation of an idealised form of socialism. Rather, Carens presents a stylised picture of how we might harness the allocative function of markets without the distributive; that is, have markets without capitalism. In this picture, it is false that powerful market actors have dominating power over others; they only act as traffic regulators of sorts. Let me explain.
Contrast a boss under capitalism (Bill) with a boss under Carensian socialism (Rosa). Bill and Rosa issue instructions to Labour, who does the cooking: 'make the omelette with mushrooms, not tomatoes'. This instruction gives Labour reasons to prefer mushrooms to tomatoes, regardless of whether Bill or Rosa utters it. But Bill's utterance has a different structural role from Rosa's. When Bill utters the instruction, Labour correctly understands that if she uses mushrooms, she might get a pay rise; if she uses tomatoes, she might get fired. When Rosa utters the instruction, however, Labour correctly understands that, if she uses mushrooms, she is saving social resources; if she uses tomatoes, she is wasting them.
So under the Carensian system, prices and wages possess a signalling function only: all they do is communicate independently valuable information, that is, information about relative scarcities. 26 It is as if the Walrasian auctioneer jumped out of the eggs, yolks running down his forehead, shouting: 'Make it with mushrooms!' In this world, there is no real private property, no real power of bosses over workers-other than that authorised by power-independent allocative requirements-and no reified control over labour; markets expressively uphold constitutively free production. Under the capitalist system, by contrast, markets assume the form of an alien power, standing over and in opposition to the direct producer; the worker and the means of production therefore become irreconcilable through the familiar Walrasian mechanism. 27 So Rosa helps Labour perform constitutively free production; Bill does not. And this is why Bill's power, unlike Rosa's, dominates Labour.

Capital entails the state
The normative part of the argument so far has gone as follows: (3) Structural domination is a triadic relation. (4) Capitalism is an instance of structural domination. ∴ (5) Capitalist structural domination is a triadic relation. (From [3] and [4]) The move from the dyadic Capital relation to the triadic Capitalist structure requires the addition of a regulator. Who is the regulator under capitalism? By the definition of a regulator, we need an agent, role, or norm that appropriately contributes to Capital's domination of Labour. One plausible candidate is the capitalist state, the main enforcer of private property in the means of production. Consider cookshops, for example. Modernity lends expression to the private ownership of cookshops through the institution of property law. 28 Property law, in turn, requires a legal system, which, in turn, requires (something like) a state. Capitalist structure, in other words, presupposes a regulator-in-chief, the capitalist state. 29 The concept of capitalism therefore entails the concept of the capitalist state. More precisely, the application of Regulated Domination to capitalism yields the concept of the capitalist state: (6) The capitalist state completes the triad of capitalist structural domination.
The capitalist state regulates the capital relation, by instituting, justifying, and enforcing, among other things: (i) Capital's control over Labour's conditions of production, (ii) Labour's control over her own labour power, and (iii) the exercise of these control rights through the institution of contract. This incidence of control rights confers on Capital power over Labour's labour, in two ways: Capital comes to control access to consumption goods and to the means used to produce them. The structural triad (dominator, dominated, regulator) thus takes the form (capitalist, worker, state). 30 So: Kant and Nozick are therefore right to deduce an idea of the state from the very idea of private property, in conjunction with certain generalisations about the nature of right. 31 But both signally and consequentially fail to note that such a state is compatible with the monetised mastery of the propertied over the propertyless and, conversely, with the monetised servitude of the propertyless to the propertied. Marx and Engels do not so fail, when they note that 'Communism deprives no man of the power to appropriate the products of society; all that it does is to deprive him of the power to subjugate the labour of others by means of such appropriations.' 32 Time to take stock. I have argued for a triadic definition of structural domination and for its application to capitalist structure. The capital relation, I argued, is monetised control over the labour of others. Insofar as this control obstructs constitutively free actions, it is an instance of domination. And when it is appropriately regulated, this domination is structural. Capital's principal regulator is the capitalist state. Table 1 broaches possible placeholders of the triadic relation in the cases of capitalism, white supremacy, and patriarchy.

Collective and individual agency
This section argues that capitalist structural domination does not, in general, presuppose shared agency or joint action. 33 On a widely held view, due to Michael Bratman and Margaret Gilbert, 34 joint action requires shared intentions, that is, P's and Q's intentions that P and Q perform an act, each by doing her part in that performance. These shared intentions must, moreover, 'mesh', in the sense that the content of P's participatory intention must cohere with Q's.
I now show that the joint regulation of a dominating power relation-its regulation by a multitude of agents-is not sufficient for shared agency or joint action, in the Bratman/Gilbert sense. If I am right, then the idea that capitalists jointly possess a power over workers does not presuppose shared agency. Capital, in other words, is collectively power conferring but agentless. Consider: Catallaxy Prison -Catallaxy prison has 990 prisoners and 10 guards. Some guards are former prisoners: they achieve guardhood by squeezing through the (few available) cell loopholes. The size of these loopholes is subject to capacity constraints, including the rate of growth of the prison and guard rule over the prison itself. The proportion of guards in the total population ends up hovering at around 1% for the whole duration of the prison's long existence, through nobody's plan or intentional design.
In Catallaxy Prison, prisoners squeeze into guardhood from the prison population, subject to capacity constraints and guard rule over prisoners. By assumption, there is a feasible prisonless alternative, in which all 1000 humans are free, but where there are no longer guards and prisoners, such that the former are worse off and the latter better off. The guards therefore have an interest in maintaining prison structure. It does not follow that they have shared and meshing intentions, that they optimise by forming them, or that they optimise by accepting some collective-decision procedure as theirs. 35 By way of illustration, consider prison guards P 1 and P 2 . P 1 knows that it is in P 2 's interest not to let too many prisoners through the prison loophole; the same is true of P 2 . By the definition of structure, moreover, all guards are symmetrically disposed. But then neither P 1 nor P 2 need have any intention that they enforce discipline or control the loophole together; it is consistent with their collective exercise of power that they lack shared intentions of the Bratman/Gilbert variety. P 1 , for example, violates no requirement of rationality if she gets on with enforcing discipline, all the while expecting-falsely, as it turns out-that P 2 will fail to do her part. More generally, the reproduction of domination in Catallaxy Prison only requires that: (i) individual guards are motivated by self-interest, (ii) what serves the guards' collective self-interest serves their individual self-interest, 36 (iii) the guards are individually capable and willing to enforce their collective self-interest, and (iv) these facts are common knowledge. Claim (iii) is subject to a free-rider objection: how do individual guards ensure that other guards enforce discipline? The exact answer is irrelevant to our purposes; perhaps, acting as regulators of the prison relation, the guards monitor each other's performance, punish defection, and reward cooperation. It does not follow that they have shared intentions or, yet more strongly, meshing shared intentions. So the collective power of prison guards does not presuppose shared agency, on their part. It is similar with the collective power of capitalists. 37 There is, however, a way that capitalism might indirectly presuppose collective agency. Consider the following argument: If (8) is true, then the main regulator of the capital relation is a collective agent. In which case it follows, by (5), that capitalism is a triadic relation between two agentless processes (capital and labour) regulated by a collective agent. I have no quarrel with (8). But I now show that the soundness of the argument just rehearsed does not impugn the idea that capitalist structural domination is agentless. This can be glimpsed by the following variant of Catallaxy Prison: Prison Warden -Same set-up as in Catallaxy Prison. The only difference is that the guards and prisoners jointly elect a warden. The warden is elected from the whole population of prisoners and guards through universal suffrage. Her mandate is to promote the quality of life of both guards and prisoners, subject to capacity constraints. Thanks to a long-lasting overlapping consensus between guards and prisoners, the proportion of guards in the total population ends up hovering at around 1% for the whole duration of the prison's long existence, through nobody's plan or intentional design.
In Prison Warden, the conditions for collective agency are satisfied: 38 the warden's election is a joint action by the population of guards and prisoners. But the existence and reproduction of prison structure itself are not (the upshot of) joint agency; nor do they presuppose any such agency. In other words, it is entirely consistent with Prison Warden that guards and prisoners lack any intention to reproduce prison structure and that each lacks any intention to do her part in that reproduction. So, once again, the collective power of the guards over the prisoners does not entail shared agency on their part.
Liberal capitalism is like Prison Warden, in that liberal capitalism allows forms of (democratic) shared agency subject to capitalist structural domination. More importantly, Prison Warden captures a distinctive feature of that domination: much like the power of the prison guard, capitalist power can reproduce itself spontaneously, without shared and meshing intentions on the part of capitalists. Perhaps for that reason, it can-by dint of periodic elections, welfare provision, trade unionism, co-determination, and other palliatives-systematically conceal its nature as a form of structural domination. 39

Conclusion
This article has argued for the cogency of the idea of structural domination and for application of that idea to capitalist economic structure. Capitalism, I have argued, is a cage-much like patriarchy and white supremacy. The article defended these claims by offering a triadic account of structural domination, according to which structural domination is a three-place relation between dominator, dominated, and regulatorany role holders or norms that contribute appropriately to the constitutive domination dyad. The article concluded that capitalist domination presupposes the capitalist state but not collective agency.  231ff, for an influential defence of this claim. In section 4, I discuss the relevant sense of 'need' and reject Cohen's specification of it as too weak. 8 Types ('man', 'woman') are, I assume, extensionally individuated. 9 Thomas Wartenberg suggests that all power relations involve regulators-he calls them 'peripheral social others' (Wartenberg, 'The situated conception of social power', Social Theory and Practice 14: 317-343). It follows that all power is triadic. Wartenberg's premiss is untenable: in the two-person world where Adam extorts Eve, we should think that Adam dominates Eve. Some domination is merely dyadic. Wartenberg uses the teacher-student relationship as an illustration of the putatively general point that all power is structural: 'A student's well-being is affected by the grade in the first place only through the mediation of human being situated outside of the classroom, who use the grade as a sign that results in their administering "harm" to the student' (Wartenberg op. cit., p. 322). This argument confuses individual instances of the practice of student grading, which may involve 'harm', with the general practice of student grading and evaluation. 'Harm' is not part of the constitutive conditions of grading and neither is the 'harmful' disposition of third parties. Otherwise we couldn't meaningfully speak of 'harmless', 'justified', or 'good' evaluation practices and powers. 10 That is, conceptually distinct and extensionally nonequivalent.