Power - Three Faces and What Lies Beneath

Conflict theory rests on a concept it does not itself define adequately: power. This lesson takes it directly, because power is both the discipline's most-used word and its most contested one, and because there is an unusually clear structure to the argument.

The structure is a sequence of moves, each of which says: the previous account measures only the power you can see.

Three ways to get someone to do what you want.

One. A landlord tells a tenant to leave. The tenant refuses. The landlord goes to court, obtains an order, and the bailiffs arrive. There was a conflict, it was visible, and the landlord won.

Two. A group of tenants wants to raise the question of repairs at a residents' meeting. The item does not appear on the agenda. The chair, when asked, explains that repairs are an operational matter for the management company and not a proper subject for the meeting. There is a procedure for raising operational matters; it involves a form.

No conflict occurs at the meeting. The question is never put, no vote is taken, and nobody loses an argument — because no argument happens.

Three. In a different building, nobody raises repairs at all. The residents have lived with damp for years. Asked about it, they say the building is old, that this is what you get at this rent, that it is not really anyone's fault, and that complaining would not achieve anything and might make things awkward.

There is no conflict, no suppressed item, and no visible grievance. People have adapted to their situation, and their adaptation feels to them like realism.

Now: in which building is power operating most effectively?

The answer that the whole of this lesson defends is: the third. And the reason is that in the first case power had to be exercised, in the second it had to be managed, and in the third it does not have to be used at all.

As Steven Lukes put it: "Is it not the supreme exercise of power to get another or others to have the desires you want them to have — that is, to secure their compliance by controlling their thoughts and desires?"

The dispute begins as an argument about method and becomes an argument about what power is.

The pluralist position (Robert Dahl, 1950s–60s): power should be studied where it can be observed. Identify key decisions on contested issues, find out who prevailed, and you have measured power. Anything else is speculation.

Dahl's Who Governs? (1961) applied this to New Haven and found different actors prevailing in different domains — business in redevelopment, educators in schooling, party officials in nominations — with no single group dominant. His conclusion: power is dispersed among competing minorities. He called it polyarchy (4.6.2).

The elitist position (Mills, Hunter): power is concentrated in an interlocking set of institutional positions, and the fact that it does not always show up in observable decisions is not evidence of its absence.

The methodological deadlock: the pluralists say the elitists cannot demonstrate their claim, and the elitists say the pluralists have defined power so that only its most visible form counts.

The resolution — and it is a genuine one — comes from asking what the pluralist method would miss even if perfectly executed.

The first face: decision-making

What is right about it: it is checkable. It requires evidence. It refuses to infer power from position or reputation alone, and it produced a genuine finding — that different groups prevail in different domains, which the elite theorists' model did not predict.

What is wrong with it: it can only detect power that produces a visible contest. Everything that prevents a contest is invisible to it, and the next two moves are about exactly that.

The second face: non-decision-making

Peter Bachrach and Morton Baratz, "Two Faces of Power" (1962), made the first move.

"Power is also exercised when A devotes his energies to creating or reinforcing social and political values and institutional practices that limit the scope of the political process to public consideration of only those issues which are comparatively innocuous to A."

The mechanism is agenda control — what they call the mobilisation of bias, a phrase they take from E. E. Schattschneider: "All forms of political organisation have a bias in favour of the exploitation of some kinds of conflict and the suppression of others, because organisation is the mobilisation of bias. Some issues are organised into politics while others are organised out."

A non-decision is "a decision that results in suppression or thwarting of a latent or manifest challenge to the values or interests of the decision-maker."

How it operates, concretely:

Rules of procedure determine what is a proper subject for a body to consider. Agenda-setting determines what reaches the table and in what order. Anticipated reactions — the most common form — mean that a proposal is never made because everyone knows it would fail. Definitions of the possible rule options out as unrealistic before they are stated. And the threat of sanctions, which need not be issued to be effective.

The third face: shaping preferences

Steven Lukes, Power: A Radical View (1974), made the decisive move.

His argument: both previous accounts assume that B has a preference which A overrides or suppresses. The first face requires visible conflict; the second requires a latent grievance kept off the agenda.

But what if B has no grievance at all?

"A may exercise power over B by getting him to do what he does not want to do, but he also exercises power over him by influencing, shaping or determining his very wants... Is it not the supreme exercise of power to get another or others to have the desires you want them to have?"

The third dimension concerns the shaping of perceptions, cognitions and preferences such that people accept their role in the existing order — because they can see no alternative, because it appears natural and unchangeable, or because they value it.

And its key feature: there is no observable conflict at all. Not suppressed conflict — no conflict, because the parties do not experience their interests as opposed.

A workplace where nobody has been promoted from the shop floor to management in twenty years.

First face. Someone applies, is rejected, complains, and loses. Observable conflict, observable outcome. This is measurable and, in this workplace, it has happened twice in twenty years.

Second face. The promotion criteria include a qualification that the company's own training programme does not provide and that shop-floor hours make impossible to obtain. Nobody decided to exclude anyone; the criteria were adopted for good reasons at a time when the workforce was different. The rule does the work, continuously, and no one has to. When someone proposes changing the criteria, the proposal is referred to a review, which reports in eighteen months.

Third face. Almost nobody applies. Asked why, people say management is not for them, that they would not fit, that they lack the education, that they would not want the stress, and that the people who go into management are a different sort. These are not complaints. They are self-descriptions, offered without resentment, frequently with some pride in not being that sort of person.

And here is the analytical difficulty that Lukes insists on rather than hiding. Are those self-descriptions:

(a) an accurate assessment of a real mismatch — some people genuinely do not want management, and this is a legitimate preference; (b) an adaptation to a situation, in which people have adjusted their wants to what is available — what Jon Elster calls adaptive preference formation, and what the fable of the fox and the sour grapes names; or (c) the internalisation of a definition supplied by others about who is management material?

The pluralist's objection is that you cannot tell, and that the analyst who says "they only think they don't want it" is claiming to know someone's interests better than they do.

Lukes's answer, and it is honest rather than triumphant: you cannot tell from the situation alone. What you can do is look for relevant counterfactuals — comparable workplaces where the criteria differ and applications from the shop floor are common; the same workforce at a moment when the arrangement was different; what happens when someone does make the transition. Comparison is the method, and it is imperfect.

The problem of real interests

The third face requires a concept of real interests — what B would want under conditions of relative autonomy, free from the influence of A's power.

And this is where the concept becomes normatively loaded, which Lukes states openly rather than concealing.

"The concept of power... is ineradicably evaluative and 'essentially contested'... any attempt to define it and locate it in social life is unavoidably value-laden."

The objections, which are serious:

Who decides? If the analyst determines B's real interests, the analyst is claiming privileged access to another person's good. This is the condescension that 4.1.7 identified in "false consciousness," and it is unfalsifiable in the same way: agreement confirms the theory, disagreement confirms the domination.

And the strongest version of the objection comes from a direction Lukes had to answer: if power operates by constituting people's very subjectivity — their sense of who they are and what is possible — then there is no autonomous B whose real interests could be identified. This is Foucault's position (5.9), and it dissolves the concept of real interests entirely.

Lukes's own response, in the 2005 second edition, is the most useful part of the debate and is usually not taught.

He conceded substantial ground. He accepted that his original formulation was too tied to a binary of domination and freedom; that power is not only "power over" but also "power to" — the capacity to act, which is not inherently zero-sum; and that he had underestimated the ways power can be productive rather than only repressive.

And he defended what remained: that people's compliance with arrangements harmful to them is a real phenomenon requiring explanation; that adaptive preference formation is well documented; and that the alternative — accepting all expressed preferences as final — makes it impossible to describe indoctrination, manipulation or the effects of severely restricted options, which plainly exist.

His revised formulation is more modest and more defensible: power's third dimension concerns the capacity to secure compliance by shaping perceptions and preferences, it operates in degrees rather than absolutely, and identifying it requires comparative evidence rather than assertion.

This is a model of how to hold a contested concept: state the difficulty, refuse both the position that it is unmeasurable and the position that it is obvious, and specify what evidence would count.

Go deeper

Beyond the three faces

Foucault (5.9) rejects the whole framework. Power for him is not something possessed by A and exercised over B; it is a relation dispersed throughout the social body, operating through discourse, classification, examination and self-regulation, and it is productive — it produces subjects, categories and knowledge rather than merely repressing. "Where there is power, there is resistance." The question is not who has power but how power operates, and the answer is: through the constitution of what counts as normal, true and possible.

Bourdieu (5.10) adds symbolic power: the power to impose categories of perception as legitimate, so that arbitrary distinctions are experienced as natural. Symbolic violence is the process by which the dominated accept the terms of their domination because those terms are the only ones available for thinking about it. This is close to Lukes's third face with a mechanism supplied — the habitus, formed by position, generating preferences that fit the position.

Michael Mann's The Sources of Social Power (1986–2012) distinguishes four networks — ideological, economic, military and political — that are not reducible to one another and that combine differently in different periods. It is the most substantial historical treatment, and its central methodological point is that power is not one thing.

Gramsci's hegemony (5.6) is the third face developed with a politics: consent won continuously in civil society, never total, requiring real concessions, and contestable.

And the "power to" tradition — Hannah Arendt, and Talcott Parsons, of all people — insists that power is not only domination but the collective capacity to act, generated when people combine. Arendt: "power corresponds to the human ability not just to act but to act in concert." This is a genuinely different concept sharing a word, and conflating them causes confusion.

The empirical difficulty, honestly

The three-faces framework is more useful analytically than it is tractable empirically, and this should be said.

The first face is measurable and captures the least. The second requires demonstrating that something was kept off an agenda, which usually means finding the comparable case where it was not. The third requires counterfactual claims about what people would want under conditions that do not obtain.

Which is why the best empirical work on power tends to be comparative and historical: Skocpol on revolutions, Mann on power networks, Steven Levitsky and Lucan Way on competitive authoritarianism, Gilens and Page on policy responsiveness (4.6.2, with its methodological disputes). The comparison supplies the counterfactual that a single case cannot.

And the practical rule for a reader: when someone claims power is operating, ask which face they mean and what evidence would show it. When someone claims it is not, ask whether they are only looking at the first.

"Power means getting your way." That is the first face only. The second and third concern preventing contests and shaping preferences.

"Non-decisions are deliberate suppression." The stronger version is institutional: arrangements systematically favour certain interests without anyone acting. The mobilisation of bias is a property of organisation itself.

"The third face is unfalsifiable." Lukes accepts it is contested and value-laden, and argues it can be evidenced comparatively — through relevant counterfactuals — rather than asserted.

"Real interests are what the analyst says they are." This is the standing objection and it has force. Lukes's revised position is more modest: adaptive preference formation is documented, and the alternative — treating all expressed preferences as final — makes indoctrination indescribable.

"Dahl was wrong." His finding that different actors prevail in different domains is real and was a genuine advance. His method captures one dimension.

"Foucault's power is the same as the third face." Foucault denies there is an autonomous subject whose real interests could be identified, which dissolves the concept Lukes needs. They are not compatible positions.

Why it matters

It is the discipline's central concept, and this is the clearest map of it. Three faces, each detecting what the previous one misses, with a stated method and a stated difficulty for each.

It supplies a permanently useful question: not "who has power here?" but "which face is operating, and what would count as evidence?" Applied to a workplace, a family, a government or a market, it disciplines a word that is usually used loosely.

It shows a concept being properly contested. Lukes's second edition, conceding substantial ground and defending what remained, is a model of intellectual conduct — and a demonstration that "essentially contested" does not mean "impossible to reason about."

And it names the thing conflict theory needs and does not supply. Coser and Dahrendorf describe conflict; this lesson describes what determines its outcomes and, more importantly, what determines whether it occurs at all. The conflicts that do not happen are the ones that tell you most about a society, and only the second and third faces can see them.

The first face (Dahl): "A has power over B to the extent that he can get B to do something that B would not otherwise do." Method: observable conflict over key decisions, and who prevails. Checkable, and captures only power that produces a visible contest.

The second face (Bachrach and Baratz, 1962): non-decision-making and the mobilisation of bias — "organisation is the mobilisation of bias. Some issues are organised into politics while others are organised out." It operates through rules of procedure, agenda control, anticipated reactions and definitions of the possible — and often nobody decides: the arrangement does the work.

The third face (Lukes, 1974): shaping perceptions, cognitions and preferences so that people accept their position — "is it not the supreme exercise of power to get another or others to have the desires you want them to have?" Its feature is that there is no conflict at all, not suppressed conflict.

It requires a concept of real interests, which Lukes concedes is "ineradicably evaluative and essentially contested." The objections — who decides, and the unfalsifiability problem — are serious. His 2005 revision conceded that power is not only "power over," that it can be productive, and that the third face operates in degrees and requires comparative counterfactual evidence rather than assertion.

Beyond the three: Foucault, for whom power is relational, dispersed and productive, and who denies the autonomous subject Lukes needs; Bourdieu's symbolic power and symbolic violence, which supply the third face with a mechanism; Mann's four irreducible networks — ideological, economic, military, political; Gramsci's hegemony; and the "power to" tradition (Arendt, Parsons), a genuinely different concept sharing a word.

And the practical rule: when power is claimed, ask which face and what evidence; when it is denied, ask whether only the first is being examined.

First face of power — prevailing in observable conflict over decisions.

Second face of power — preventing issues from reaching decision through agenda control.

Non-decision — the suppression or thwarting of a challenge before it becomes a decision.

Mobilisation of bias — Schattschneider's term: organisation systematically favours some conflicts and suppresses others.

Third face of power — shaping perceptions and preferences so that no grievance forms.

Real interests — what a person would want under conditions of relative autonomy; the contested premise of the third face.

Adaptive preference formation — the adjustment of wants to what is available; Elster's sour grapes.

Polyarchy — Dahl's term for rule by multiple competing minorities.

Symbolic power / symbolic violence — Bourdieu's terms for imposing categories of perception as legitimate, and for the dominated accepting the terms of their domination.

Power to / power over — collective capacity to act, versus domination; distinct concepts sharing a word.

One — find a non-decision. In an institution you know, identify something that never reaches the agenda. Ask what rule, procedure or assumption keeps it off, and whether anyone had to do anything.

Two — test for adaptive preference. Find a case where people say they do not want something they have no access to. Then find a comparable group with access. If their preferences differ, you have your counterfactual.

Three — ask which face. Take three claims about power you have read this week. For each, identify which face is being described and what evidence is offered. Most journalism uses the first exclusively.

Four — look for the mobilisation of bias. Take any meeting's standing agenda. Ask what categories of issue it has no slot for, and whose interests that serves. Nobody will have intended it.

Five — read Lukes's afterword. The 2005 edition's response to critics is about forty pages and is one of the better examples of a scholar revising a position under pressure without abandoning it.