The previous essay asked where power comes from. The answer: mutual recognition combined with asymmetry, operating across three overlapping layers. This essay moves to the next question — once power exists, what forms does it take? And how does it actually work when two people (or two groups, or a person and an institution) come into contact?

These are distinct questions. The first is structural: how is power distributed and organized? The second is mechanical: in any given encounter, what are the parties doing, and how do their actions shape each other? They belong together, though, because they point toward the same conclusion: power is never static. It has shape, but shapes shift. It has logic, but the logic is always generating new conditions.

Concentration and Dispersal

The most basic structural distinction is between concentrated and dispersed power. A single individual can hold power over thousands; or power can be spread across a network of many nodes, with no single node holding all of it. Neither arrangement makes power more or less real. They are different structural configurations with different properties.

Concentrated power moves fast. Instructions flow efficiently from the center, and the response to disobedience is swift. But it is fragile: remove the center and the whole structure may collapse. Dispersed power is harder to kill — damage one node and the network reroutes around it — but it is slow and expensive to coordinate. Most real power structures are some mixture: there are central nodes, but those nodes depend on a distributed network of secondary actors. Purely concentrated or purely dispersed arrangements are rarities.

This matters for a practical reason. The intuition "remove the leader and the power structure falls" is often wrong. If power is distributed through a network, the removal of the top node simply opens competition among the secondary nodes. The structure reconstitutes. The faces change; the game continues. Many revolutions discover this the hard way.

Three Media

Power needs something to travel through. Corresponding to the three source layers described in the previous essay, there are three media through which power operates.

The first is force and coercion. Soldiers, police, physical restraint, the threat of pain or confinement. Force acts directly on the body. It requires no belief from the subjected party — only an accurate understanding of consequences. It is fast, visible, and efficient in the short run. Its disadvantage is cost: maintaining a credible threat requires constant resources, and force is reactive by nature, triggered only after a violation has already occurred.

The second medium is narrative and symbol. Titles, ceremonies, stories about who deserves obedience and why — these are far cheaper than force because they don't require a separate payment for each act of compliance. If you genuinely believe the king was appointed by heaven, no soldier needs to escort you to your knees. But narrative requires maintenance. A story that is never retold fades; a story that is publicly contradicted loses power. This is why those who hold power spend such effort on ritual, on controlling education, on managing the public record. These are not vanity projects. They are upkeep on the medium that power travels through.

The third medium is recognition itself — the direct, uncoerced acceptance of someone's authority as legitimate. This is the most efficient medium and the most fragile. When a person's authority is genuinely accepted — not from fear, not from habit, but from actual conviction — the power relation requires almost no maintenance. But recognition can be withdrawn in an instant. Unlike force, it leaves no residue when it disappears. Unlike narrative, it has no inertia. Recognition is a live relationship: given now, revocable now.

Any functioning power relation draws on all three media at once, in varying proportions. A modern state has police (force), legal codes and national symbols (narrative), and some degree of genuine public acceptance (recognition). When one medium weakens, the others bear more weight — which is why authoritarian regimes that have lost legitimacy become ever more reliant on force, and that reliance further erodes what remained of narrative and recognition.

Stacked and Flat Structures

A further structural dimension: power can be arranged in layers, or it can operate across a flat network of peers.

Stacked structures are the most common arrangement for large-scale power. An emperor above provincial governors above county officials above local strongmen. Instructions pass down; resources pass up. Each layer exercises power through the layer below. The advantage is reach — a single emperor can "govern" hundreds of millions of people he will never meet, through sufficiently many layers of intermediaries. The disadvantage is distortion: every handoff is an opportunity for instructions to be reinterpreted, filtered, or quietly redirected to serve the intermediary's own interests.

Flat structures — where power holders are peers who coordinate rather than command — are more resilient. There is no apex to be removed. The costs are coordination and speed: flat networks take longer to move and are harder to point in a single direction. Democratic systems of checks and balances, market competition, and certain forms of organized crime all embed flat-structure logic alongside their hierarchical elements.

Here is a practical implication that follows from both arrangements: institutions do not exercise power. People at key positions in institutions exercise power through those institutions. The distinction sounds subtle but it changes what we expect from institutional reform. When an institution is abolished, the individuals who held key positions within it are freed — free to reconstitute power relations through a different institutional form. This is why replacing one set of institutions with another, without changing the underlying recognition relations among individuals, so often produces the same outcome in new clothing.

Three Positions in Every Encounter

Now the shift from structure to mechanism. In any concrete encounter where power is at work — a boss issuing instructions, a state imposing a restriction, a parent setting a rule — there are always three positions.

The first is the one exercising power. This party initiates the asymmetric shaping: makes the demand, issues the directive, arranges the situation so that the other party's available choices narrow. The exercising party sets the frame of the encounter.

The second is the one subjected to power. This is the field in which the asymmetry registers. The subjected party is not a blank surface waiting to be inscribed. They have an inner life, a capacity for judgment, a set of ongoing calculations. The crucial insight is this: how the subjected party registers the power being exercised determines the nature of the power relation.

Fear and admiration are both forms of registration. Resentful compliance and genuine acceptance are both forms of registration. But they produce radically different power relations. Compliance driven by fear is unstable — reduce the threat and it begins to dissolve. Compliance driven by genuine conviction requires no continued threat to sustain, but it is revocable: change the conviction and the compliance stops. This is why power holders cannot be indifferent to the inner states of those subjected to their power. Those inner states are not external to the power relation. They are partly constitutive of it.

The third position is the observer. Third parties who witness the exercise of power are not irrelevant bystanders. Their registration matters, sometimes decisively. A public exercise of power that humiliates the subjected party damages that party's standing in the witnesses' eyes — diminishing their recognition more broadly. A public display of authority that witnesses find appropriate and proportionate reinforces the authority by having been seen. This is why power is so attentive to public performance. The pomp of courtrooms, the choreography of state ceremonies, the staging of military parades — these are not decoration. They are management of the observer position.

The Active Subjected Party

The most consistently underestimated aspect of power's operation is the activity of the subjected party.

We instinctively picture power as an arrow fired downward into a stationary target. But the subjected party is not stationary. They are interpreting the exercise of power. They are calculating: What happens if I comply? What if I don't? Is there a way to appear compliant while actually evading? They develop strategies — open compliance, surface compliance with private evasion, quiet resistance, outright refusal when conditions allow. Every exercise of power generates a response; every response changes the field for the next exercise.

Power relations are not one-time events. They are ongoing, adjusting interactions — radically asymmetric, often, but interactive nonetheless. This is why the most comprehensive systems of control tend to generate the most elaborate cultures of evasion. When power blocks every visible exit, the pressure does not dissipate. It finds lateral and underground channels. Censorship produces coded language. Surveillance produces double lives. Enforced conformity produces expert dissemblers. The subjected party's agency does not disappear when their options are compressed. It changes form.

The Claim to Justification

Every exercise of power carries an implicit claim: I have the right to ask this of you. This claim can be accepted, questioned, or rejected.

The most efficient state for this claim is one where it is never questioned — not because dissent is suppressed, but because no one has thought to raise the question. The idea that the king rules by divine mandate is not experienced as a proposition requiring defense; it is experienced as obvious background. Power in this mode is maximally low-cost. But this state is historical, not permanent. Once someone asks "by what right?", the claim becomes a question that needs answering.

When the justification claim is consistently challenged by enough parties, the power relation becomes unstable. The power holder faces a choice: provide new justification, escalate force to suppress the challenge, or attempt both. Each option has costs. Providing new justification implicitly concedes that the old justification was insufficient. Escalating force increases resource requirements and draws down whatever remained of the narrative and recognition media. Neither option restores the comfortable state where the claim went unquestioned.

This is why power cannot rest on force alone. A power relation sustained purely by coercion faces costs that scale with the number of subjected parties and grow over time as resentment accumulates. The next essay will examine what exactly accumulates — and why that accumulation is not a practical inconvenience but a structural feature of every power relation.