Anarchy Is Not Chaos

Order without a sovereign, at every scale

There is a genre of political conversion story that circulates endlessly, and it always has the same shape. One version I encountered ran like this:

I really thought communism was the answer to all of society’s problems until I tried to run a community garden.

To which another writer added:

When I was younger, anarchy looked a lot more feasible to me until I ran a Burning Man theme camp for a few years. Interpersonally, I still like treating everyone like they’re sovereign, but anything I’m in charge of is now a benevolent dictatorship.

My reply was short:

To be fair, anarchies do not preclude organizations run like benevolent dictatorships.

That reply is this chapter. The conversion story feels devastating because it seems to be an empirical refutation: I tried anarchy, it collapsed into squabbling, so I appointed myself dictator, and things finally worked. Case closed. But look at what the story demonstrates. A group of people voluntarily formed a project. When flat consensus failed them, they voluntarily adopted centralized leadership. Anyone who disliked the arrangement was free to walk away and camp somewhere else. Nobody was conscripted, nobody was taxed, nobody was forbidden to leave. The theme camp did not refute anarchy. The theme camp was anarchy — anarchy discovering, as it reliably does, that some tasks want a hierarchy.

The mistake is in the definition, and the definition matters enough to get right, because on it turns the entire question of what a stateless order can and cannot be.

What Anarchy Actually Means

Anarchy is not chaos. It is not the absence of organization, and it is not a utopian demand that all human activity be conducted by endless consensus meetings. Anarchy means the absence of a coercive monopoly on authority — the absence of an institution that declares itself sovereign over everyone in a territory regardless of their consent. That is all it means. Within that space, people are free to organize however they wish, and very often they choose hierarchies.

Notice what the definition does not prohibit. It does not prohibit leaders, bosses, commanders, or founders with total operational control. It does not prohibit rules, contracts, discipline, or expulsion. It prohibits exactly one thing: the structure that claims authority over people who never agreed to it and punishes them for trying to leave. The previous chapter made the underlying logical point — rights enforcement entails coercion, but coercion does not entail monopoly. This chapter draws out the organizational consequence: hierarchy does not entail monopoly either. Authority and sovereignty are different things, and almost every argument against anarchy trades on confusing them.

This is institutional sovereignty: a claim to final, coercively enforceable authority within a jurisdiction. It is not the reflective sovereignty of an agent whose own reasons govern its transitions, developed in Volume 4. An institution can claim political sovereignty without possessing a unified mind, and an agent can be sovereign in the reflective sense while living under external constraint. The shared word marks supremacy in two different domains; it does not make them the same property.

A “benevolent dictatorship” within an anarchic framework is therefore not a contradiction. It is an organization people freely opt into, submit to for as long as the arrangement benefits them, and retain the right to exit. A startup founder who overrules her engineers, a theme camp lead who assigns the shifts nobody wants, an orchestra conductor whose interpretive word is final — all of these are dictators in the operational sense, and all of them are fully compatible with anarchy, because their authority extends exactly as far as their followers’ continued consent and not one step further. The moment authority becomes coercive — the moment leaving is punished or forbidden — it ceases to be anarchic. Not before.

This is why functioning anarchies, historical and contemporary, contain a mosaic of organizational forms rather than a monoculture of consensus circles: worker co-ops beside entrepreneurial ventures, communes beside trade unions, flat collectives beside charismatic leaders who attract devoted followings. Human coordination requires different structures in different contexts — sometimes horizontal, sometimes steeply hierarchical, usually a hybrid — and no committee can know in advance which structure a given task wants. What all the forms in the mosaic share is what they lack: the monopolistic claim to rule everyone, consenting or not.

So the lesson of the community garden is not that anarchy fails. The lesson is that flat consensus is one organizational tool among many, useful for some tasks and hopeless for others, and that a free order lets people discover which is which and reorganize accordingly. Anarchy does not mean doing away with leadership. It means leadership must always remain provisional, conditional, and chosen.

The Exit Criterion

Condense the argument and one indispensable test remains: is exit realistically available and unpunished? It is not the only test. Dependency, sunk costs, monopoly control of necessities, information, legal capacity, and effects on third parties can make formal permission to leave a fiction. The theme camp and the totalitarian state remain distant cases, but much institutional life occupies the terrain between costless exit and a shot at the border.

Internal structure, rules, dependency, voice, and remedy still matter. Cheap, informed exit strongly disciplines authority because continued participation must be re-earned. Blocked exit is powerful evidence of domination. Between those poles, leaving may be legal but destructive of livelihood, family, identity, accumulated assets, or access to essential infrastructure. Governance must improve exit while also constraining abuses that cannot be cured by telling the burdened person to move.

I flag this test here because it is the load-bearing principle of everything that follows in this volume. Exit is not one liberal value among others; it is the mechanism that keeps all the other values honest. When I argue later that large-scale coordinating structures escape evaluation and control precisely as they close off exit — and that the only architecture worth building treats exit supremacy and asset portability as constitutive, not decorative — that is this criterion, scaled up: The Limits of Leviathan. And when the volume ends with governance designed from scratch, for the hardest environment there is, engineered exit is the first requirement written into the constitution: Axiocracy. The principle enters the book in a theme camp. It does not stay there.

The Largest Anarchy in the World

Now scale up — past the camp, past the firm, past the nation — to the largest coordination system humans operate. Here the argument stops being hypothetical, because at the top of the world’s political architecture there is no sovereign at all, and there never has been.

No world sovereign stands above states, but it does not follow that international law is etiquette. Treaties create legal obligations; customary law can bind without a new signature; UN members are bound by the Charter; Article 25 requires members to carry out Security Council decisions;¹ and parties undertake to comply with ICJ judgments in cases to which they are parties.² Enforcement is decentralized, selective, and power-sensitive. That makes international law unlike domestic law, not unreal.

Apply the definition. Anarchy means the absence of a sovereign authority above the parties involved. By that definition the international system is anarchic to its core. States form treaties, submit disputes to tribunals, gather in the halls of the UN — but they do all of it voluntarily. No higher authority compels them to join, and no higher authority can compel them to stay. The only real constraint on a state is the threat posed by other sovereigns.

The contrast with domestic law makes the structure vivid. Domestic law rests on institutions of enforcement: police, courts, and prisons supply a standing deterrent, so that murder statutes, while they do not eliminate murder, reliably raise its expected cost for everyone in the jurisdiction. International law has no such machinery — no global police force, no prisons, no monopoly of violence. Enforcement is contingent, improvised, and political. Small states are sometimes punished for violations. Great powers ignore adverse judgments with impunity unless some other sovereign chooses, at its own risk and expense, to intervene.

Major powers have repeatedly used force despite serious legal objections, demonstrating weak and unequal enforcement. But “the law prevented nothing” is too strong: legal rules shape coalitions, sanctions, access, domestic authority, diplomatic cost, reparations claims, and the language through which conduct is contested, even when they fail to stop an invasion. Power conditions enforcement; law also conditions the exercise and evaluation of power.

Coordination Mistaken for Sovereignty

It would be a mistake — the opposite mistake, and just as disabling — to conclude from this that international order is an illusion. The order is real. Treaties stabilize trade across oceans between parties who will never meet. Diplomatic conventions let bitter adversaries exchange messages without exchanging fire. Postal systems interoperate, aircraft cross borders on standardized protocols, embassies enjoy an immunity that is violated rarely enough to make each violation a scandal. Most states comply with most of their commitments most of the time, because predictability is valuable and a reputation for keeping agreements is an asset that takes decades to build and one betrayal to spend.

Much of this order arises from mutual interest, reciprocity, legitimacy, domestic incorporation, institutional routine, and the shadow of retaliation rather than subordination to a world sovereign. That is still governance and law in a polycentric form. The instructive point is not that international law is fake, but that legal order can exist with incomplete and distributed enforcement — and that the distribution leaves predictable power asymmetries.

And this is the fact that should reframe every instinctive dismissal of stateless order: we already live in global anarchy. Every statist on Earth conducts his life inside a functioning anarchic system and hardly notices. The world has no sovereign above the state, no binding parliament, no supreme court with a police force at its disposal — only a lattice of agreements sustained by reciprocity and shadowed by power, sovereignty all the way down. The trade that feeds him, the treaties that keep his airspace navigable, the conventions that make his passport mean something abroad: all of it is order without a ruler. The argument between archist and anarchist is therefore not over whether order can exist without a sovereign — it demonstrably does, at the largest scale there is. The argument is over where else the same architecture may be trusted.

Real, and Asymmetric

Honesty requires that I not end on that triumphant note, because the international system demonstrates something else too, and it cuts the other way.

The order that anarchy generates is real, but it is not symmetric. Reciprocity constrains states in proportion to what other states can credibly do to them — which means it constrains the middle of the power distribution far better than the top. A mid-sized state that tears up its treaties faces sanctions, isolation, and coalitions of the offended; a hegemon that does the same faces strongly worded letters. I examined this structure in Law and Power: law binds only where power cannot override it, and at the international scale that truth operates without even a pretense of concealment. The gravity of the system weakens with altitude. The etiquette of anarchy is enforced on those who can be retaliated against, and merely recited at those who cannot.

So the international system is simultaneously the existence proof and the cautionary tale. It proves that large-scale order needs no sovereign — that reciprocity, reputation, and mutual deterrence can sustain cooperation among armed parties with no referee, for centuries. And it proves that unstructured reciprocity alone leaves the strongest actors effectively unbound, free to write rules for others that never reach themselves. Anarchy solves the problem of the tyrant above; by itself it does not solve the problem of the giant among.

I do not take that as a refutation. I take it as a design specification. The question it poses — how to build orders in which constraint stays reciprocal even when power is unequal, in which the exit criterion binds the strong as firmly as the weak — is the question the rest of this volume works toward: first mapping the region of political phase space where such orders can live at all, then designing one. The theme camp got the micro-architecture right by accident: leadership provisional, conditional, and chosen, with exit unpunished. The remaining work of political philosophy, as I see it, is getting the same properties at scale — on purpose.