Why do we drive on the right? Not because the right is better than the left — half the world does the opposite, and fares no worse — but because everyone else does, and to do otherwise is to die. This is a convention: a rule that binds not because it is wise or just or natural, but simply because it is kept. Each of us follows it because all the others do, and all the others follow it for the same reason, so that the whole vast coordination hangs in the air, supported by nothing but the mutual expectation that it will go on. A convention is arbitrary in its origin and iron in its grip — and once we see how much of social life runs on conventions, we will not look at custom the same way again.
The example of the road is chosen for its clarity, not its importance. What it shows in pure form is the structure of a coordination problem: a situation in which what each of us should do depends entirely on what the others do, and in which there is more than one way for us all to succeed. If everyone drives on the right, I had better drive on the right; if everyone drives on the left, I had better drive on the left; and the two arrangements are, in themselves, equally good. What matters is not which we choose but that we choose the same, for the cost of failing to coordinate falls on all alike. A convention is whatever settles such a problem: a regularity in conduct that we all conform to, expect the others to conform to, and conform to partly because we expect it of them. It requires no agreement in words, no signed compact, no authority to impose it — only that the expectation, once it exists, be mutually held, so that each believes the others will keep to the pattern, and finds it in his own interest to keep to it too.
Once the structure is seen, conventions appear on every side, most so familiar we have ceased to notice them as choices. Language is the great example: that a particular sound should mean a particular thing is a matter of pure convention, as the babel of human tongues makes plain, and yet within a community the assignment is binding past argument, for a speaker who coins his own meanings is not free but unintelligible. Money is a convention of the same family — a token passes for value because everyone accepts it, and would lose all value the instant the acceptance ceased. The measures we trade by, the calendar we keep, the side on which a coat buttons, the words with which we greet and part: each is one selection from many that would have served, frozen into custom and kept in place by the expectations of all. The world we move through is built, to a degree we seldom suspect, of solutions to coordination problems we have long forgotten were ever problems.
The modern understanding of all this we owe to two thinkers a generation apart. The economist Thomas Schelling, studying coordination, observed that people are often strikingly good at it even without communication, converging on whatever solution somehow stands out; and the philosopher David Lewis, taking up the puzzle, gave the convention its rigorous definition. A coordination problem, in their analysis, is a game with several equilibria — several combinations of choices, like all-drive-right and all-drive-left, from each of which no single player would wish unilaterally to depart. A convention is a means of always landing on the same one. And its central property, the one that sets it apart from a law, is that it is self-enforcing: once the convention holds, conforming to it is already in the interest of each, provided the others conform, so that no policeman is needed to make it stick. The arrangement John Nash had defined in the abstract — a point from which no player can profitably deviate alone — turns out to describe the most ordinary facts of custom and manners.
Three features deserve to be named, for the whole argument of the volume turns on them. The first is that a convention is arbitrary: not that it does not matter — coordinating matters enormously — but that some other convention would have done as well, there being nothing in nature that prefers the right of the road to the left. The second is that, for all its arbitrariness, a convention is binding; once in place, the freedom to depart vanishes, and the deviant meets not liberty but a wall, paying for his independence in inconvenience or ruin. And the third is that the convention sustains itself, resting on nothing firmer, and nothing weaker, than mutual expectation — each conforming because he looks for the others to conform, and looks for them to look for it in him, in a tower of expectation that philosophers call common knowledge. No force outside this lattice of belief holds the convention up. Shake the expectation and the rule dissolves, for it never had any other foundation.
Now the convention's significance for the larger argument comes into view. A price is a verdict on a hidden contest, mistaken for a fact of nature — and the convention is that same mistake caught at its simplest, the place where it can be seen with the naked eye. For a convention wears precisely the disguise the whole work is concerned to strip: it presents what is merely one arrangement among many, kept in being by nothing but our keeping it, as though it were the necessary order of things. With the road the disguise is thin and slips at a touch, for no one truly believes the right side is written into the world. But the same form of mistake, in deeper conventions, runs far harder to detect; and where the convention is not a matter of pure coordination but carries stakes — where one of the possible equilibria rewards some party at the expense of another — the appearance of naturalness conceals something more than an arbitrary choice. It conceals a contest — over whose convenience and whose advantage the common rule shall serve — settled long ago and frozen into the way things simply are. The convention is the master mistake in miniature: visible, diagnosable, and a model for all the rest.
It must be said at once, in fairness, that a great many conventions hide no contest whatever. The purest cases — which side to drive on, which sound to attach to which thing, which day to call the first of the week — carry no distributional stakes at all: nobody is advantaged by the right rather than the left, and the one thing of value is that we agree. Convention of this innocent kind is among the great achievements of social life, quietly disposing of the countless questions to which reason gives no answer, and freeing thought for the things that warrant it. The volume has no wish to deny this, nor to cast every greeting and handshake as a mask for power. Its claim is narrower: that the innocence of a convention cannot be presumed from its feeling natural, since feeling natural is exactly what a convention does whether it is innocent or not — so that whether a given custom is mere coordination or something with a victim becomes a question to investigate, never a thing to assume. Even the innocent conventions, moreover, can harden past their use, the arbitrary choice locking in and outliving its reason, so that a people finds itself bound by an arrangement none would now choose, simply because all expect all to keep it.
What the analysis leaves in the hand is less a doctrine than a discipline — a question to be put to whatever presents itself as simply the way things are done. Must it be done so, the question runs, or does it merely happen to be? Asked of the road, the question is idle and the answer obvious. Its edge shows only when it is carried into country where the answer is not given in advance: to the tasks that fall, by custom, to the wife not the husband; to the deference rank is held to be owed; to the shape a creditable life must take. In each the same possibility arises that the road made plain — that what is lived as a fixed feature of the world may prove, on inspection, a convention that has lost the memory of being one, an equilibrium no one now defends because no one now sees there was ever a choice. To have learned merely to ask the question is the first step, and the whole of this plate's work; the asking itself, pressed into the harder cases, is the labor of the plates that follow.
And the choice to begin here, with the convention we can all see through, is the strategy of the volume in little. A price is a verdict on a hidden contest, mistaken for a fact of nature; the convention is the gentlest illustration of the claim, the case in which the contingency is too obvious to deny, offered first so the principle may be granted before its full reach is felt. For the very analysis that dissolves the necessity of the road can be turned, with patience, upon rules sunk immeasurably deeper — into the cadences of manners, the verdicts of conscience, the things a people holds sacred — where the made character is buried beneath generations and feels not like a rule at all but like the grain of reality itself. The convention teaches the move in the one case where it costs nothing to accept. What is left is to follow that move inward, into the norms that resist it, and to find how much of what we take for the order of nature is convention that has forgotten its own name.
A convention is a rule that binds for no reason but that it is kept — one possible arrangement among many, frozen into the way things simply are, and held up by nothing in the world but our expectation of one another.
So the first plate sets down the atom of the normative. A convention is a regularity we keep because the others keep it; arbitrary in that another would have served, binding in that this one now governs, and self-enforcing in that no hand outside our mutual expectation sustains it. It is the simplest of the unwritten rules, and the one in which the volume's whole thesis can be read at a glance: the made arrangement mistaken for the given order, the agreement mistaken for the necessity. The question the volume set itself — whether the rules of conduct are anything more than conventions we have taken for laws of nature — has, for this plainest instance, a plain answer: the convention is exactly that, a necessity that is only an agreement, a law of the world that is only a habit of expectation. What remains in doubt is how much else, looked at squarely, will prove to be the same. The argument remains open.