We use the word “authority” to mean lots of things — police and state actors are “the authorities”, an expert may be “an authority on the matter, etc. But I want to suggest that it is very useful to think of authority as a characteristic of information in a social context. In particular, information is “authoritative” when some community of people to coordinate upon it and behave as if it were true, regardless of whether or not the information is in fact true, or even of whether the individuals doing the behaving personally believe it to be true. If information is authoritative, members of the community behave as if the information is true even despite strong, often opposing interests in the question. When we claim that someone “is an authority”, we are claiming that the information they produce will (or should) alter behavior within some human community. Authority subsists in the relationship between information and behavior in a social context.
Let’s take an example. A judge, in the context of a trial, is an authority. Suppose a judge pronounces a defendant guilty, despite her protestations of innocence. Both parties have produced information. But it is the information produced by the judge that guides the behavior of the vast preponderance of the community. Suppose the bailiff, who was present for the trial, privately came to a different conclusion than the judge, and believes that the defendant is in fact innocent. The bailiff will nevertheless behave as if she were guilty, taking her back into custody rather than setting her free.
More often than not, there is not so much cognitive dissonance. Most of us, most of the time, take a huge variety of conjectural “social facts” as given, condition our behavior as if they were true, and to the degree that we even give them a second thought, we believe them to be true. I log into my bank’s website, and check the balance of my account. Most of the time, I take the number presented as an authoritative representation of how much money I have “there”. I would prefer, quite strongly, that the number be millions larger, and my deposit balance at a bank is nothing more or less than what the bank acknowledges that it owes to me, so it is in a small way extraordinary that the bank and I are so willing to agree, despite diametrically opposed economic interests on the matter. But the miracle of authority is that it quells many disputes so thoroughly that parties don’t even imagine that there is any ambiguity or question to argue about. Authoritative information presents itself as factual, even when it (like a bank balance) has no external, empirical referent and is purely a social construction.
As surely as we depend upon the laws of physics to suspend us in our fifth floor apartments, we depend upon authority to give structure to our social and economic lives. Our very identities — our names, our credentials, the entity to whom our properties belong — exist as “social fact” by virtue of authority. The production of authority is the production of the social reality upon which we all coordinate. We talk sometimes about “defying authority”, by which we mean resisting some particularly crude and visible attempts to render information authoritative. But for the most part, to fail to coordinate on the “same facts” our community has settled upon comes off not as courageous but as insane. As Ijeoma Oluo writes:
A lot of things in our society are social constructs — money, for example — but the impact they have on our lives, and the rules by which they operate, are very real. I cannot undo the evils of capitalism simply by pretending to be a millionaire.
It’s hard to defy the authority of your bank account, even though the value that ends up there is the result of myriad social and institutional contingencies and is in a certain sense quite arbitrary. Of course we can, and under some circumstances we do, claim our bank balances are wrong. But whatever we believe the “true” figure should be is irrelevant as a practical matter unless and until an authoritative source (in this case, the bank itself) produces it. As individuals, we can dissent, but what makes information authoritative is how a larger community treats it, which often renders our own private judgements immaterial.
If authority is defined by information that a human community behaves “as if” is true, one might conjecture some relationship between processes that produce authoritative information and practices that might be colorably argued to be truth-generating. In most societies, a judge pronounces guilt, rightly or wrongly, after some kind of trial in which evidence is gathered and presented and the facts of the case are argued. In some societies, we might imagine the truth-generating power of legal procedure to be pretty good. In other societies not-so-much and we might mumble dismissively about “show trials”. As an anthropological matter, it’s clear that having some sort of narrative that connects the information we coordinate upon as “true” to processes that might mean they actually are true is helpful to the production of authority. Let’s call this “soft power”. On the other hand, if there are people with economic resources they can withhold to starve you, or with the physical capacity to harm and imprison you, they can, um, persuade the collective you to behave as if the information they produce is true. Let’s call this “hard power”. In nearly all societies, authority is generated by a combination of hard and soft power. We have a dispute. Is this house your house or my house? I can show you the deed to the property, evidence of a transfer of funds for its purchase, all of those things. But perhaps you can do the same. Our economic interests are opposed, and our standards of evidence are unlikely to be neutral. If we bring our dispute to the attention of the broader community, is it hard power or soft power that wins the day? Who knows? In most societies hard power is usually deployed under a fig leaf of soft power (the police evict me or they evict you following a trial with evidence and all of that). But sometimes this fig-leaf is so thin as to be meaningless. Even under procedures we consider decent, the ultimate “truth of the matter” will very often remain uncertain and contestable after all of the formalities have been deployed. A verdict will nevertheless be pronounced, and we will collectively behave as if the unknowable truth is known. Sure, that’s true in part because we believe hard power might eventually be deployed against those who defy the decision. But then, if the procedures were truly decent, you can argue that it is those who manage the institutions of soft power that determine the direction of the gun. And in practice, it’s rare for any overt hint of the exercise of hard power to be required to persuade most of use to behave as if some set of social facts is true.
It is a mistake — an easy, common, and foolish mistake — to imagine that hard power tells the whole story, that “how many divisions does he have?” is the beginning and end of the question of authority. The exercise of hard power is expensive. Even from the perspective of a “rational bandit” (ht Elaine Ou) whose ultimate source of legitimacy is the barrel of the gun, producing information about the world that causes people to behave in the ways you would like them to behave is cheaper and more efficient than frog-marching everybody everywhere all of the time. You’ll have more firepower available to defend your domain and plunder new lands if you can point your guns outward and your subjects still do what you want, then if you have to be pointing your guns inward at everyone. The law of the jungle selects for “voluntary compliance”. Further, relying upon the exercise of hard power butts up against the same informational limits that give rise to the economic calculation problem. No leader or ruling junta can even figure out what even they want the millions of people they rule to be doing all the time, let alone stand behind them with a gun and make them do it.  It’s much better if you can shape social reality so that people behave in roughly the way you’d like them to behave without your even having to tell them specifically what to do all the time, let alone point your scarce guns at them.
Communities want authoritative information on which they can coordinate. All sorts of valuable forms of collaboration are practical only when we are not bickering over every contingent and contestable social fact. Even flawed authority is better than no authority, and authority has network effects (the more people act as if some set of information is true, the more costly it is for others not to also act as if it were true). Nevertheless people dislike the cognitive dissonance associated with acting “as if” certain facts are true when they privately believe them to be false. We denote authority “Orwellian” when it is clumsy, when under threat of hard power or overwhelming convention it becomes in our interest to behave as if things we think false are true. Much more powerful (and so potentially dangerous) is authority that is not Orwellian at all, whose “soft power” is sufficiently persuasive that we privately believe nearly all of the social facts that we collectively coordinate upon. 
Authority, like most coordination problems, is relatively easy at small scale. We can choose a wise woman to judge and declare. However, the benefits of coordination grow nonlinearly with scale (“agglomeration effects”). Economic and military power accrue to polities that are able to produce authoritative information that coordinates behavior over large geographies and populations with minimal exercise of costly hard power. Modern, developed countries devote a significant fraction of their energies to the production of authority. Much of the work of the legal and accounting professions in the private sector, and of courts and the regulatory state in the public sector, is devoted to the production of authority. Finance, which concerns itself with contentious questions of who owns what and how scarce resources should be invested, is necessarily intertwined with the machinery of authority. The court system, the training and professional standards that apply to law and accountancy, the bureaucratic procedures that surround the operation of the regulatory state, all embody complicated sets of compromises between interests (which try to shape the social facts we coordinate upon for their own benefit) and the broader necessity of maintaining “credibility” and “legitimacy” so that recourse to hard power in shaping social behavior is rare. The production of “soft power” authority is the sine qua non of the modern state, and a source of competitive advantage for those who do it well.
The production of authority is a socio-technological problem, albeit a far-from-neutral technological problem (but technologies are never neutral). Although “soft power” authority is cheaper than resorting frequently to hard power to manage behavior, the systems by which we currently manage the production of authoritative information remain extraordinarily expensive — lawyers and judges and regulators and bankers don’t come cheap! Contemporary practices are also discriminatory. Most of the work of producing authority is done by a particular professional class, which is often socially and geographically segregated from the rest of the polity. Enfranchisement in the production of authority is skewed towards those within that class or capable of accessing (and paying) members of that class. This is problematic on technical grounds (those whose interests and perspectives are not included in the production of authority are more likely to privately dissent, diminishing the effectiveness of authority at coordinating behavior and increasing the degree to which hard power may be required), and on ethical grounds (the facts upon which we coordinate social behavior largely determine social outcomes, the determination of those facts is never neutral and always to a very large degree arbitrary).
The entropy of an individual human body is extraordinary large. It is a miracle, the degree to which even people we lock up as batshit crazy control and manage that entropy to yield elaborately functional behavior. The entropy of a human community or society is many of orders of magnitude larger, the space of potential social behavior is incomprehensibly vast and multidimensional. The behavior of so many bodies must be improbably constrained and synchronized to yield functional societies, which requires elaborate social coordination. Authority is an invisible drummer that helps to organize this dance. We construct authority. How we construct it is among the most important social, ethical, and technological problems we face.
 As with questions surrounding socialism and economic calculation, there is an case to be made that emerging information technology will render practical more pervasive and direct forms of state compulsion. So, um, exciting.
 When we are in it we are in it, but while we are thinking about authority from a distance, let’s remind ourselves that the absence of cognitive dissonance does not imply the presence of truth from some larger perspective. History is full of communities that produced authority effectively (in the sense that the “facts” that conditioned social behavior were widely privately believed), but which we now look back upon as having been egregiously in error, scientifically or morally. We might be “wrong” too. But authority is not about truth or falsehood in the eyes of God. It is about coordinating human behavior.