Customary Law and Popular Sovereignty

Following are lightly edited remarks delivered at a panel on “Unwritten Law,” held at the annual meeting of the Association of American Law Schools on January 6, 2022. These informal remarks are of course not intended to be rigorous or comprehensive, merely suggestive. Many thanks to organizer Robert Leider and fellow panelists Jeremy Waldron, Steve Sachs and Ashraf Ahmed for their thoughts and contributions.


It’s difficult to know how to discuss the topic of unwritten law in twelve minutes. The topic is an incredibly heterogeneous one, full of analytic complications that require endless preliminary distinctions, such as the extremely pellucid and not at all confusing distinction, beloved of legal historians, between legal custom and customary law. So naturally I propose simply to heroically ignore all those conceptual preliminaries and tell you how I changed my mind on a fundamental substantive issue: the relationship between popular sovereignty and custom. 

I used to think that, given a commitment to popular sovereignty and democratic principles, customary law was at best suspect. Now I think it’s not true that custom is systematically more suspect than written legal instruments, such as statutes, from the standpoint of popular sovereignty and democratic principles, although of course in particular cases one may be more or less suspect than the other. 

I’ll focus the issue by reading a single, very famous passage from the classical law, one that I’ve been mentally wrestling with for some time. It is from the Digest of Justinian, compiled and promulgated in the 6th century A.D. by the Emperor Justinian (in Alan Watson’s translation). The passage is quoting the Roman jurist Julian, and it says the following: 

“Age-encrusted custom is not undeservedly cherished as having almost statutory force, and this is the kind of law which is said to be established by usage. For given that statutes themselves are binding upon us for no other reason than that they have been accepted by the judgment of the populace, certainly it is fitting that what the populace has approved without any writing shall be binding upon everyone. What does it matter whether the people declares its will by voting or by the very substance of its actions?”

As of 2015, I thought that this grounding of customary law in popular sovereignty, giving it (almost) statutory force, was highly suspect, even downright wrong. And I published a piece called “Conventions in Court” in the Dublin Law Journal, criticizing the democratic credentials of unwritten law. Let me explain my argument and why I now think it is wrong, indeed fallacious. 

The argument was that conventions suffer from a democratic deficit with respect to principles  such as accountability, responsiveness, deliberation and transparency. I wrote that “there is no well-defined office, official or institution who chooses among possible … conventions and promulgates them.” Furthermore, conventions or customs are fundamentally equilibria, which by definition cannot be changed by the unilateral action of any citizen. But nothing in the nature of equilibria, even equilibria followed from an internalized sense of legal obligation rather than fear of sanctions, guarantees that they are actually good according to any conception of the general welfare. What if the equilibrium is bad, or at least a citizen believes it to be bad?  In that case, I argued, democratic values “require that there be, at least in principle, some well-defined institutional mechanism by which citizens might put their claims before the polity, with at least a theoretical chance of persuading fellow citizens to change the rules accordingly.” To adapt a line from the great legal theorist Henry Kissinger, if I want to change a statute, I can call my legislator, but if I want to change a convention or custom, who exactly do I call? 

I now think there was a basic fallacy pervading this argument, one that I have diagnosed in the works of others and that I should have detected in my own argument. I refer to Harold Demsetz’s “Nirvana fallacy.” I had implicitly compared an idealized case of the enactment of a written legal instrument, like a statute, issued by an accountable, responsive and deliberative institution, on the one hand, to a jaundiced, worst-case example of custom on the other. We should however compare like with like, ideal with ideal or real with real or both. To that end, let me both state some of the ideal justifications in principle that the Roman lawyers give to ground the customary law in popular sovereignty, and, conversely, pour some skeptical acid on the democratic credentials of statutes. 

As to an idealized version of custom: the first justification of customary law in terms of popular sovereignty is unanimous or near-unanimous consent. Justinian’s textbook for lawyers says that “Long-standing custom founded on the consent of those who follow it is just like legislation.” Now, accountability and responsiveness and so on are all important democratic values, but they have an indirect or second-order character; they come to the fore when citizens select representatives who make law on their behalf. We ought not overlook the more direct, first-order, simpler and more basic possibility that everyone or nearly everyone simply agrees on rules of behavior with internalized normative force for the commonwealth. If we imagine all citizens looking at each other and saying “well of course such and such is and should be the rule of behavior for us,” it’s hard to imagine improving on that from a democratic standpoint. And perhaps, in some cases anyway, custom is like that. 

Note that there is a deeper point here, I think, than the sheer democratic credentials of unanimous or near-unanimous consent. Consider the following claim from the Digest, quoting the jurist Paul:

“This kind of law [i.e. customary law] is held to be of particularly great authority, because approval of it has been so great that it has never been necessary to reduce it to writing.”

Here the point is that statutes are often written down precisely because the democratic community is bitterly divided over them. Conversely, when the consensus of the community is sufficiently profound and sustained, no one thinks it necessary to embody that consensus in a formalized written instrument. Indeed they may well not even contemplate doing so — and after all, why should they? No democratic code could or should possibly capture all the background assumptions of any community about normative behavior, even legally normative behavior, and it would be kafkaesque to suggest that it do so.

The second justification is the test of time – the thought that customs observed over a long period of time must reflect an enduring popular consensus. The Digest again:

“But we also keep to those rules which have been sanctioned by long custom and observed over very many years; we keep to them as being a tacit agreement of the citizen, no less than we keep to written rules of law.”

Here I pointed out that there is no general mechanism that ensures, systematically, that a long-term equilibrium actually tracks democratic consensus, that it really does represent a “tacit agreement of the citizenry.” The test of time is a mere possibility theorem, not a robust general mechanism. I still think that argument was not wrong, as far as it went, but it was fatally noncomparative; it didn’t compare the possible failures of the test of time with the possible failures of democratic consensus underlying either statutes enacted by representative legislatures, or even legal instruments enacted by processes of direct democracy, such as referenda.

The problem here is that there is also no robust general mechanism ensuring that statutes track democratic consensus or some sort of agreement of (a majority of) the citizenry. The whole thrust of modern public choice theory is to point out a series of possibility theorems and mechanisms by which statutes, however formally enacted, may not reflect broad democratic consensus and may impose diffuse costs on the affected citizenry to benefit the interests of a small minority. Of course particular customs may be elite customs, but statutes may also be elite projects, and with some frequency are. So that if we really think through the Nirvana fallacy problem, we see that again there is no systematic difference between custom and statutes in this regard. Or at least we see that there are plausible cases on both sides of the divide in which customs and statutes are, or are not, elite projects; the elite-popular line is different from, and perhaps orthogonal to, the custom-statute line. The answer to the Kissinger question, then, is that I may be able to call my legislator, but my legislator may then inform me (at least if he is given truth serum) that extracting resources from me and most of the diffuse majority, or violating what we took to be our rights, was precisely the point of the legislation, rather than some sort of unintended consequence, and that nothing can be done about it. In general, if we compare like with like, nonideal with nonideal, there is no general reason to think that custom fares systematically worse on this dimension of tracking genuine democratic consensus or even just majority views.

It’s true that unwritten equilibria are difficult for those harmed to change unilaterally, but of course statutes are too. As St. Thomas Aquinas puts it, “although each single person cannot make laws, yet the whole people can.” And by laws, he means here both written and customary laws. For both statutes and customs, there are all sorts of ways an individual or small group may propose change, including by violating the statute or custom. If a critical mass approves of the proposal, or at least does nothing to enforce the prior norm, then the law in any operative sense changes at least de facto, but this process is not different with respect to custom and statutes. There is of course an elaborate body of theory and practice on change of customary law by objection or non enforcement in international law. Aquinas here helps us avoid a fallacy of composition: it may be true that I cannot change custom unilaterally, and also that you cannot change custom unilaterally, but it may nonetheless still be true that we (or a critical mass of us) can change custom together, by joint action or by our interaction. And so too with statutes; there is no general difference between the two in that regard.

So, to conclude, I now see no general reason to think that custom is more suspect from the standpoint of popular sovereignty and democratic values than are formal written instruments like statutes. What follows from that, if anything, for the hierarchical relationship between customary law and enacted law — in particular, what if anything follows for the longstanding issue whether custom can repeal statutes, de jure or de facto, under the doctrine of desuetude — is a separate set of questions, which I don’t have time to take up here, but I can discuss later if anyone is interested.

Adrian Vermeule