How to Answer Dworkin’s Argument from Theoretical Disagreement Without Attributing Confusion or Disingenuity to Legal Officials

Abstract

Ronald Dworkin’s argument from theoretical disagreement remains a pressing challenge for legal positivists. In this paper, I show how positivists can answer Dworkin’s argument without having to attribute confusion or disingenuity to legal officials. I propose that the argument rests on two errors. The first is to assume that positivism requires legal officials to converge on precise grounds of law when convergence on more general grounds will do. The second is to construe judicial speech too literally. If we pay attention to the pragmatics of judicial speech, we see that judges do not disagree over what the grounds of law are; they at most disagree over how courts should proceed when agreed-upon, though imprecise, grounds of law underdetermine what the content of the law directs in the case at hand.

Introduction

Suppose that, as sometimes happens, two judges disagree over how to interpret a statute: one says that the statute should be interpreted according to its ordinary meaning; the other says that it should be interpreted according to the legislature’s intent. Ronald Dworkin claims that disputes like this are ‘theoretical disagreements’ over the grounds of law and that they pose a problem for legal positivism. The problem is roughly this: According to positivism, a fact is a ground of law only if legal officials agree that it is. But legal officials engaged in theoretical disagreement disagree precisely over whether some fact—say, a statute’s ordinary meaning—is a ground of law. So positivists must conclude that these officials are either confused (they are debating whether a fact is a ground of law even though their very disagreement precludes it from being so) or else disingenuous (they are just pretending to engage in such a debate).

Dworkin presented this argument in 1986, ¹ but at first little was written on it. ² That changed when Scott Shapiro revived the argument in 2007, calling it “the most serious threat facing legal positivism at the beginning of the twenty-first century.” ³ Brian Leiter then offered an influential response, contending that theoretical disagreement rarely occurs in legal practice, such that positivists should not be concerned about attributing confusion or disingenuity to those who engage in it. ⁴ But recent years have seen pushback against that response from Dale Smith, Barbara Baum Levenbook, and others. ⁵ These writers claim that (i) Leiter underestimates the amount of theoretical disagreement in legal practice and, in any event, (ii) there is prima facie reason to reject a theory of law that requires viewing legal officials as confused or disingenuous.

My aim in this paper is to show how positivists can answer the argument from theoretical disagreement without having to attribute confusion or disingenuity to legal officials. ⁶ I propose that the argument rests on two errors. The first is to assume that candidate grounds of law must be facts like a statute’s ordinary meaning or legislative intent. The grounds of law need not be so precise. Legal officials agree on the grounds of law but at a higher level of generality: they agree, for instance, that a statute or other enacted text’s meaning (its communicative content) is a ground of law, and that a judicial opinion or other precedential text’s holding is as well. These are platitudes, which are stated at such a high level of generality that no one would dispute them. But that is partly the point: positivists expect the grounds of law to be platitudinous.

Legal officials’ agreement on these grounds of law is ‘coarse-grained,’ in the sense that, though they agree that legal texts’ meanings or holdings are grounds of law, they disagree over what meanings or holdings are—and in particular, disagree over which more basic facts (semantic facts, contextual facts, legal actors’ intentions, etc.) ground meanings or holdings. Part of my project is to show that positivism, or at least one version of it, requires only this sort of coarse-grained agreement on the grounds of law. Legal officials must agree on which types of facts ground the content of the law, but that is consistent with them disagreeing over what makes any fact a token of those types. Notably, this coarse-grained agreement on the grounds of law suffices for legal officials to agree on what the content of the law directs in many cases.

The first error, in short, is to assume that positivism demands agreement on precise grounds of law when agreement at a more general level will do. Recognizing this error is important ground-clearing work but admittedly does not yet answer Dworkin’s objection. If legal officials agree that the grounds of law are as general as I just suggested, then they should also agree that the content of the law is indeterminate as to some cases. But as Dworkin observes, judges tend to speak as if the content of the law is determinate as to every case, suggesting that they understand the grounds of law to be more precise than positivists can allow. This leads to the second error that I wish to correct, which is to consider judges’ speech without regard to the context in which they are speaking, i.e., to ignore the pragmatics of judicial speech.

The pragmatics of judicial speech is an underexplored topic in legal philosophy, and I can only begin to address it here. Nearly everyone who writes on theoretical disagreement assumes that, if judges speak as if the content of the law is more determinate than they believe it to be, then they are behaving disingenuously. ⁷ But disingenuity has connotations of deceit or bad faith, and there is nothing of the kind in the judicial speech at issue. It is no secret that, when judges speak in their official capacities, they are fulfilling an institutional role that generally requires them to speak as if the content of the law determinately directs a court’s decision even when they believe the content of the law to be unsettled. Speaking in this way is not disingenuous; it is simply how lawyers, and possibly even many laypeople, expect judges to talk.

Given widely shared contextual beliefs, a reasonable hearer will often understand judges to be using their speech not to assert what the content of the law is but rather to advocate for what it should be—to be advancing lawyerly arguments for or against a court’s disposition of a legal issue. Below, Part I lays out the argument from theoretical disagreement and some recent discussion of it. Part II contends that legal officials, at least in the United States but presumably elsewhere as well, exhibit coarse-grained agreement on the grounds of law. Part III argues that this coarse-grained agreement is all that positivism requires. Finally, Part IV turns to judicial speech in hard cases and explains why that speech neither undermines agreement on the grounds of law nor requires attributing confusion or disingenuity to legal officials.

I. The Argument from Theoretical Disagreement

In this part, I lay out the argument from theoretical disagreement, first as Dworkin presents it, and then charitably reconstructed as an objection to the Hartian idea that the rules of recognition at the foundation of any legal system are social rules. I also discuss Leiter’s response to the argument and the obstacles that his response faces.

A. Dworkin’s Presentation of the Argument

Dworkin prefaces the argument from theoretical disagreement by introducing two terms: “propositions of law” and “grounds of law.” ⁸ Propositions of law are propositions about the content of the law, i.e., about what the law entitles, requires, permits, or empowers its subjects to do or have. For instance, ‘California law forbids driving over 55 miles per hour on a highway’ is a proposition of law. What Dworkin means by ‘grounds of law’ is somewhat less clear: he states that propositions of law are true or false in virtue of other, more “familiar” propositions, which “furnish” the grounds of law. ⁹ Although Dworkin was writing before talk of grounding became popular in metaphysics, I take it that he intends to invoke the sort of metaphysical dependence relation that metaphysicians today would call a ‘grounding’ relation. ¹⁰

We need not delve into contemporary controversies surrounding grounding for purposes of laying out Dworkin’s argument. I will assume—as Dworkin seems to assume—that social reality is hierarchically structured from more basic, or foundational, facts to increasingly derivative ones. The grounds of law are the more basic facts that ground the content of the law for some legal system. That is, the grounds of law are the facts in virtue of which the content of the law for some legal system obtains (and thus the facts in virtue of which propositions of law for that system are true or false). ¹¹ Everyone agrees that these facts include some descriptive facts about individuals’ or institutions’ actions and attitudes; the more contentious question is whether they also include evaluative facts about what is right, good, etc.

Dworkin claims that much of legal practice involves debating which propositions of law are true, and he identifies two forms such disagreement can take. ¹² In empirical disagreements, parties agree on the grounds of law “but disagree about whether those grounds are in fact satisfied in a particular case.” ¹³ For instance, parties might agree that the speed limit is 55 mph in California if there exists a California statute to that effect but disagree about whether there is in fact such a statute. ¹⁴ By contrast, in theoretical disagreements, parties disagree about what the grounds of law are. ¹⁵ For example, they might agree on “what the statute books and past judicial decisions have to say about compensation for fellow-servant injuries” but disagree over “whether the statute books and judicial decisions exhaust the pertinent grounds of law.” ¹⁶

The cases that Dworkin uses to illustrate theoretical disagreement are all cases in which judges appear to disagree over what it is about some statute or other legal text that grounds the content of the law. ¹⁷ In Riggs v. Palmer, ¹⁸ the New York Court of Appeals held that New York law prohibited a murderer from inheriting under his victim’s will. Judge Earl, writing for the majority, acknowledged that the relevant statutes would, if literally construed, direct that the murderer inherit. ¹⁹ But he reasoned that what the statutes literally said must yield to what the enacting legislature would have directed if “such a case had been present to their minds” and surely the enacting legislature would have directed that the murderer not inherit. ²⁰ Judge Gray dissented, arguing that the court should enforce what the statutes literally said. ²¹

Similarly, in TVA v. Hill, ²² the question that divided the U.S. Supreme Court was whether the Endangered Species Act (ESA) required enjoining operation of the Tellico Dam to prevent eradication of the snail darter—an endangered species of fish of little scientific or aesthetic interest. Chief Justice Burger, writing for the majority, held that the ESA’s “ordinary meaning” required halting the dam project, regardless of the attendant waste of public funds. ²³ Justice Powell argued in dissent that this conclusion was “absurd” and that the Court should adopt “a permissible construction [of the statute] that accords with some modicum of common sense and the public weal.” ²⁴ As Dworkin reads these cases, the judges disagreed over the grounds of law; that is, they were engaged in theoretical disagreements. ²⁵

Dworkin argues that cases like Riggs and TVA pose a problem for “the plain fact view of law”—a view that he attributes to H.L.A. Hart and other positivists. ²⁶ On this view, the grounds of law are matters of “plain historical fact,” over which no one can sensibly disagree, making theoretical disagreement perplexing at best. ²⁷ There are difficulties, however, with attributing the plain fact view to positivists. It is not entirely clear what the view holds. What does it mean for the grounds of law to be matters of ‘plain historical fact’? Nor is it clear that any positivist subscribes to it (Hart, for one, disclaimed holding such a view). ²⁸ In what follows, I will focus on a reconstructed version of Dworkin’s argument that more squarely objects to a central tenet of Hartian positivism, namely to the idea that rules of recognition are social rules. ²⁹

B. Reconstructing the Argument

Let us stipulate that a legal system’s rules of recognition supply the criteria of legal validity for that system, i.e., the criteria governing what counts as a legal norm in that system. Sophisticated legal systems presumably have numerous such rules. ³⁰ For example, one rule of recognition in the United States may be something like, ‘Any directive that the U.S. Constitution communicates is a legal norm.’ As this example illustrates, rules of recognition specify which facts—e.g., the fact that the U.S. Constitution communicates a certain directive—ground legal norms. There is thus a tight connection between rules of recognition and grounds of law: a legal system’s rules of recognition state what the grounds of law for that system are; they specify which types of facts legal officials must consider to identify the content of the law.

Many positivists, following Hart, maintain that rules of recognition are social rules. ³¹ It seems that a necessary condition for a social rule to exist is that members of the community whose rule it is must, by and large, subscribe to the rule, in the sense of largely conforming their behavior to the rule and accepting it as a standard of criticism. It is plausible (albeit not entirely uncontroversial) that a social rule cannot exist in a community where no such convergent pattern of conduct and attitudes obtains. This line of reasoning leads many positivists to conclude that, for a rule to be a rule of recognition, legal officials must generally treat it as such. More specifically, many positivists take themselves to be committed to the following constraint:

Convergence Constraint. A fact ϕ is a ground of law for some legal system only if the officials of that system converge on a rule of recognition that makes ϕ a ground of law.

By ‘converge on a rule of recognition,’ I mean that legal officials generally behave as though a rule is a rule of recognition and are inclined to criticize those who do not behave that way. What proportion of legal officials must so converge? No positivist would require total unanimity; yet a bare preponderance does not seem like enough. There is no bright-line rule: the convergence constraint is vague, and there will be borderline cases where there is no answer as to whether it is satisfied and thus no answer as to whether a fact is a ground of law. That said, a legal system cannot function without clear convergence on its central rules of recognition (those that identify many legal norms or very important ones), so we should expect borderline cases of convergence to arise only as to peripheral rules of recognition, at least in stable legal systems.

When would theoretical disagreement pose a problem for proponents of the convergence constraint? It seems that three conditions must obtain for any problem to arise. First, legal officials must disagree over whether a rule is a rule of recognition. (For simplicity, I will speak of legal officials generically and put aside questions of who those officials are or whether the behaviors and attitudes of non-officials are also relevant.) ³² Second, the disagreement must be widespread enough for there to be a clear failure of convergence. If only a few legal officials disagree, the result may be a borderline case of convergence, and a proponent of the convergence constraint can easily account for disagreement in such a case. Third, the disagreement must be authentic. It cannot be just a pretense to cover up some other form of disagreement.

Going forward, let us take ‘theoretical disagreements’ to refer to disagreements that meet these conditions, i.e., to widespread, authentic disagreements among legal officials over whether a rule is a rule of recognition. The problem for proponents of the convergence constraint is this: If legal officials fail to converge on whether a rule is a rule of recognition, and thus on whether a fact is a ground of law, the convergence constraint entails that the fact is not a ground of law. Proponents of the convergence constraint must therefore conclude that legal officials engaged in theoretical disagreement are confused: they are advancing facts as grounds of law that cannot be grounds of law because their very disagreement precludes those facts from being so. According to Dworkin, however, legal officials do engage in such disagreements and are not confused.

How much of a problem this is turns partly on how common theoretical disagreements are: while some confusion over peripheral rules of recognition that rarely make a difference in practice might not be surprising, confusion over central rules of recognition that impact many cases would be. ³³ So it is significant that Dworkin claims to offer examples of theoretical disagreement over central rules of recognition. As noted above, these are cases in which judges seem to disagree over what it is about a statute or other legal text that grounds the content of the law. How legal texts ground the content of the law matters to many cases—not just ‘hard’ ones in which legal officials disagree over what the content of the law directs in the case at hand, but also ‘easy’ ones in which legal officials agree but may agree for different reasons.

C. Leiter’s Response

There are several responses to the argument from theoretical disagreement. ³⁴ I will focus on just one influential response here. Leiter argues that there is little theoretical disagreement in practice and that whatever theoretical disagreement there is can be explained away by either of two strategies: the “Disingenuity” or “Error Theory” strategies. ³⁵ First, positivists can deny that apparent theoretical disagreements are ever authentic: when legal officials seem to be engaged in theoretical disagreements, they are really arguing about what the content of the law should be. ³⁶ Second, positivists can just accept that legal officials engaged in theoretical disagreements are confused. ³⁷ It would be a hard bullet to bite to conclude that legal officials are disingenuous or confused in many cases, so Leiter stresses that disagreement in legal practice is quite rare:

One may think of the universe of legal questions requiring judgment as a pyramid, with the very pinnacle of the structure captured by the judgments of the highest court of appeal … and the base represented by all those possible legal disputes that enter a lawyer’s office. This is, admittedly, a very strange-looking pyramid, as the ratio of the base to the pinnacle is something like a million to one. It is, of course, familiar that the main reason the legal system of a modern society does not collapse under the weight of disputes is precisely that most cases that are presented to lawyers never go any further than the lawyer’s office; that most cases that lawyers take do not result in formal litigation; that most cases that result in litigation settle by the end of discovery; that most cases that go to trial and verdict do not get appealed; and that most cases that get appealed do not get appealed to the highest court…. Why the preceding is true is familiar to anyone knowledgeable about law and litigation: there is massive and pervasive agreement about the law throughout the system. ³⁸

Leiter is surely right that lawyers and judges typically agree on what the content of the law directs in specific cases. But it is important to distinguish this sort of agreement from theoretical agreement on the grounds of law. Let us call agreement on what the content of the law directs in specific cases ‘legal agreement.’ And let us say that legal agreement is ‘broad’ if nearly all legal officials agree on what the content of the law directs in any one case, and that it is ‘pervasive’ if some officials agree on what the content of the law directs in nearly all cases arising in practice. A question for Leiter is whether the fact of broad and pervasive legal agreement—the fact that nearly all legal officials agree on what the content of the law directs in nearly all cases—can show what he wants it to show, namely that there is little theoretical disagreement.

Barbara Baum Levenbook and Dale Smith both push Leiter on this point. They note that, just because legal officials agree on what the content of the law directs in specific cases does not mean that legal officials agree on the grounds of law: legal agreement does not entail theoretical agreement. ³⁹ Legal officials may agree on what the content of the law directs in a given case but agree for different reasons. It is possible (albeit unlikely) that significant theoretical disagreement could underlie even broad and pervasive legal agreement. If so, then proponents of the convergence constraint would have to conclude that legal officials are pervasively confused and that easy cases are happy coincidences, in which officials’ different but equally mistaken beliefs about the grounds of law happen to lead to the same conclusion. ⁴⁰

Other commentators raise an independent objection: even granting that theoretical disagreement is limited to cases like Riggs and TVA, Leiter attributes more disingenuity or confusion to legal officials than we should accept. ⁴¹ The cases that Dworkin thinks involve theoretical disagreement may be proportionately rare relative to the number of disputes that enter a lawyer’s office, but there are still quite a few of them in absolute terms. These cases also tend to be highly salient, in the sense that they receive far more attention from legal officials and the general public than most other disputes do, making persistent disingenuity or confusion in such cases perplexing. A theory of law that does not require attributing disingenuity or confusion to legal officials in all such cases would be prima facie preferable to one that does.

I think that Leiter is right to emphasize broad and pervasive legal agreement, and my own response to Dworkin will emphasize such agreement as well. Where I believe that Leiter’s response falters is in conceding that cases like Riggs and TVA involve real or apparent theoretical disagreement at all. In what follows, I will argue that cases like Riggs and TVA do not evince theoretical disagreement; to the contrary, the judges in these cases clearly agree on the grounds of law and disagree only over how courts ought to proceed when those grounds fail to fully determine what the content of the law directs in the case at hand. So whereas Leiter allows that there is some evidence of theoretical disagreement and tries to explain that disagreement away, I will argue that there is no evidence of theoretical disagreement, at least in stable legal systems.

II. Coarse-Grained Agreement on the Grounds of Law

The upshot of the preceding part is that the argument from theoretical disagreement remains a problem for positivists committed to the convergence constraint. Before proceeding further, let me clarify the dialectical position that I take myself to be in. My goal in this paper is not to convince Dworkinians of the truth of positivism: there are many objections to positivism, and I cannot address them all here. My goal is more limited and defensive: it is just to show that positivists committed to the convergence constraint can answer this one objection without incurring the theoretic cost of attributing confusion or disingenuity to legal officials. I will thus be assuming, without argument, a broadly positivist picture of law and attempting to show how that picture can persuasively account for the cases that interest Dworkin.

My answer to the argument from theoretical disagreement has three steps. The first step is to show that legal officials do agree on the grounds of law, albeit at a higher level of generality than Dworkin and his supporters suppose. The second step is to contend that this sort of ‘coarse-grained’ agreement on the grounds of law is all that the convergence constraint requires. And the third step is to argue that legal officials do not wrongly suppose, or mislead others into supposing, that the grounds of law are any more precise than those I identify here. The present part of the paper concerns the first of these steps. I first draw a distinction between legal texts and legal content. I then argue that legal officials agree not only on what counts as a legal text but also—contra Dworkin—on what it is about any legal text that grounds the content of the law.

A. Legal Texts and Legal Content

We can distinguish at least two senses in which we use the word ‘law.’ We sometimes use it to refer to legal texts, like constitutions, statutes, regulations, or judicial opinions. Let us stipulate that these texts are all speech by institutions that (a) have authority to make legal norms, and (b) intend to exercise that authority. That is, ‘legal texts’ are speech acts whereby institutions with authority to make legal norms intend to exercise their authority to do so. Note that while these institutions are often collectives composed of many members (e.g., legislatures or courts), there is no mystery in attributing speech acts to them. What enables these institutions to speak with one voice is that they have voting procedures, which aggregate the expressed preferences of their members into an institutional decision to speak a text and thereby create legal norms. ⁴²

We also sometimes use ‘law’ to refer to what legal theorists call ‘the content of the law’ or ‘legal content.’ Legal content is what the law entitles, requires, permits, or empowers its subjects to do or have; it is the set of all legal norms in a legal system, the set of all legal rights, duties, permissions, and powers for that system. Whereas Title VII of the Civil Rights Act of 1964 is a legal text, any legal norm that Title VII grounds—e.g., a norm that employers shall not discriminate on the basis of an individual’s race, sex, etc.—is part of legal content. ⁴³ A crucial question is: What is the metaphysical relationship between legal texts and legal content? Or as I will sometimes put it: How do legal texts contribute to legal content? Positivists have not always focused on this question, and even today, it is not clear that they have fully answered it. ⁴⁴

Significantly, the mere fact that a speech act x is a legal text cannot by itself tell us what the content of the law is; we must also know what it is about x (what property of x) that grounds the content of the law. It is not enough for a legal system’s rules of recognition to include rules governing what counts as a legal text (call these ‘text-identification rules’). They must also include rules specifying how any legal text contributes to the content of the law (call these ‘text-contribution rules’). ⁴⁵ My goal in what follows is to show that legal officials converge on both sorts of rules. As convergence on text-identification rules is fairly uncontroversial, I address it only briefly before turning to convergence on text-contribution rules in greater depth. While I focus on the United States, I suspect that much of what I say is true of any stable legal system.

B. What Counts as a Legal Text?

Let me begin by forestalling a potential mistake: disagreement over the content of constitutional law is not disagreement over text-identifying rules of recognition. In Shelby County v. Holder, the U.S. Supreme Court divided over the constitutionality of the “coverage formula” in § 4(b) of the Voting Rights Act, which regulated whether state or local governments required federal preclearance for changes to their election procedures. ⁴⁶ A majority of the Court held that the coverage formula violated certain constitutional norms; a dissent disagreed. But all of the justices agreed that § 4(b) was part of a statute. The justices took for granted, as legal officials nearly always do, that the text at issue was a legal text—that it was spoken by an institution that has authority to make legal norms with the intention of exercising that authority.

Disputes like Shelby County may raise issues concerning how a constitution contributes to legal content, but they do not call into question whether a given text is a legal text and so do not evince a failure of convergence on text-identifying rules of recognition. This is not just an arbitrary consequence of our definition of legal texts; it also coheres with how legal officials generally treat legal texts in practice. Legal officials will usually enforce whatever norms a legal text creates regardless of constitutional defect until a court either enjoins enforcement of those norms or declares them to be unconstitutional. Moreover, even after a court declares a norm to be unconstitutional, the originating text will remain ‘on the books’—will continue to be recognized as a legal text—unless it is repealed by the enacting institution. ⁴⁷

For legal officials to actually disagree over the rules governing what counts as a legal text, they would have to disagree over the criteria that specify (a) which institutions have any authority to make legal norms (which institutions are legal institutions), or (b) which facts must obtain for those institutions to intend to exercise their authority (who must do what to generate the relevant institutional speech act). I will not attempt to articulate what those criteria are here (though the most central criteria clearly have to do with authority to make law being traceable to either the U.S. Constitution or a State Constitution). I will only note that there is no evidence of widespread disagreement in this regard in the United States today, given how extraordinarily rare it is for legal officials to disagree over what counts as a legal text. ⁴⁸

We can, of course, imagine widespread disagreement over text-identifying rules of recognition, and such disagreement may arise in politically unstable jurisdictions. For instance, there could be widespread disagreement over which of two constitutional documents creates institutions with authority to make law. ⁴⁹ But such disagreement does not challenge the convergence constraint. Rather, the convergence constraint helps to explain what is going on: the lack of convergence on a rule of recognition has caused a breakdown in the legal system, which should be obvious to everyone involved. In such a case, legal officials would understand that their disagreement is a political one over what the rules of recognition should be—the sort of disagreement that can only be resolved by political force or persuasion, not by legal argument.

C. How Does Any Legal Text Contribute to Legal Content?

I just suggested that there is hardly any disagreement over text-identifying rules of recognition in stable legal systems. Dworkin does not contest that conclusion. ⁵⁰ The rules that he believes to be subject to disagreement are rather text-contribution rules of recognition, i.e., rules governing how any legal text contributes to legal content. What I want to show next is that there is also hardly any disagreement over these rules. To that end, it will help to distinguish two kinds of legal texts. Some texts, like constitutions, statutes, and regulations, are spoken with the intent that the directives that they communicate directly become part of legal content. Call these ‘enacted texts.’ Other texts, like some judicial opinions, are spoken with the intent of justifying a decision, which is then taken to be precedential. Call these ‘precedential texts.’

Legal officials agree that an enacted text’s meaning and a precedential text’s holding is identical to what each contributes to legal content—to what each adds to or subtracts from the set of all legal norms. Put a bit more carefully, legal officials agree that (i) any directive that an enacted text communicates just is its contribution to legal content, and (ii) the holding of a precedential text just is its contribution to legal content. I take these to be platitudes, which are stated at such a high level of generality that no legal official would dispute them. What Dworkin and others overlook is that the convergence constraint does not require agreement on anything more precise than this. Since agreement on these two text-contribution rules of recognition is key to my answer to Dworkin, let me say a bit more about each.

1. Enacted Texts

My claim is roughly that an enacted text’s meaning is identical to its contribution to legal content. ⁵¹ Of course, ‘meaning’ is a notoriously ambiguous word that can refer, among other things, to what a text literally says, to all that it communicates, or to its upshot for a certain situation. ⁵² I will try to precisify a little what lawyers and judges mean by ‘means,’ but a central contention of this paper is that it is a mistake to pursue that project too far. We should not expect a philosophical level of precision in legal practice. Legal practice is less concerned with precision than it is with prospectively settling our rights, duties, etc., with respect to one another, and agreeing to use terms imprecisely can be conducive to that end. ⁵³ The key is that lawyers and judges agree on what ‘meaning’ means well enough to get by for the most part.

When lawyers and judges talk about an enacted text’s meaning, I take them to be referring to what it communicates. Or rather, I take them to be referring to the directive or directives that it communicates—to what I will call the text’s ‘imperatival communicative content.’ ⁵⁴ I make this minor clarification because the text’s non-imperatival communicative content obviously cannot be a legal norm. For example, the Americans with Disabilities Act communicates the legislative finding that “some 43,000,000 Americans have one or more physical or mental disabilities.” ⁵⁵ It would make no sense to say that this finding just is a legal norm. Legislative findings or preambles are evidence of the directives that a legislature intends to communicate, but it is only the directives themselves that are identical to legal norms.

One might cash out an enacted text’s imperatival communicative content in different ways. It could be understood in terms of the directives that the enacting institution intended to assert (along the lines of Gricean speaker meaning). Or it could be understood in terms of the directives that a reasonable hearer would infer that the enacting institution intended to assert. It could also be thought to comprise some or all of what the text implicates (e.g., by conversational implicature or presupposition). Accordingly, the facts that arguably ground an enacted text’s imperatival communicative content include at least: the speaker’s communicative intention, conventional meanings of the text’s words, rules of syntax, and contextual beliefs that a reasonable hearer would have (including beliefs about any applicable conversational norms).

Significantly, these facts sometimes underdetermine which directive an enacted text communicates. This may happen, for instance, due to a conflict between what the enacting institution most likely intended to assert and what a reasonable hearer would most likely infer that the institution intended to assert; due to unresolved ambiguity or vagueness in the text’s language; or due to uncertainty over the applicable conversational norms and thus over what the text implicates. While a more precise conception of imperatival communicative content would eliminate some of this underdeterminacy, there is little pressure for legal officials to converge on a more precise conception because this underdeterminacy only rarely produces disagreement over what the content of the law directs in the run of cases arising in practice. ⁵⁶

Most enacted texts are carefully drafted to provide clear direction in an anticipated range of cases and it is precisely those cases that tend to arise most often. The result is that, much of the time, anyone familiar with the usages of American legal English can read an enacted text and immediately and unreflectively see what the text directs in the case at hand without having to interpret the text, in the sense of choosing between multiple competing directives that the text could plausibly communicate. ⁵⁷ In these easy cases, legal officials agree not only on what legal content directs in the case at hand (legal agreement) but also on the grounds of law that make it so (theoretical agreement). In particular, they agree that the directive or directives that the relevant enacted text communicates just is its contribution to legal content.

When the facts that arguably ground an enacted text’s imperatival communicative content underdetermine which directive the text communicates, legal officials can sensibly disagree over what that directive is and thus over what legal content directs in some cases (legal disagreement). But even in these hard cases, legal officials still agree that the text’s imperatival communicative content just is its contribution to legal content (theoretical agreement). Although there is no single right answer to which directive the text communicates in these cases, there are still infinitely many wrong answers, and the text’s imperatival communicative content continues to mark the limits of appropriate legal argument. Any argument that runs counter to what the text could plausibly communicate cannot be an argument about what it contributes to legal content.

If an enacted text requires interpretation (in the sense defined above), legal officials can sensibly disagree over how to proceed. This brings us to debates over legal interpretation. These debates are sometimes offered as proof that theoretical disagreement is rampant, ⁵⁸ so it is crucial to explain why they show no such thing. Painting with broad strokes: textualists about statutory interpretation hold that courts should enforce what a reasonable hearer would be most likely to infer that the enacting institution intended to communicate by a statute; intentionalists hold that courts should enforce what the enacting institution most likely actually intended to communicate by a statute; and purposivists hold that courts should enforce what a reasonable, public-spirited institution would most likely have intended to communicate by a statute. ⁵⁹

There are many variations of these theories and other theories besides these. The point is that every mainstream theory of statutory interpretation offers some means of choosing among competing understandings of what a statute plausibly communicates. Oftentimes, this involves privileging some of the facts that arguably ground a statute’s communicative content over others. Thus, textualists would have courts focus on the conventional meanings of words, rules of syntax, and contextual beliefs that a reasonable hearer would have; intentionalists would have courts focus on the speaking institution’s communicative intention (and evidence thereof); and purposivists would have courts focus on what the text asserts or implicates in light of the conversational norm that legal institutions ought to speak reasonable, public-minded directives.

Theories of statutory interpretation are just that—theories of interpretation. They are theories about how courts should choose between multiple competing directives that an enacted text might plausibly communicate. For the most part, ⁶⁰ those who engage in debates over legal interpretation recognize that the debates turn not on metaphysical arguments about the grounds of law but on moral-political arguments about how courts should proceed when a legal text’s imperatival communicative content, and hence contribution to legal content, is nonobvious. ⁶¹ Thus, debates over legal interpretation do not evince a failure to agree on the grounds of law; to the contrary, these debates take the shape they do because legal officials agree that an enacted text’s imperatival communicative content just is its contribution to legal content.

2. Precedential Texts

Let me turn next to precedential texts, though I will have less to say about these because they figure less prominently in the literature on theoretical disagreement. Like enacted texts, precedential texts have communicative content. But unlike enacted texts, precedential texts’ communicative content is less straightforwardly connected to their contribution to legal content. Legal officials agree that what matters when it comes to precedential texts is their ‘holding’—roughly, the rule or principle explaining the speaking institution’s decision, ⁶² though even this formulation may be controversial. ⁶³ It is often harder to identify a precedential text’s holding than it is to identify an enacted text’s imperatival communicative content, and I would guess that the majority of legal disagreement stems from disagreement over precedential texts’ holdings.

Legal officials no more converge on a precise conception of a holding than they do on a precise conception of imperatival communicative content. ⁶⁴ Courts sometimes say that a holding consists only of whatever rules or principles are necessary to reach a decision but at other times say that a holding may include rules or principles that explain a decision without being strictly necessary. ⁶⁵ In addition, legal officials may disagree over how to weigh what the speaking institution actually said against what the institution should have said to best effectuate some policy or purpose underlying its decision. ⁶⁶ These, however, are at most disagreements over which sorts of facts ground a precedential text’s holding; they do not evince disagreement over whether a precedential text’s holding just is its contribution to legal content.

There is undoubtedly much complexity here. But we should not let that complexity cause us to lose sight of the fact that legal officials do converge on an articulable text-contribution rule for precedential texts. Moreover, it is at least sometimes possible for legal officials to read a precedential text and immediately see what its holding requires in another case, without having to consider and evaluate different candidate holdings. For instance, if a court concludes in a precedential opinion that a murderer cannot inherit under their victim’s will, then lower courts will ordinarily have no trouble ascertaining what legal content directs them to do in the next case involving someone who murders the testator of a will under which they stand to inherit. That is so even though some nonmaterial facts of the next case will, of course, be different.

III. Coarse-Grained Agreement and the Convergence Constraint

One might worry that the agreement identified above is illusory. Suppose that two judges consistently look to different facts to discern an enacted text’s imperatival communicative content and hence contribution to legal content: Judge Smith always looks to what a reasonable hearer would infer that the speaking institution intended to assert by the text, while Judge Brown always looks to what the speaking institution actually intended to assert by the text. To mark this disagreement, let us say that Judge Smith looks to what an enacted text ‘communicates*’ to identify an enacted text’s contribution to legal content and Judge Brown looks to what an enacted text ‘communicates†’ to do so. As it happens, however, what an enacted text communicates* rarely differs in any practically significant respect from what it communicates†.

In easy cases, these two judges will reach the same conclusion but for different reasons: Judge Smith will base decisions on what an enacted text communicates*, and Judge Brown will base decisions on what the text communicates†. And in hard cases, they will reach different conclusions for different reasons. Assuming that their disagreement is representative of a wider disagreement among officials of their legal system, it seems to follow from the convergence constraint that there is no fact of the matter about how enacted texts contribute to legal content. That brings us back to the conclusion that we were trying to avoid—namely, that legal officials are pervasively confused and that easy cases are mere coincidences in which officials’ equally mistaken beliefs about the grounds of law happen to lead to the same conclusion.

The problem with this objection is that it mischaracterizes matters to say that these two judges are following different text-contribution rules. They agree that an enacted text’s meaning just is its contribution to legal content. At the same time, however, their agreement is coarse-grained, in the sense that they disagree over which facts ground meaning. Recall that Dworkin assumes—I think correctly—that social reality is hierarchically structured from more basic facts to increasingly derivative ones. We can agree on how certain derivative facts relate to certain more basic facts without agreeing all the way down to the most basic facts. ⁶⁷ Legal officials can agree that legal content is grounded in enacted texts’ meanings without agreeing on which facts ground those meanings, and on which facts ground those facts, and so on.

Coarse-grained agreement on rules of recognition, and thus on grounds of law, is all that the convergence constraint requires. Remember that the idea animating the convergence constraint is that rules of recognition are social rules. For a rule to be a social rule, it should generally suffice that members of the relevant community converge (in the sense defined above) on articulable criteria that permit them to agree on most applications of the rule while structuring their disagreement over the remaining applications. Thus, for a rule to be a rule of recognition, it should generally suffice that legal officials converge on articulable criteria that allow them to agree on how the rule applies most of the time while structuring their disagreement over how it applies the rest of the time. Coarse-grained agreement on the grounds of law does just that.

Our imagined judges’ behavior suggests that they converge on the rule that an enacted text’s meaning (its imperatival communicative content) just is its contribution to legal content. While their behavior also suggests that they understand ‘meaning’ differently, the difference is marginal. It is not as though they disagree over which of a pair of homonyms (like ‘bank’ as in a riverbank versus ‘bank’ as in a financial institution) grounds legal content. If that were the case, then their agreement would be truly illusory. Instead, they disagree over which of two related senses of ‘meaning’—with mostly overlapping extensions—grounds legal content. So long as judges agree on what enacted texts mean most of the time, this coarse-grained agreement on the grounds of law can still account for broad and pervasive legal agreement.

Even in hard cases, Judge Smith and Judge Brown will take themselves to be arguing over the same thing: they will take themselves to be arguing over what an enacted text’s meaning is. Judge Smith may not be moved by arguments about what the text communicates†, just as Judge Brown may not be moved by arguments about what the text communicates*. But surely they will recognize that both kinds of argument are legal arguments because both concern how to conceive of an agreed-upon ground of law—how to conceive of an enacted text’s meaning. In this way, coarse-grained agreement on the grounds of law structures legal disagreement; it creates the framework within which such disagreement can occur. Much the same can be said of agreement on a precedential text’s holding being its contribution to legal content.

With these ideas in mind, let us return to Dworkin’s sample cases to see why they do not undermine coarse-grained agreement on the grounds of law. Justice Powell began his dissent in TVA by stating: “If it were clear from the language of the Act and its legislative history that Congress intended to authorize this result, this Court would be compelled to enforce it.” ⁶⁸ But “where the statutory language and legislative history … need not be construed to reach such a result,” it is “the duty of this Court to adopt a permissible construction that accords with some modicum of common sense and the public weal.” ⁶⁹ I take this to show that Justice Powell, no less than Chief Justice Burger, agreed that the ESA’s meaning was its contribution to legal content. Where the justices disagreed was over whether there was room to interpret the ESA.

Chief Justice Burger argued that the statute—which required federal agencies to “insure that actions authorized, funded, or carried out by them do not jeopardize the continued existence [of an endangered species]”—conveyed a clear directive that prohibited the TVA from operating the dam. ⁷⁰ Justice Powell countered that it was actually not clear whether ‘actions’ referred to all actions or only to prospective actions begun after the ESA became effective. ⁷¹ He contended that, given the absurd result that would follow from the former sense of ‘actions,’ Congress must have intended to employ the latter sense, with the result that the ESA did not prohibit operating the dam (construction on which began before the ESA became effective). ⁷² In short, he argued that the ESA required interpretation and that the Court should have interpreted differently.

This sort of argument based on absurdity comes up regularly, so we should consider it more closely. One conversational norm widely believed to govern legislative speech is that a legislature should not speak any directive that requires an absurd result. That norm has been adopted by courts and is thus a legal norm as well. ⁷³ Legal officials sometimes use this norm to select among multiple related senses that a word in a statute can bear (as Justice Powell did). ⁷⁴ They also sometimes rely on it to find that a statute conversationally implicates an exception to what would otherwise seem to be a categorical directive (as we will soon see that Judge Earl did in Riggs). Although legal officials often disagree over how this norm applies to particular cases, that is hardly surprising, since absurdity is a vague concept.

Justice Powell’s absurdity argument was not in opposition to the ESA’s clear meaning but rather in service of showing that the ESA’s meaning was not clear. His contention was that the Court should have leveraged underdeterminacy in the statute’s meaning, and hence contribution to legal content, to reach a decision that comports with “common sense and the public weal.” Chief Justice Burger, for his part, disagreed about the ESA being open to interpretation, stating that “Congress has spoken in the plainest of words.” ⁷⁵ The point for our purposes is that this disagreement over whether the ESA’s meaning was open to interpretation (or how to proceed to the extent that it was open to interpretation) is consistent with agreement that the ESA’s imperatival communicative content just is its contribution to legal content.

Turning to Riggs, we should clarify what Judge Earl was arguing. He took a kitchen-sink approach, contending that: (1) courts should enforce what the legislature would have directed if it had anticipated the case at hand and, here, the legislature would have directed that a murderer cannot inherit from their victim; ⁷⁶ (2) statutes should not be interpreted to direct absurd results and allowing a murderer to inherit would be absurd; ⁷⁷ (3) “maxims of the common law,” like “[n]o one shall be permitted … to take advantage of his own wrong,” prohibit a murderer from inheriting; ⁷⁸ and (4) whereas civil-law jurisdictions have statutes that expressly direct this outcome, common-law jurisdictions generally have no such statutes because their legislators presume that “the maxims of the common law” are already “sufficient to regulate such a case.” ⁷⁹

Importantly, the relevant New York statutes did not expressly state whether a murderer may or may not inherit from the victim. If they had done so, this would have been an easy case and would never have made it to New York’s highest court. But since they did not, there was some room to reasonably disagree over the statutes’ imperatival communicative content. We can understand Judge Earl’s first and second arguments as advocating that, given the legislature’s presumed communicative intention, or given the conversational norm that legislatures should not speak directives that would require an absurd result, the court should interpret the statutes to implicate an exception to what would otherwise seem to be a categorical rule, i.e., to implicate an exception for cases where, as here, it would be absurd for courts to enforce a will.

Similarly, we can understand Judge Earl’s third and fourth arguments as arguing that the legislature did not intend to communicate any directive that would displace established ‘maxims of the common law’ (i.e., legal norms grounded in caselaw or custom), one of which precluded the murderer from inheriting. By contrast, Judge Gray claimed that this case did not come within any implicated exception for absurdity because allowing the murderer to inherit was not (in his view) absurd. ⁸⁰ Moreover, he believed that the relevant statutes implicated that they occupied the field, thus displacing any contrary common-law norms regulating wills. ⁸¹ In short, neither TVA nor Riggs undermines legal officials’ convergence on the text-contribution rules discussed above; to the contrary, these rules help to explain legal disagreement in these cases.

IV. The Pragmatics of Judicial Speech

I have argued that legal officials, at least in the United States but presumably elsewhere too, exhibit coarse-grained agreement on the grounds of law and that such agreement satisfies the convergence constraint. This theoretical agreement accounts for not only legal agreement in the vast majority of cases but also legal disagreement in exceptional cases like Riggs and TVA. Still, we have yet not answered Dworkin’s argument. If legal officials only agree on the imprecise grounds of law just discussed, then on a positivist picture of law, it is practically inevitable that the grounds of law will not fully determine what legal content directs in some cases. Positivists take this conclusion to be a feature, not a bug, of their view because it seems intuitively right that there sometimes is no fact of the matter as to what legal content directs in a given case.

But then a puzzle arises: judges nearly always speak as if there is a fact of the matter; indeed, they speak this way even when it is obvious that legal texts’ meanings or holdings do not fully determine what legal content directs in the instant case. Does this speech imply that judges believe the grounds of law to be more precise than the convergence constraint can allow? If so, then it seems that proponents of the convergence constraint must conclude that judges are either mistaken or only pretending that the grounds of law are so precise. That is better than having to conclude that judges are wholly mistaken or disingenuous about the grounds of law, but it is still counterintuitive. Thus, notwithstanding everything that we said above, it seems that proponents of the convergence constraint might still be saddled with some theoretic cost.

This puzzle, however, depends on a wooden, literalistic understanding of judicial speech that ignores the context in which that speech occurs. We have to consider what judges are doing when they speak in their official capacities. Quite often, what they are doing is not stating what the content of the law already is (though parts of their opinions may do that) but rather making lawyerly arguments for or against a court’s disposition of a legal question. The issue here concerns the pragmatics of judicial speech, i.e., the information that judges convey in virtue of using linguistic expressions in a certain context rather than in virtue of the semantically encoded content of those expressions. This is a complex and underexplored topic in legal philosophy, and my discussion of it will be, admittedly, a bit rough and preliminary.

The context of judicial speech includes certain conversational norms grounded in our expectation of how judges ought to speak when they speak in their official capacities. Those norms may vary from time to time or place to place. We can imagine legal systems in which judges are expected to explicitly recognize in their opinions that legal content is indeterminate when they believe it to be so. But there are at least two reasons why that is not the expectation in the United States. The first concerns judicial opinions’ function. Such opinions give a court’s authoritative answer to a legal question: they state the ‘law of the case’ and may have a preclusive effect in future litigation involving the same parties. Moreover, if they are precedential, their answer to the legal question will also be binding in future cases involving different parties.

The way judicial opinions perform this function is by stating an answer to the legal question under consideration with the understanding that the very statement of the answer makes it correct, i.e., that the court’s declaration of what legal content directs in the case at hand makes it so, at least pending appeal or being subsequently overturned. Judicial opinions usually first lay out the facts of the case; then identify the relevant legal texts; then, to the extent that those texts require interpretation, advocate for one plausible interpretation; and finally, apply the chosen interpretation to the facts of the case. Once a court selects an interpretation to apply, it would be confusing and counterproductive to the authoritative resolution of the legal question to explicitly acknowledge that legal content did not determinately require that answer until the court said so.

The second reason is psychological and has to do with judges’ backgrounds. Judges (at least in the United States) are all lawyers, and they do not stop thinking like lawyers when they put on their robes. Lawyers know that it is rhetorically more effective to argue that the law requires a court to decide ϕ than it is to argue that the law permits the court to decide ϕ or ψ but that the court morally should decide ϕ. When lawyers argue before a court, they nearly always frame their arguments in the former way. Naturally, when those lawyers become judges, they continue to do the same. Judges, of course, have a duty of impartiality. But once they reach a decision impartially, it is no surprise that they advocate for their decision using much the same rhetoric that any lawyer would. That is what we expect judges to do.

In addition, there may also be a political reason why judicial opinions do not explicitly recognize legal indeterminacy. As Andrei Marmor puts it, the fact that judges must sometimes make new law is, though widely recognized, “an inconvenient truth” because it sits uneasily with a popular view of the division of governmental power. ⁸² Perhaps judges do not speak explicitly of legal indeterminacy in their opinions because they think it would be unseemly to do so. Still, it would be a mistake to conclude that judges therefore intend to deceive us about the extent of legal indeterminacy, for it is highly unlikely that they would be fooling anyone. Certainly, no lawyer would think that judges exercise zero discretion in hard cases just because judges say so; and I doubt that the general public would think so either. ⁸³

All of this bears on how we understand what judges are doing when they speak in their official capacities. By analogy, when you go to the theater, you do not expect the actors’ speech to reflect their own views about reality; you expect them to act—to speak as though they are characters in a play. When you go to a debate competition, you do not expect the debaters to express their own views about the issue they are debating; you expect them to compete—to defend a randomly assigned position with the goal of beating their opponents. Closer to home, when lawyers argue in court, you do not expect them to express their privately-held views about the content of the law; you expect them to advance lawyerly arguments—to argue their client’s side of the case as forcefully as the grounds of law permit.

More specifically, when the relevant legal texts are open to interpretation, we generally expect lawyers to speak as if the only plausible interpretation is the one that favors their clients, and as if the only relevant underlying facts are those supporting that interpretation. If lawyers were to regularly argue in court that the content of the law is indeterminate, we would likely think that they are not representing their clients zealously enough. Notably, however, our expectations are different when those same lawyers advise their clients confidentially. In the latter context, if the relevant legal texts are open to interpretation, then any responsible lawyer would advise the client that the content of the law is to some degree unsettled. In short, we expect lawyers to expressly acknowledge legal indeterminacy in some settings but not others.

Much the same is true of judicial speech. When judges speak for a court, we recognize that they are not speaking as they would in ordinary conversation but rather performing an institutional role. That role sometimes demands that they make lawyerly arguments for or against a court’s disposition of a legal question. We do not expect such arguments to track judges’ beliefs about legal indeterminacy, for in a different context, judges would likely speak differently. When conferencing amongst themselves or discussing a case with their clerks in chambers, they would presumably speak as any lawyer advising a client would—they would expressly acknowledge that the content of the law is sometimes indeterminate. Judicial speech thus neither undermines agreement on the grounds of law nor implies that judges are confused. ⁸⁴

Is this ultimately just an elaboration of Leiter’s ‘Disingenuity’ strategy? Perhaps, but then, the ‘Disingenuity’ label is inapt. ‘Disingenuity’ has connotations of deceit or bad faith. But there is nothing deceitful about the fact that, when judges make law, they speak as if the law that they are making had already been made. To call this disingenuous would be like saying it is disingenuous for an actor performing a play to speak as if they are the character. We expect judges to speak this way in their official capacities; we know that they are performing an institutional role in which it is considered common and appropriate to speak as if legal content determinately requires a decision even when it does not. ⁸⁵ Taking this context into account, judicial speech in hard cases does not provide even a prima facie reason to reject the convergence constraint.

Conclusion

My conclusion is fairly modest. I have not answered every objection to positivism, nor have I shown that it better explains legal practice than other theories. All that I hope to have shown is that Dworkin’s argument from theoretical disagreement is not fatal to positivism and, in particular, does not challenge the idea that rules of recognition are social rules. Positivists have a coherent story to tell that explains both legal agreement in easy cases and legal disagreement in hard ones. Legal officials agree on imprecise grounds of law that permit agreement on what legal content directs in most, albeit not all, cases. When these grounds underdetermine what legal content directs in a given case, judges recognize that legal content is unsettled. Still, we do not expect them to expressly acknowledge as much in their official capacities; we expect them instead to speak as any lawyer would, i.e., to advocate for an answer to a legal question.