2. The Case for Theoretical Disagreement

Since I shall argue that the U.S. statutory interpretation debate does not involve a theoretical disagreement, it is helpful to first explain why one might naturally suppose that it does.

According to Hartian positivism, there is, at the foundation of any legal system, a rule that specifies the conditions under which a particular norm (e.g., “No vehicles in the park”) is legally valid in that system. Hart dubbed this rule the “rule of recognition.”Footnote ¹⁶ A rule of recognition R exists in a particular legal system S in virtue of two facts: (1) the officials of S routinely apply the criteria of validity set forth in R when identifying valid legal norms; and (2) the officials of S regard these criteria of validity as ones that they and other officials ought to apply.Footnote ¹⁷ I will refer to the conditions of validity set forth in a system’s rule of recognition as that system’s ‘ultimate criteria of legal validity’.

Proponents of different schools of statutory interpretation clearly disagree, to some extent, about how to identify valid legal norms. Textualism and purposivism, the two most prominent schools of interpretation in the United States, prescribe different techniques for deriving valid legal norms from legislative enactments. Two conclusions might seem to follow from this fact. The first is that proponents of these views disagree about the conditions under which a norm is legally valid. Each maintains that a norm is valid when it is derivable from a statute by their own preferred method of interpretation, yet each subscribes to a different method. The second conclusion is that the U.S. rule of recognition must be silent about which interpretive method is correct. There is, after all, no consensus among U.S. officials about the correct method. A rule of recognition cannot, by definition, incorporate a criterion of validity about which officials pervasively disagree.

However, if one accepts that the U.S. rule of recognition is silent about which of these methods of interpretation is correct, then all of the difficult questions surveyed in the Introduction immediately follow. Why do textualists speak as if there is a fact of the matter about whether textualism is the correct method of statutory interpretation? Why do neither their colleagues nor other persons deeply acquainted with the legal system correct these mistaken assertions?

To avoid these difficult questions, a Hartian positivist might be tempted to resist the conclusion that U.S. officials are in fact engaged in a theoretical disagreement. But it is not at all obvious how to do so. Bill Watson makes a valiant but unsuccessful attempt.Footnote ¹⁸ He argues that, although textualists and purposivists disagree about many things, they all accept certain “platitudes” about statutory interpretation.Footnote ¹⁹ In particular, they all agree that a statutory text’s “meaning” is the appropriate target of interpretation.Footnote ²⁰ They simply disagree about what ‘meaning’ means.Footnote ²¹

Watson acknowledges the objection that such ‘agreement’ might not really be agreement at all, since there is no single proposition that all participants in the debate accept.Footnote ²² But he argues that the force of this objection is blunted by the fact that the word ‘meaning’ is polysemous, and that the meanings assigned to it by textualists and purposivists are “related.”Footnote ²³ Although proponents of these views

This reply does not, however, seem adequate to overcome the objection. According to Hartian positivism, the rule of recognition of a legal system S incorporates a criterion of validity c only if the officials of S: (1) routinely apply c in identifying valid norms; and (2) believe that they ought to do so. But on Watson’s proposal, there does not seem to be any single criterion of validity that all judges apply and accept: They merely use the same words to express their acceptance of different criteria.

It may be true that the word ‘meaning’ is polysemous, and that textualist and purposivist judges assign ‘related’ meanings to it. But ‘agreement’ of this kind generally does not suffice for the existence of a social rule.Footnote ²⁵ ‘Meaning’ may be polysemous, but so is ‘elk’, which can refer both to the living animal and to its meat. If, in a given community, half of the population thought it polite to bring elk (meat) to a barbeque, and the other half thought it polite to bring (living, breathing) elk, there would be no sense in which a single social rule governing politeness existed. Acceptance of the same sentence does not entail acceptance of the same rule, unless the parties assign the same meaning to that sentence. Polysemes are no exception to this general principle.

One might also take issue with Watson’s claim that the difference between a textualist’s and a purposivist’s understanding of ‘meaning’ is “marginal.” What Watson seems to mean is that the textualist’s assigned meaning and the purposivist’s assigned meaning have “mostly overlapping extensions.”Footnote ²⁶ But even if this is true, it does not follow that the meanings are themselves remotely similar. To borrow an example from Quine, “creature with a heart” and “creature with a kidney” also have mostly overlapping extensions.Footnote ²⁷ Nevertheless, they mean very different things. If, in a given community, half of the population thought it polite to doff their caps to creatures with hearts, and half thought it polite to doff their caps to creatures with kidneys, they might all doff their caps to the very same things. But as Hart observes, the existence of a social rule requires more than such a convergence of behavior.Footnote ²⁸ It also requires shared acceptance, from an internal point of view, of a standard characterizing that behavior as appropriate.Footnote ²⁹

Hart argues that a social rule exists in a community only when: (1) there is a shared “critical reflective attitude to certain patterns of behavior as a common standard”; and (2) this reflective attitude “display[s] itself in criticism (including self-criticism), demands for conformity, and in acknowledgement that such criticism and demands are justified.”Footnote ³⁰ It is difficult to see how the members of a community could satisfy these requirements without jointly accepting some proposition (or set thereof) describing how they ought to behave. If they accept differing propositions, they are apt to disagree about which “criticism and demands are justified.” Notably, a requirement of shared propositional content is common to many different philosophical analyses of social norms and conventions.Footnote ³¹

Watson argues that a social rule can exist in a group even if its members assign subtly different meanings to the words they use to express the rule.Footnote ³² He offers the following example. Suppose a group converges on the rule that “a war is just only if the harm it aims to avoid is proportional to the harm it causes.”Footnote ³³ Its members might still disagree about “how to compare the severity of harms.”Footnote ³⁴ I agree with Watson that such a situation is clearly possible. But I do not think it follows that a group can converge on a rule while disagreeing about the meanings of its constituent terms. In Watson’s example, group members may simply disagree about the extension of the word ‘proportional’.

Suppose, by analogy, that two people disagree about the best quick-and-dirty heuristic to use to determine which of two large, unwieldy pieces of furniture is heavier. The heuristics (e.g., visual size-up vs. partial lift) lead them to opposite conclusions about the right answer. Clearly, it does not follow from this disagreement that they disagree about the meaning of the word ‘heavier’. They may simply disagree about which ordered pairs of furnishings fall into the extension of ‘heavier’. Similarly, the individuals in Watson’s example may disagree not about the meaning of ‘proportional’, but just about the pairs of harms that fall within the extension of this word. Disagreeing about the right method for determining a word’s extension does not require disagreeing about its meaning.

Watson draws his proportionality example from Joseph Raz, who uses the example to illustrate what he calls the “relative independence of inter-related concepts.”Footnote ³⁵ Raz points out that people who share the very same “concept of a just war” can nevertheless disagree about “the truth conditions of propositions applying it,” owing to an underlying disagreement about the truth-conditions of propositions employing the concept of proportionality.Footnote ³⁶ This claim seems correct, but it does not follow from it that people can converge on a rule while disagreeing about the meanings of its constituent terms. The reason, in short, is that there is a difference between (i) disagreements about truth-conditions, and (ii) disagreements about meaning.

As Raz points out, truth-conditions can be characterized at different “levels of abstraction.”Footnote ³⁷ As an illustration, suppose that you and I understand the meaning of the descriptive name, “Jack the Ripper,” in the same way: It refers to whichever individual committed certain murders in the Whitechapel district of London in 1888 (i.e., the so-called “canonical five”). But suppose we disagree, for a variety of substantive reasons, about who committed these murders. I am convinced that it was Mr. Aaron Kominski; you believe it was Prince Albert Victor. Do we then agree or disagree about the ‘truth-conditions’ of the sentence, “Jack the Ripper was a prince”? The question is ambiguous, but on at least one ‘level of abstraction’, we disagree. On my view, it is true if and only if Mr. Kominski was a prince; on yours, if and only if Prince Albert Victor was a prince. This disagreement, however, is fully compatible with agreement about the meaning of the sentence.Footnote ³⁸ And notably, there are more abstract ways of characterizing the truth-conditions of this sentence that we would both endorse.

So too for proportionality and just wars. I agree with Watson that members of a community can converge on the rule that “a war is just only if the harm it aims to avoid is proportional to the harm it causes,” even while disagreeing—at some level of abstraction—about the truth-conditions of sentences involving the word ‘proportional’. But I do not think it follows from this that they can converge on this rule while disagreeing about the meaning of ‘proportional’. As the above example shows, a disagreement about truth-conditions (at one level of abstraction) can co-exist with agreement about meaning.Footnote ³⁹ If it became clear that there was no characterization of the truth-conditions of sentences involving the word ‘proportional’ that all members of a community accepted, then it would be clear that they understood its meaning differently. But in such a case, I would not be tempted to say that they had successfully converged on a social rule.Footnote ⁴⁰ As such, I do not think this example weighs very heavily against the commonplace view that shared acceptance of a sentence only suffices for the existence of a social rule when the members of a group assign the same meaning to that sentence.

Although I do not think that Watson’s explanation of why textualists and purposivists are not having a theoretical disagreement ultimately succeeds, I am in full agreement that this argumentative strategy is a promising one. The remainder of this paper will attempt to show that participants in the statutory interpretation debate may not, in fact, disagree about what the ultimate criteria of legal validity are. They may instead be engaged in what I call a ‘denotational’ disagreement. The next section will attempt to clarify this concept.

3. Denotational Disagreements

As observed above, from the premise that U.S. officials disagree about how to derive a valid legal norm from a statute, it is very natural to infer that they must also disagree about what the ultimate criteria of legal validity are. This section, however, will demonstrate that this natural inference is invalid. Officials who disagree about the correct method of statutory interpretation may nevertheless be in perfect agreement about the ultimate criteria of legal validity.

If the natural inference is invalid, then it becomes far less obvious why the U.S. statutory interpretation debate is a problem for Hartian positivism. If textualists and purposivists did disagree about the ultimate criteria of legal validity, then it would indeed be challenging for Hart to explain why they do not resolve their disagreement by ascertaining which criteria of validity most U.S. officials accept. But if a disagreement about the correct method of statutory interpretation does not entail a disagreement about the ultimate criteria of validity, then there is no obvious reason to think that textualists and purposivists are having a theoretical disagreement. Dworkin’s challenge to Hartian positivism is then significantly undermined.

To demonstrate why the natural inference is invalid, this section introduces two fictional examples of what I call ‘denotational’ disagreements. By a ‘denotational’ disagreement, I mean a disagreement in which the parties agree about what the ultimate criteria of legal validity are, but disagree about the denotation of a term appearing in a criterion that they all accept. To illustrate this notion, I begin with an example involving an empirical disagreement among officials. I then turn to an example involving a moral disagreement.

4. The U.S. Rule of Recognition

I have so far shown that what I called the ‘natural’ inference is logically invalid. One cannot validly infer—from the premise that the officials of a legal system disagree about the correct method of statutory interpretation—that they also disagree about the ultimate criteria of legal validity. This raises an important question about the U.S. statutory interpretation debate: Why should we think that it involves a theoretical disagreement? Might not U.S. officials, like the officials of Factlandia and Moralandia, be having a purely denotational disagreement? If they are, there is no reason to expect that they would try to resolve the dispute via official headcount, or that they would refrain from making obviously controversial assertions about what the correct method of statutory interpretation is.

The reader might, however, find it difficult to imagine how the U.S. statutory interpretation debate could be a purely denotational one. The goal of this section will be to illustrate this possibility more vividly. The proposal I advance will necessarily be sketchy, since without conducting a detailed empirical investigation of the attitudes and practices of U.S. officials, I could not hope to specify the exact content of the U.S. rule of recognition. My aim will be considerably more modest: I will merely argue that it is plausible that the U.S. statutory interpretation debate stems from a denotational disagreement.

Before proceeding, it will be useful to dwell for a moment on what it is that Factlandia and Moralandia have in common. In both nations, there is significant disagreement among officials about what the legally correct method of statutory interpretation is. But there is nevertheless widespread agreement about the desiderata a method of statutory interpretation must satisfy to be legally correct. In Factlandia, the sole desideratum is a historical one: It must be the method that the founder of Factlandia endorsed. In Moralandia, the desideratum is a normative one: The method must be morally correct.

An important lesson to be drawn from these examples is that a shared acceptance, among officials, of a set of ultimate criteria of legal validity need not manifest itself as a shared acceptance of any particular method of statutory interpretation. It may instead manifest itself as a shared acceptance of a desideratum, or set thereof, that a legally correct method must satisfy. This raises the question of whether there are any generally accepted premises among U.S. officials about the conditions that a method of statutory interpretation must satisfy to be legally correct.

In both Factlandia and Moralandia, officials manifest their shared acceptance of desiderata through common methods of argumentation. Factlandian officials, for example, attempt to vindicate their claims about the correct method of interpretation by advancing historical arguments about who the founder of Factlandia was. If there are shared assumptions among U.S. officials about the desiderata that a method must satisfy, it is natural to expect that they, too, would manifest themselves in common methods of argumentation. And indeed, there are striking similarities in how U.S. officials attempt to vindicate their assertions about which method of statutory interpretation is legally correct. Both when advocating for a particular method of interpretation and when critiquing alternatives, officials appeal—with great frequency—to the value of faithful agency.Footnote ⁴⁷

As numerous legal scholars have observed, it is a generally accepted premise among U.S. judges that, when interpreting a statute, courts “must act as Congress’s faithful agents.”Footnote ⁴⁸ Influential textualist John Manning has termed this view the “faithful agent theory.”Footnote ⁴⁹ On this view, the relationship between an enacting legislature and an interpreting court is a special kind of principal-agent relationship, in which the court, as fiduciary, must discern and execute the will of its multi-member principal, subject to the limits imposed by the U.S. Constitution.Footnote ⁵⁰ Obvious as this proposition might sound to American ears, it clearly does not express any kind of necessary truth about the legal duties of courts to legislatures. A legal system could, for example, delegate to courts the responsibility to construe statutes so as to make the legal system as morally just as possible, without regard for the enacting legislature. American judges would likely regard such an interpretive method as lawless; but this reflects only a contingent fact about the American legal system. The faithful agent theory is therefore a substantive one, in the sense that it expresses something about the American system that could have been otherwise.

While the faithful agent theory enjoys widespread support among U.S. officials, there is nevertheless significant disagreement about what exactly follows from the theory. As many scholars have observed, although textualists and purposivists both endorse the principle that courts engaged in statutory interpretation “must function as Congress’s faithful agents,” they sharply disagree about how exactly a faithful agent would proceed in interpreting a statute.Footnote ⁵¹ Textualists, who begin from the premise that a statute’s actual text is “the only reliable indication of congressional intent,” unsurprisingly conclude that a faithful agent will give effect to a statute’s text whenever its meaning is clear.Footnote ⁵² Purposivists, who reject the claim that the only intention of Congress that is knowable is the intention to enact a particular text, unsurprisingly draw the opposite inference.

The fact that textualists and purposivists agree about the desideratum that an interpretive method must satisfy, but disagree about which method satisfies that desideratum, raises the possibility that the U.S. statutory interpretation debate may not be so different from the Factlandian or Moralandian debates. In particular, it raises the possibility that U.S. officials may accept, from an internal point of view, the same criterion of legal validity, but disagree about the denotation of a concept embedded in that criterion.Footnote ⁵³

Here is a natural way of giving flesh to this hypothesis. U.S. officials accept, as one of their ultimate criteria of legal validity, the proposition that is semantically expressed by (3):

But textualists and purposivists strongly disagree about what follows from this proposition, since they have very different views about the denotation of the concept ‘the method of statutory interpretation a faithful agent of Congress would employ’. Each side thinks it denotes their own preferred method. The disagreement does not arise because the parties assign different meanings to the words “faithful agent”; both construe the “faithful agent” as an interpreter who gives effect to the will of the enacting Congress. The disputants simply have radically different views about which interpretive methods can reasonably be expected to have this effect. Consequently, they strongly disagree about which methods the faithful agent would employ.Footnote ⁵⁴

Understood in this way, the U.S. statutory interpretation debate has the same essential structure as its Factlandian and Moralandian counterparts. Officials accept, from an internal point of view, the same ultimate criterion of validity (i.e., (3)) but disagree about how that criterion applies to the facts. If the truth of (3) were itself the subject of official dispute, then by Hartian lights it would be perfectly sensible for officials to resolve the dispute via official headcount. But since, by hypothesis, officials are not disagreeing about the ultimate criteria of validity, they have little use for such a procedure.

In section 2, I examined Watson’s view that textualists and purposivists assign distinct but related interpretations to the word ‘meaning’, and I argued that agreement of this kind is insufficient to ground the existence of a social rule. It could reasonably be asked whether my own response to Dworkin is subject to the very same critique. After all, is a word’s denotation not part of its meaning? And even if it is not, how is it possible for two people to disagree about a word’s denotation without also disagreeing about its meaning? Does not meaning determine denotation?

It is certainly plausible that the meaning of some expressions is their denotation. A good example is so-called ‘directly referential’ terms, such as numerals. How could two people disagree about the denotation of the numeral ‘7’ without also disagreeing about its meaning? Plausibly they could not. Things stand differently, however, with other expressions—for example, many definite descriptions. Suppose we are in perfect agreement about the meaning of the description ‘the tallest building in the world’—i.e., both about the meanings of its constituent words and about how they compose to form the meaning of the whole. Would it follow that we also agreed about its denotation? No. You might (correctly) believe that it denotes the Burj Khalifa in Dubai, and I might (mistakenly) think that it denotes Merdeka 118 in Kuala Lumpur, for the simple reason that I was incorrectly told the Burj Khalifa burned down yesterday.Footnote ⁵⁵ Our disagreement would be not about meaning, but about the state of the world. It would therefore be implausible to identify the meaning of a definite description with its denotation. Agreement about the former does not entail agreement about the latter.Footnote ⁵⁶

Notably, on my account, the phrase that is the subject of the denotational disagreement among U.S. officials is a definite description—namely, “the method of statutory interpretation that a faithful agent of Congress would employ.” I will say more about the semantics of this description below.Footnote ⁵⁷ But given the behavior of other definite descriptions, it should not be surprising that two people could disagree about the denotation of this phrase without also disagreeing about its meaning. On my account, this is just the situation in which U.S. officials find themselves: They disagree about the denotation of a description whose meaning they understand in the same way.

* * *

In the previous section, I showed that a denotational disagreement among legal officials might take the form of either an empirical disagreement, as in Factlandia, or a moral disagreement, as in Moralandia. In light of this distinction, it is natural to wonder whether, on the view recommended here, the denotational disagreement among U.S. officials is Factlandian or Moralandian in character. I have argued that textualists and purposivists disagree about what interpretive methods a faithful agent of Congress would employ. But is this a disagreement about matters of empirical fact? Or is it rather a disagreement about morality?

The question must be framed carefully, since disagreements that are fundamentally factual can sometimes give rise to the appearance of fundamental moral disagreements. For example, two individuals might be in perfect agreement that an action is morally right if and only if it best promotes happiness. They might even be in perfect agreement about what happiness is. Nevertheless, they might still frequently disagree about whether various actions are morally correct, because of factual disagreements about which actions best promote happiness. Their disagreement would be fundamentally factual, and only derivatively moral.

Our question, then, is whether the disagreement among U.S. officials about which method of interpretation a faithful agent would employ is a fundamentally factual disagreement, a fundamentally moral disagreement, or both. Each of these possibilities seems worthy of further consideration. But it seems to me that the most obvious answer—and the answer that is consistent with the broadest range of positivist views—is that the disagreement is a fundamentally factual one.Footnote ⁵⁸

Here is one way of developing this idea. U.S. officials agree that the correct method of statutory interpretation is the method that a “faithful agent of Congress would employ.” Officials also agree that a “faithful” agent is one who reliably identifies and carries out the commands of the principal. But officials disagree about an important factual question: Which method of interpretation will enable reviewing courts to reliably do this? I will refer to this question, for short, as the question of which method of statutory interpretation is ‘reliable’. Call a method m ‘reliable’ just in case courts employing m will, across the long run of cases, correctly identify the shared communicative intentions of enacting majorities more often than courts employing any alternative method.

By a ‘communicative intention’, I mean an intention to express a message of some kind using a particular set of words. Communicative intentions can be very specific, such as when a speaker uses a sentence with the intention to express a particular proposition. But they can also be quite general and abstract. For example, a group of legislators might enact a statutory provision with the shared de dicto intention to convey ‘the message it expresses’—even while disagreeing about which message that is. Their shared de dicto intention would nevertheless count as a communicative intention, as I am using this term.Footnote ⁵⁹

To recapitulate, on the view I am proposing, the disagreement between textualists and purposivists is a disagreement about which method of interpretation is ‘reliable’. The question of whether a method is ‘reliable’, in my sense, is a purely factual one. To see this, note first that it is a factual question whether the congressional majority that enacted a statute shared a given communicative intention in so doing. It is also a factual question whether courts employing a particular method of interpretation are statistically more likely to correctly identify the communicative intentions of enacting majorities. But since these factual questions are quite difficult ones, U.S. officials unsurprisingly have not achieved consensus about their answers—i.e., about which method of statutory interpretation is ‘reliable’. Textualists maintain that a statute’s text is “the only reliable indication of congressional intent.” Purposivists disagree. Consequently, there persists significant denotational disagreement about which methods of interpretation a faithful agent would employ.Footnote ⁶⁰

It could be objected that my view’s emphasis on shared communicative intentions renders it questionably consistent with textualism. After all, is it not a commitment of textualists that enacting majorities frequently do not understand statutory texts in the very same way and frequently have differing understandings of their purposes? This commitment, however, is perfectly consistent with the view I have articulated. As pointed out above, in enacting a statute, legislators can share a de dicto communicative intention without even agreeing about what the text means—much less about what purposes it is intended to achieve.

It might also be thought that my view conflicts with textualism for another reason. According to textualism, it might be objected, legislative intent is not a determinant of the content of the law. It therefore should not feature in a criterion of legal validity. This objection, I think, rests on a more serious misunderstanding. Textualists acknowledge that legislative intent is a determinant of the content of the law. Consider, for example, the doctrine of the scrivener’s error, which both textualists and purposivists accept, and which Justice Scalia repeatedly endorsed.Footnote ⁶¹ The doctrine provides that, when it is absolutely clear that an enacting majority used a different word than the one it intended, courts should give effect to “the meaning genuinely intended but inadequately expressed” and “correc[t] [the] technical mistake.”Footnote ⁶² This doctrine clearly acknowledges the legal relevance of communicative intent. Lawrence Solum is therefore right that “textualists should not, do not, and cannot reject the role of communicative intentions in the production of statutory meaning.”Footnote ⁶³

If the view I have offered is correct, then what distinguishes textualists from purposivists is not that textualists regard communicative intent as legally irrelevant. Nor do textualists and purposivists disagree about the ultimate criteria of legal validity. Both accept (3), and both understand a “faithful agent of Congress” to be an agent who employs a reliable method of statutory interpretation. However, because they disagree about which method of interpretation is reliable, they also disagree about the denotation of the phrase, “the method of statutory interpretation that a faithful agent of Congress would employ.” Each thinks it denotes their own preferred method.Footnote ⁶⁴

For simplicity’s sake, I have so far written as though textualism and purposivism are the only plausible candidates for being “the method of statutory interpretation that a faithful agent of Congress would employ.” It is worth noting that this is not really the case. A judge could reasonably think that faithful agency counsels in favor of a more eclectic, context-sensitive method, which more closely resembles textualism in some cases and purposivism in others. Abbe Gluck and Richard Posner have shown that eclecticism of this sort is the actual practice of many appellate judges.Footnote ⁶⁵ The widespread use of such hybrid methodologies may indicate that, in the U.S. legal system, a commitment to faithful agency runs deeper than a commitment to any particular school of interpretation. On that supposition, it is unsurprising that judges would display a strong tendency toward eclecticism: Faithful agency might require the use of different tools in different cases.Footnote ⁶⁶

Although (3), by its terms, applies only to statutory interpretation, it might be wondered whether a similar ‘faithful agency’ criterion could be used to dissolve apparent theoretical disagreements in the realm of constitutional interpretation—e.g., the debate between originalism and living constitutionalism. I take no stance on that difficult question here.Footnote ⁶⁷ Importantly, I do not pretend to have shown that there are no theoretical disagreements in the U.S. legal system; I have merely tried to show that certain apparently theoretical disagreements within the realm of statutory interpretation may not be theoretical after all.