The actions of legal institutions

Greenberg never defines “the actions of legal institutions.” He at most gives examples like legislatures enacting statutes or courts rendering decisions.Footnote ¹⁰ Perhaps he intends the phrase to denote every social fact that can generate legal content, but if so, the phrase is both under- and overinclusive. It is underinclusive because custom can generate legal content but is not an action of a legal institution. The British doctrine of parliamentary sovereignty, for instance, is part of the content of British law but does not arise from any action of a legal institution (unless we stretch “action” and “institution” beyond recognition to include longstanding acceptance of the doctrine by officials). The extension of the phrase “the actions of legal institutions” is thus too narrow to encompass every social fact that we commonly deem capable of generating legal content.

The phrase is also overinclusive. Legal institutions can take numerous actions. A legislature can draft a bill, debate it, vote on it, and enact it. The police can enforce a statute by ticketing or arresting violators. All of these—drafting, debating, voting, enacting, enforcing—are plausibly actions of legal institutions. But they do not all generate legal content. Mere preparatory actions like drafting, revising, and debating cannot generate legal content. These actions may, once a statute is enacted, be evidence of legal content. But they do not by themselves generate legal content.Footnote ¹¹ Other actions, like enforcing, also cannot generate legal content. Indeed, police can only enforce a statute if they already know roughly what the legal content generated by the statute requires.

Greenberg often uses statutory-interpretation cases to illustrate his theory, and it is clear that what most interests him is statutes and other legal texts like regulations and judicial opinions. We can clarify his theory—and more importantly for my purposes, sharpen his theory's contrast with the standard picture—by understanding it as an account of how legal texts determine legal content.Footnote ¹² By “legal texts,” I mean oral or written speech by an individual or institution (1) having authority to make legal content and (2) speaking with the intention of exercising that authority. A legal text, in other words, is a speech act by an individual or institution having authority to make law and intending to exercise that authority.Footnote ¹³ It is such speech acts that are the relevant “actions of legal institutions.” Thus, consider:

Creating the moral profile

Greenberg coins the term “moral profile” to refer to any moral norm—any moral obligation, power, permission, etc.Footnote ¹⁵ He gives several examples of how legal texts can “create” the moral profile.Footnote ¹⁶ Legal texts can establish a criminal justice system and thereby morally empower the state to use retributive force and morally disable private persons from using such force. Legal texts can precisify preexisting moral norms: a statutory-rape law, for instance, precisifies a preexisting moral obligation not to sexually abuse children by specifying the age of consent. Legal texts can also solve coordination problems and thereby morally obligate us to abide by that solution: a traffic regulation telling us which side of the road to drive on morally obligates us to drive on that side. And democratically enacted legal texts can morally obligate us to comply with what was democratically agreed upon.

Closer examination of these examples, however, reveals that Greenberg's use of the word “create” is misleading. Legal texts need not create moral norms out of whole cloth to generate legal content on his view. What his examples share in common is not so much legal texts creating moral norms as legal texts providing moral reasons to follow certain norms rather than others.Footnote ¹⁷ Thus, legal texts establishing a criminal justice system provide moral reasons for the state to have the power to retributively punish offenders and for private persons to have no such power. Legal texts precisifying a preexisting moral norm provide moral reasons to follow the precisified norm. Legal texts solving coordination problems provide moral reasons to follow norms supporting those texts’ solutions. And legal texts resulting from fair democratic processes provide moral reasons to follow what was democratically agreed upon.

In short, it seems that the way legal texts have a moral impact on Greenberg's view is by affecting our moral reasoning. When Greenberg says that a legal text creates part of the moral profile and thereby generates legal content, we should understand him as saying the text provides a moral reason to follow a norm and that norm just is legal content. If the text provides a moral reason to follow an obligation, the obligation is a legal obligation; if it provides a moral reason for an institution to have a power, the power is a legal power; if it provides a moral reason for someone to have a permission, the permission is a legal permission; and so forth. This suggests:

The second revised formulation raises an important issue: When a legal text provides a moral reason to follow a norm, does it matter whether the reason is pro tanto or all things considered? Greenberg “tentatively” suggests that “the relevant moral obligations” must be “all-things-considered, rather than pro tanto.”Footnote ¹⁸ I take this to mean that a legal text must provide an all-things-considered moral reason to follow a norm if it is to generate any legal content at all. But that tentative position is highly implausible. Suppose a city council enacts an ordinance stating “No vehicle shall park on a city street after a snowfall.” What is the ordinance's moral impact? Greenberg could say that the ordinance provides an all-things-considered moral reason not to park on a city street after a snowfall, but that would be absurd. In an emergency, there could be defeating reasons to park on a city street anyway.Footnote ¹⁹

Does that mean the ordinance fails to generate legal content? If so, most legal texts will fail to generate legal content because few, if any, provide all-things-considered moral reasons. That cannot be what Greenberg intends. He could instead say that the ordinance provides an all-things-considered moral reason to follow a loose obligation not to park on a city street after a snowfall, except in an unspecified range of cases where there is sufficient reason to park there anyway. But, then, we must know what someone should, all things considered, do in any situation before we can know what the ordinance prescribes in that situation. Determining what someone should, all things considered, do becomes prior to determining what the content of the law is. Surely, it is the other way around: we must know the content of the law before determining what we ought to do on the balance of reasons in situations where the law applies.

In sum, Greenberg's tentative position that a moral reason must be all things considered to generate legal content implies either that most legal texts do not generate legal content or that determining what we should, all things considered, do is prior to determining what the content of the law is. Neither result is tenable. As a friendly amendment to the theory, I propose that only a pro tanto moral reason is necessary to generate legal content:

The legally proper way

Greenberg says little about “the legally proper way,” but he apparently includes this restriction for just one purpose. Under certain circumstances, a legal text will provide a moral reason obligating us to undermine that text. Suppose a statute orders citizens to engage in a form of wrongful discrimination. The statute provides a moral reason for citizens to call for the statute's repeal or otherwise frustrate its purpose. But the statute's contribution to legal content cannot be an obligation to do those things; the statute's contribution to legal content cannot be an obligation to undermine the statute. Greenberg posits that, insofar as legal texts provide this sort of moral reason, they do not operate in “the legally proper way.”Footnote ²⁰ We can express the same idea more clearly by referring to non-undermining reasons, that is, reasons not manifestly at cross-purposes with the legal text:

The fourth revised formulation is, I think, as clear a statement of the moral impact theory as we can expect and one that is mostly faithful to what Greenberg intends. One might worry that I have merely substituted my own jargon for Greenberg's, but I believe we have made progress. In particular, we have clarified that: (1) the moral impact theory is best understood as addressing how legal texts determine legal content; (2) the way that legal texts determine legal content is by providing moral reasons to follow norms; (3) on the most defensible version of the theory, those moral reasons need only be pro tanto; and (4) those moral reasons cannot undermine—cannot be manifestly at cross-purposes with—the legal text.

The objection from gravely immoral legal content

A frequent objection to antipositivist theories of law is that they cannot account for gravely immoral legal content; they struggle to explain, for instance, how the Fugitive Slave Act—which directed state officials to cooperate in the return of escaped slavesFootnote ²⁷—generated legal content. As Greenberg presents his theory, legal texts must provide all-things-considered moral reasons to follow norms for those texts to generate legal content. Legal texts like the Fugitive Slave Act obviously provide no such reasons. Thus, the Fugitive Slave Act did not generate any legal content on Greenberg's view. He concedes as much, concluding that there are no “truly evil legal norms.”Footnote ²⁸

Yet it seems that the Fugitive Slave Act did create legal content—gravely immoral legal content, but legal content nonetheless. The judges and lawyers who lived under it certainly did not regard it as a legal nullity. Although Greenberg does not share the intuition that texts like the Fugitive Slave Act create legal content,Footnote ²⁹ he still owes us an explanation of why that intuition is so widely held. That is, he owes us an account of the widely held belief that not just gravely immoral legal texts but also gravely immoral legal norms can and do exist.Footnote ³⁰ While Greenberg notes that “[t]rue theories often have counterintuitive consequences,”Footnote ³¹ that does not absolve him of the need to explain why so many people have the intuitions that they do.

But what about the more plausible version of Greenberg's theory advanced above, on which legal texts need only provide pro tanto moral reasons? The objection from gravely immoral legal content presents less of a problem for this version of the theory.Footnote ³² One can argue that a text like the Fugitive Slave Act does not provide even a pro tanto moral reason to follow a norm. But it is difficult to distinguish between an extremely weak pro tanto moral reason that is easily defeated and the absence of any pro tanto moral reason whatsoever. I doubt that we have clear intuitions about this. Accordingly, while the objection from gravely immoral content is a strong objection to Greenberg's theory as he originally presents it, it is less forceful against the most plausible version of the theory requiring only pro tanto moral reasons.

The objection from the possibility of widespread legal error

Andrei Marmor offers the following thought experiment as an objection to Ronald Dworkin's theory of law as integrity.Footnote ³³ Suppose the US Supreme Court tries to resolve a question of legal interpretation in the manner prescribed by law as integrity. A majority of the justices might morally err: they might decide that the answer to the question is M, when the best constructive interpretation of the relevant legal and political decisions is not-M. After the Court's decision, practitioners would uniformly agree that the content of the law is determinately M. Nevertheless, law as integrity might still hold that the content of the law is determinately not-M. Although this may not be troubling as an isolated occurrence, we can imagine it reoccurring over and over again. Dworkin's theory thus suggests that it is possible for practitioners to be mistaken about much of the content of their law.

Mutatis mutandis, we can apply the same thought experiment to the moral impact theory. In resolving a contested question of legal interpretation, a majority of the Court might morally err and conclude that a legal text's moral impact is something other than it really is. Practitioners would then uniformly agree that the content of the law is determinately one thing, while the moral impact theory might still say it is determinately something else.Footnote ³⁴ This could happen over and over again. Thus, the moral impact theory also suggests that it is possible for a legal community to be mistaken about much of the content of its law—which is a hard pill to swallow. As Marmor writes, “Surely, at some point one would have to doubt whether a theory that renders a great part of the law to be a legal error is really a theory that tells us what the law is.”Footnote ³⁵

This has been, no doubt, too quick a discussion of these two familiar objections, but they are not my main focus here. The objection from gravely immoral legal content does not clearly threaten the most plausible version of the moral impact theory; moreover, we know that Greenberg is aware of the objection and finds it unconvincing. The objection from the possibility of widespread legal error is, I think, an effective objection against any version of the moral impact theory, and Greenberg has not, to my knowledge, addressed it. Yet because the objection is already a familiar one, I do not want to belabor it. Instead, I want to turn next to two objections that are more tailored to Greenberg's specific brand of antipositivism. These objections are significant not only because they show what Greenberg's theory fails to explain about legal practice but also because they highlight what the standard picture explains quite well.

The objection from legal reasoning

Greenberg acknowledges that there is a standard picture of how legal texts determine legal content and that the moral impact theory is not it.Footnote ³⁷ On the standard picture, to discover legal texts’ legal content, we must look to what those texts communicate. Not surprisingly, this is what we observe practitioners doing when they reason and argue about legal content. Practitioners often reason and argue about what the words in legal texts mean, what the lawmaker who promulgated those texts intended, or what the purpose of promulgating those texts was—all of which plausibly relates to what those texts communicate. By contrast, we hardly ever observe practitioners arguing about legal texts’ moral impact.Footnote ³⁸

To make the objection more precise, we can distinguish easy from hard cases. This distinction sometimes tracks whether existing sources of legal content determinately resolve a legal dispute, but that is not how I will use it here. Rather, let us say that easy cases are ones in which virtually any fair-minded practitioner with full knowledge of the relevant sources of legal content would agree on what the governing legal content is.Footnote ³⁹ All other cases are hard cases. Easy cases do arise; indeed, I argue below that most legal questions that come up in practice are easy cases. What we observe in easy cases is this: practitioners look to the relevant legal texts’ communicative content, and because what those texts communicate is sufficiently clear to resolve the case at hand, practitioners agree on what the governing legal content is. We do not observe practitioners arguing or reasoning about legal texts’ moral impact.

To be fair, practitioners sometimes mention, and are perhaps even motivated by, legal texts’ moral impact. A plaintiff's lawyer will sometimes frame her case to suggest that a legal text provided a moral reason for the defendant to be obligated to do such and such. But a competent lawyer would never rest her whole case on that framing alone. If a lawyer mentions a legal text's moral impact in an easy case, it is because she hopes it will motivate the court to look favorably on her other arguments pertaining to what the text communicates; it is not because she takes the text's moral impact to be dispositive of what the governing legal content is. Similarly, if a judge considers a legal text's moral impact in an easy case, it is likely because she wishes to show that she is aware of the moral stakes of her decision and, ideally, that her decision is morally right. It is not because she sees the moral impact as legal grounds for her decision.

Hard cases are different because, there, courts do commonly engage in moral reasoning. The usual positivist explanation of moral reasoning in hard cases is that courts are making new legal content. But the present objection to the moral impact theory does not even require that assumption because the moral reasoning courts use in hard cases is almost never moral-impact-style reasoning. The moral arguments that practitioners and courts make in hard cases usually concern the downstream consequences of deciding a case a certain way. They argue, for instance, that deciding a case a certain way would give the wrong incentives to future litigants, would increase administrative costs, would undermine the rule of law, etc. These are arguments about what the morally best resolution of a case is; they are not arguments about what the moral impact of legal texts is. Evidence of moral-impact-style reasoning in judicial opinions is rare.

In sum, we do not observe practitioners treating legal texts’ moral impact as grounds for deciding either easy or hard cases. This poses an especially pressing problem for the moral impact theory because the theory does not simply make morality a necessary condition for legality, as some natural law theories do. Rather, the theory purports to explain how certain social facts determine legal content. It thus goes to the heart of legal reasoning. If the theory is right, then practitioners are not merely mistaken about the legality of a few immoral norms, they are mistaken in how they practice law. They should be arguing about what the moral consequences of legal texts are, but that is not what we see them doing. This is at least prima facie reason to reject the moral impact theory; it puts the burden on Greenberg to explain why practitioners are pervasively mistaken in how they reason and argue about legal content.

Greenberg raises a couple of responses to this style of objection. One is that we should distrust how practitioners theorize about their practice: “[P]ractitioners are notoriously bad at theorizing their own practice. So we should give little weight to what practitioners say when they put on their philosophy hats.”Footnote ⁴⁰ Even if that is true, it misconstrues the objection. The issue is not how practitioners theorize about law but how they practice law; the objection is not that the moral impact theory fails to accord with practitioners’ jurisprudential theories but that it fails to accord with how they behave in practice. If Greenberg's theory were right, we would expect practitioners to reason and argue about legal texts’ moral impact, but we observe no such thing.

A second response points to substantive canons of statutory interpretation—like the rule of lenity—as evidence of courts deciding cases based on legal texts’ moral impact.Footnote ⁴¹ In Section III.B, I show how the standard picture better explains courts’ use of substantive canons. For now, it suffices to say that substantive canons are not evidence of courts looking to legal texts’ moral impact. First, courts apply substantive canons in only a fraction of cases; they more often decide cases without resort to any such canon. Second, the application of substantive canons is sharply limited by legal texts, which state how and when a given canon applies.Footnote ⁴² Courts applying substantive canons are not engaged in an unbounded inquiry into legal texts’ moral impact; they are rather applying those canons in the manner communicated by other legal texts (a point that I hope will become clearer when I return to it in Section III below).

The objection from pervasive agreement

It is commonly believed that a theory of law must be able to explain how practitioners can disagree over legal content.Footnote ⁴³ But it is equally, if not more, important that a theory of law be able to explain why practitioners agree on legal content as often as they do.Footnote ⁴⁴ In practice, we observe pervasive agreement over most legal content. Consider how the vast majority of legal disputes are resolved: most never make it to court; of those that do, most settle;Footnote ⁴⁵ of those that proceed past discovery, many end in judgment as a matter of law; and of the few that are appealed, most are affirmed by unanimous panels. Think also of all the issues that parties to litigation rarely dispute (e.g., parties often take personal jurisdiction and venue for granted). Some issues are virtually never disputed (e.g., that a judge was properly appointed). Such pervasive agreement calls out for an explanation.

The problem is that the moral impact theory cannot explain pervasive agreement over legal content. A legal text's moral impact at least arguably depends on a variety of factors, such as the fairness of the process by which the text was enacted, whether the text achieves an on-balance socially beneficial result, and how beneficial or oppressive that result is for particular parties. I take no stance on what factors actually determine a moral impact; I merely note that there is considerable room for disagreement as to both what those factors are and how they should be aggregated and weighed against one another. It is highly unlikely that, if the moral impact theory were right, we would observe anywhere near as much agreement over legal content as we do. Parties to a legal dispute have every incentive to disagree over legal content, and the moral impact theory suggests that there is much for them to reasonably disagree about.

To make the point more concrete, consider a garden-variety contract dispute with a statute-of-limitations issue. Suppose the Widget Corporation contracts to sell one hundred widgets to a buyer. The contract requires delivery by January 1, 2010, but the Widget Corporation fails to deliver by that date. The relevant statute of limitations is six years, but the buyer does not sue for breach until February 1, 2016. No extraordinary circumstance prevented him from suing earlier; he simply slept on his rights, as people sometimes do. This is an easy case: the trial judge will certainly grant summary judgment for the Widget Corporation (if the case even proceeds that far). What is more, easy disputes like this are common. If there is anything unusual about this particular dispute it is only that it made it into court—the buyer's lawyer should have told him to drop his claim and move on with his life.

The moral impact theory has trouble explaining why this is an easy case. The theory claims that the statute of limitations’ legal content is what its moral impact is. But what is the statute's moral impact? The Widget Corporation could argue that the democratic enactment of language referring to a bright-line rule, together with such a rule's ability to efficiently settle disputes, provides a moral reason to preclude people like the buyer from suing for breach more than six years after their claim accrues. The buyer could respond, however, that the statute provides a moral reason to follow a more discretionary norm—one that allows a court to take into consideration, say, the relative financial position of the parties and the ease with which the Widget Corporation can pay the buyer's damages. Which party has the better argument is at least not obvious. Even if the buyer's argument is weak, it is at least a colorable argument about what the statute's moral impact, and hence contribution to legal content, is. A court adhering to Greenberg's theory could not simply dismiss the buyer's argument out of hand.Footnote ⁴⁶

I purposefully chose a statute-of-limitations example because it illustrates how legal clarity is crucial to law achieving its coordination and settlement functions. Not only does the moral impact theory fail to explain pervasive agreement over most legal content, it fails to account for how law can coordinate action and settle disputes.Footnote ⁴⁷ For law to successfully perform these functions, there must be relative clarity as to what legal content is; people must be able to know what the law requires of them. But, as we just saw, the moral impact theory precludes precisely that sort of clarity: every bright-line rule we now agree upon could at least arguably be a vague standard on Greenberg's theory. This is once again prima facie reason to reject the theory; it puts the burden on Greenberg to explain why we so often agree on legal content when it follows from his theory that legal content is usually open to reasonable dispute.

Greenberg's strategy, particularly in recent work, is to minimize any difference between his theory and the standard picture. The moral impact theory, he says, is the “fundamental” account of how legal texts determine legal content, while the standard picture is the “surface” account.Footnote ⁴⁸ Practitioners can get away with behaving as if the standard picture is correct because the standard picture generally reaches the same result as the moral impact theory.Footnote ⁴⁹ I will reply to this point at the end of the paper, but we first need to get a better grasp of the standard picture.

The objection from indeterminacy

As we saw above, there are multiple ways of interpreting a legal text's communicative content; we might, for instance, look to its ordinary meaning, to what its author intended to say, or to what its author should have said in light of the text's purpose or the goal of maximizing social welfare. Sometimes (though, in practice, not very often), these different methods of interpretation yield multiple plausible interpretations of a legal text's communicative content. The trouble, according to Greenberg, is that when there are multiple plausible interpretations of a legal text's communicative content, the standard picture does not tell us which interpretation coincides with legal content.Footnote ⁶⁵ (The moral impact theory, he says, “supplies what is missing” by showing us why communicative content matters—it matters only insofar as it bears on a legal text's moral impact—thereby giving us a normative criterion by which to see which interpretation coincides with legal content.Footnote ⁶⁶)

Greenberg uses Smith to illustrate.Footnote ⁶⁷ Assuming that the majority and dissent had plausible readings of the statute's communicative content, the standard picture does not tell us which reading coincides with legal content. The standard picture does not answer whether the statute prescribes a sentencing enhancement for using a firearm in any way or only for using a firearm as a weapon. Greenberg concludes that the standard picture is therefore “badly underspecified.”Footnote ⁶⁸ Unlike Greenberg, I take the standard picture's neutrality in such cases to be a feature, not a bug. It is to the standard picture's credit that it does not try to resolve hard cases but instead explains why hard cases are hard. Smith was a hard case (after all, it divided the Supreme Court). It was hard because the statute's communicative content was indeterminate.

I take it that a desideratum of a metaphysical account of what determines legal content is that the account be consistent with any mainline prescriptive theory of legal interpretation. Such an account should not aim to reduce disagreement between textualists, intentionalists, and the like, but only to explain such disagreement where it occurs. The standard picture does just that: if a legal text's communicative content is indeterminate, its contribution to legal content is too, and there is room to reasonably disagree over how a court should apply the text. Someone who subscribes to the standard picture can still hold that, after communicative content runs out, courts should engage in moral reasoning (even moral-impact-style reasoning) to make new legal content. The standard picture only denies that such moral reasoning tells us anything about existing legal content.

The inconsistency objection

A legal text's communicative and legal content sometimes seem inconsistent. William Baude and Stephen Sachs imagine a criminal statute stating “Any person who sends live geese through the mails shall be fined under this title or imprisoned not more than two years, or both.”Footnote ⁶⁹ The statute seems to communicate that all and only those people who send geese through the mails are guilty of the offense. But we know this is not the whole story. Someone who sends geese through the mails but is insane or acting under duress is not guilty of the offense. Moreover, someone who does not send geese through the mails but attempts or conspires to do so is guilty of an inchoate offense. The statute's legal content is thus limited by defenses and augmented by liability for inchoate crimes that apply by default to all criminal offenses—none of which is reflected in the statute's communicative content.

Lawrence Solum raises a similar objection using the First Amendment. He writes: “[T]he content of free speech doctrine is legal content, but this doctrine far outruns the communicative content of the First Amendment.”Footnote ⁷⁰ More problematic still, free speech doctrine contradicts the amendment's communicative content. The amendment states that “Congress shall make no law . . . abridging the freedom of speech.”Footnote ⁷¹ Thus, its communicative content appears to concern only the legislative branch. Yet free speech doctrine limits the power not just of the legislative branch but of the entire federal government (and by incorporation via the Fourteenth Amendment, limits the power of entire state governments).Footnote ⁷² So it seems that legal content not only outruns the amendment's communicative content but also contradicts its communicative content.

The problem with this style of objection is that it ignores how legal texts interact. Legal texts hardly ever function in isolation. A statute and judicial opinions interpreting that statute commonly bear on the same legal norm. The statute may generate indeterminate legal content by itself, while the statute and judicial opinions together generate more determinate legal content. The same goes for constitutional provisions and judicial opinions interpreting those provisions or statutes and regulations promulgated pursuant to those statutes. For this reason, it is imprecise to speak, as we have been, of “a legal text's legal content.” Doing so too easily leads to the mistaken impression that legal texts must each give rise to distinct legal norms. We should instead speak of “a legal text's contribution to legal content”—meaning its contribution to the entire set of legal norms for a given legal system. This suggests:

The answer to Baude and Sachs's objection should by now be clear. They conflate a single legal text's contribution to legal content with legal content after all such contributions are taken into account. The communicative content of the statute they imagine is exactly what they think it is—a general prohibition on sending geese through the mails. On the standard picture, that communicative content just is what the statute contributes to legal content. The existence of defenses and inchoate offenses do not undermine that conclusion because the communicative content of other legal texts fixes those defenses and inchoate offenses.Footnote ⁷³ Multiple legal texts bear on the goose-mailing offense that Baude and Sachs imagine, and not surprisingly, we must look to what all the relevant legal texts communicate to know what legal content requires.

The response to Solum is similar. The First Amendment, like many constitutional provisions, is vaguely worded. Because its communicative content is indeterminate, its legal content is too.Footnote ⁷⁴ Most free speech doctrine arises not from the First Amendment itself but from judicial opinions interpreting the amendment.Footnote ⁷⁵ Now, this still does not answer Solum's main point, which is that free speech doctrine seems to contradict the amendment's communicative content. But the standard picture has resources to explain even this apparent contradiction. First, we sometimes understand an utterance to communicate something other than what its words ordinarily mean. We observe this in everyday conversation whenever we perceive that our interlocutor is misspeaking: our interlocutor says one thing, but we know he means another. It is possible that courts understand the First Amendment to be an instance of misspeaking: it says “Congress” but actually communicates a limit on the power of the entire federal government.

Second, and more importantly, even if we assume courts have misinterpreted the First Amendment's communicative content, the rule of recognition in the United States makes the Supreme Court the supreme interpreter of the Constitution. The Court's opinions are thus controlling in matters of constitutional law, and what they communicate, not what the Constitution communicates, ultimately determines the governing legal norm. The Court's opinions applying the First Amendment against the entire federal government thus explain why free speech doctrine is as it is: those opinions’ contribution to legal content supersedes the contribution of the First Amendment itself. The gap between free speech doctrine and the First Amendment's communicative content therefore does not undermine the standard picture.Footnote ⁷⁶

The objection from substantive canons

Greenberg also claims that the standard picture cannot account for common canons of interpretation, such as the rule of lenity (ambiguity in criminal statutes should be construed in favor of the accused), the absurdity doctrine (courts should not interpret a statute to yield absurd results), or the practice of reading a mens rea requirement into a criminal statute that does not expressly contain one.Footnote ⁷⁷ These canons, Greenberg says, do not aim to uncover legal texts’ communicative content but are instead “rules of thumb for working out the moral consequences of statutes.”Footnote ⁷⁸ He concludes that the standard picture therefore cannot explain how courts interpret legal texts in practice.

Although this objection is more complex than the prior one, it is related, and the response is similar. Some canons—like expressio unius est exclusion alterius, the expression of one thing implies the exclusion of others—are what some commentators call “semantic” canons.Footnote ⁷⁹ These are essentially Gricean maxims that courts use to work out pragmatically communicated content.Footnote ⁸⁰ Because semantic canons help decipher what a legal text's communicative content is, they pose no problem for the standard picture. The canons that Greenberg cites, however, are different. They are “substantive” canons that prescribe presumptions for or against deciding cases in certain ways. One response to Greenberg's objection is that, insofar as legislatures intentionally legislate against the backdrop of substantive canons or can be reasonably understood as doing so, those canons are relevant to legal texts’ communicative content. That is, perhaps substantive canons are context that partly determines communicative content.

But we need not rely on anything so empirically tenuous to refute this objection because substantive canons are themselves legal norms grounded in legal texts. Substantive canons are legal norms directing courts to take certain considerations into account when interpreting legal texts whose communicative content is indeterminate. When courts apply substantive canons to privilege one plausible interpretation of a legal text's communicative content over another, they are not searching for a legal text's moral impact—they are just following the law. On the standard picture, the communicative content of certain legal texts (mostly judicial opinions) determines the legal norms we call “substantive canons.” Those canons, in turn, direct courts to privilege certain interpretations of other legal texts, where it would otherwise be indeterminate what those texts’ communicative content, and hence contribution to legal content, is.

Dale Smith gives a more nuanced version of this same objection based on what he calls “retrospectively operating modifier laws.”Footnote ⁸¹ For instance, in the United Kingdom, §3(1) of the Human Rights Act 1998 provides that “primary legislation and subordinate legislation must be read and given effect in a way which is compatible with [the European Convention on Human Rights].” Significantly, this section controls courts’ interpretation of not only future legislation but also previously enacted legislation. Although §3(1) does not alter any legal text's communicative content, its effect may be to retrospectively modify some legal content. Suppose a statute was enacted prior to §3(1) with communicative content φ. The standard picture holds that the statute's contribution to legal content is also φ. After the passage of §3(1), the statute's communicative content remains φ, but legal content may have changed.

This version of the objection fails for much the same reason given above. Section 3(1) communicates that courts must interpret legal texts a certain way and so, on the standard picture, creates a legal norm requiring courts to interpret legal texts that way. That norm may require courts to revise their prior interpretation of some legal texts. Even though the express language of those texts has not changed, courts may be legally required to enforce a different plausible interpretation of the texts’ communicative content than they did previously. That is not to say that courts may now ignore communicative content. It is only to say that, where legal texts’ communicative content is indeterminate, courts must now privilege plausible interpretations of communicative content that are consistent with the Convention on Human Rights over those that are not. All this shows is that we must look to what all the relevant legal texts communicate to know what the content of the law is. None of this poses a problem for the standard picture.

The overgeneration objection

Another objection is that, clearly, not everything a legal text communicates is legal content.Footnote ⁸² For instance, statutes often include legislative findings. The Americans with Disabilities Act includes the legislative finding that “some 43,000,000 Americans have one or more physical or mental disabilities.”Footnote ⁸³ This proposition obviously is not legal content. The same problem occurs in judicial opinions, which often stretch to dozens of pages and begin with a detailed recitation of the facts of the case. No one thinks that everything a judicial opinion communicates is legal content. Rather, the received view is that a judicial opinion's contribution to legal content is the ground on which it was decided, its ratio decidendi. In these and similar ways, it seems that the standard picture overgenerates legal content.

Defining the standard picture in terms of communicative content—understood as all the information a legal text conveys in the context in which it is uttered—is too crude. Declaratory communicative content in a legal text may be evidence of what that text contributes to legal content. But declaratory communicative content does not contribute anything itself to legal content. What interests us is not everything a legal text communicates but rather just the text's imperatival communicative content—or as I will say, the directive that the text communicates. (Here, I intend the extension of “directive” to include not only commands but also the bestowal of powers, permissions, etc.) Properly understood, it is the directive that a legal text communicates that the standard picture identifies with the text's contribution to legal content:

This is what I take to be the true standard picture—the picture that best represents how practitioners standardly believe legal texts determine legal content. So where does that leave us? Assuming that the answer to the question of legal validity is something like a Hartian rule of recognition—which I have not argued for here—a legal system works roughly as follows. The rule of recognition, a complex social practice, picks out every legal text and other source of legal content for that legal system. Each legal text communicates some directive, which according to the standard picture, just is what the text contributes to legal content. Insofar as it is determinate what the directive is, the text's contribution to legal content is determinate; but insofar as it is indeterminate what the directive is, the text's contribution to legal content is indeterminate.

We must look to all relevant legal texts (and customs) to ascertain what the content of the law is. Sometimes, even though it is indeterminate what directive one legal texts communicates, a directive from another legal text (e.g., a substantive canon) resolves that indeterminacy and renders legal content determinate anyway. At other times, different legal texts’ directives conflict with one another, and unless the rule of recognition specifies that one directive supersedes the other, the result is that legal content as a whole is indeterminate.Footnote ⁸⁴ Much more needs to be said about what counts as a legal text and how various legal texts interact, but this at least gives an idea of how the standard picture fits into a broader positivist theory of law.

Recall Greenberg's contention that the moral impact theory is the “fundamental” explanation of how legal texts determine legal content while the standard picture is the “superficial” explanation. His claim is that, as it happens, the moral impact theory and the standard picture are roughly coterminous: a legal text's moral impact is usually the same as the directive it communicates, so both accounts pick out roughly the same norms as legal content. Practitioners can behave as though the standard picture is true because the standard picture happily coincides with the moral impact theory most of the time. The problem with this response should by now be clear: it is unmotivated. This sort of response makes sense only if objections to the standard picture succeed. If, as argued above, those objections are unfounded, then there is no reason to posit a more fundamental explanation behind the standard picture.

As a final caveat, I have been assuming that there is a single answer to how all legal texts determine legal content, notwithstanding the diversity of legal texts in most legal systems. The standard picture easily explains how certain legal texts like statutes, regulations, executive orders, and judicial orders determine legal content. But it struggles more to explain how judicial opinions determine the content of common law. Certainly, courts must communicate something to create common law, but they need not explicitly communicate any directive to do so. Perhaps all judicial opinions at least implicitly communicate some directive (based on the context in which they are rendered, which includes a practice of precedential reasoning). But this idea would have to be further fleshed out. None of this detracts, however, from the standard picture's explanation of how legal texts like statutes and regulations determine legal content.