IS THE THEORY OF LAW EVALUATIVE?

SETTING THE STAGE FOR THE ARGUMENT

The distinction between law and the theory of law stands at the basis of some of the most influential recent defenses of legal positivism and is shared by many positivists who are otherwise in disagreement over much else. If we accept the prevalent view among positivists and what was perhaps Hart's implicit view, we will conclude that the theory of law necessarily requires reliance on evaluative considerations (for brevity I will use the expression “legal theory is evaluative” as shorthand for this claim) whereas knowing what the law is necessarily, or at least contingently, does not. I want to challenge the last claim and argue, to put it most succinctly, that the same kind of considerations that lead legal philosophers to believe that legal theory is evaluative should lead them to the conclusion that knowing what the law is requires evaluative considerations as well.

Before I begin to specify in greater detail the argument, let me say what it is not. I do not take a stand on the question whether evaluative considerations are part of the law or can be “incorporated” into the law. I do not consider this boundary question particularly helpful for understanding law, but whether I am right about this or not, it is not my concern here. It might then be thought that in that case I do not engage the legal positivists' arguments, because legal positivists are concerned exactly with this question. But this is a mistake: Andrei Marmor, for instance, argued that all legal positivists adhere to the thesis that “determining what the law is does not necessarily, or conceptually, depend on moral or other evaluative considerations.”⁹

Second, it is important to stress right at the beginning a point that will be further clarified as we go along: I do not intend to argue that laws cannot be immoral, or even that they are usually moral. It is a mistake to think that the only way of explaining the fact that laws are often immoral is by arguing that evaluative considerations are never required for knowing the law. As I will try to show, the fact that evaluative considerations (and even moral considerations) are required for knowing the law does not entail that what the law requires is always moral.

Another clarification is that my thesis is not particularly concerned with the distinction between “evaluative” and “descriptive.” These terms are vague and ambiguous, and again, my interest is not in labels. In one sense, to say that, for instance, English unjust enrichment law currently treats mistake of fact and mistake of law in the same way can be plausibly understood as a descriptive claim: it is a statement that conveys a piece of information that with certain qualifications is right now true. It is thus a “description” of part of English law. This by itself, however, does not undermine the claim that getting to know this may require taking evaluative considerations into account. The analogy with legal theory is helpful: suppose, for instance, that Hart's union of primary and secondary rules is indeed a fundamental truth about law. In that case, it is a “description” of the nature of law, but, as all positivists mentioned earlier contended, in order to know that, Hart had to rely on evaluative considerations of importance.

Finally, I present below a sketch of the argument that I will advance in the following sections. Even though it concludes that “legal positivism is false,” the term “legal positivism” should be treated as label for a particular thesis defined in premise (1) of that argument. I will show that this thesis is indeed held by many contemporary legal positivists, but I do not know whether it captures the “true” essence of legal positivism, whatever that might mean. If there is some account of law that deserves that label and is consistent with my argument, so be it. In fact, in the final sections of the paper I offer an outline of an account that in some respects is closer to contemporary legal positivism than to rival theories. The important question, in other words, is not whether my argument, if successful, refutes “legal positivism,” but rather its implications for understanding law.

We can now turn to my argument (which I will simply call the Argument in what follows). Here it is in skeletal form:

1. Legal positivism is the thesis that one does not need to rely on evaluative considerations in order to identify what the law is.
2. Judgments of “what the law is” are ambiguous: they are either about validity or about content. Therefore there are two possible versions of positivism, one concerned with legal validity, the other with legal content.
3. Legal positivism about validity is either an uninteresting thesis or one that turns out to be about what the law requires us to do in particular circumstances—what I propose to call legal content. (Besides, even if positivism about validity is an interesting thesis, some positivists have explicitly argued that their thesis is true about content as well, so the Argument may be successful with regard to them independently of my argument for premise [3].)
4. Determining the content of legal norms requires judgments of importance.
5. To quote Raz, “every judgment of importance is evaluative.”

Hence (6) Legal positivism is false.

The Argument is not exactly an analogical argument and has independent force regardless of the question of the need for evaluative considerations in the theory of law. However, I think it is particularly urgent for anyone who accepts that the theory of law necessarily requires evaluative considerations. In elaborating on premises (4) and (5), I will press it upon legal positivists to explain what it is about law that makes it so different from judgments about legal theory (or indeed about any theory).

As is clear from the presentation of the Argument, one important element of it is the distinction between validity and content, so I should clarify the difference between the two right at the outset. What does “knowing what the law is” mean? There are two ways of understanding this: legal positivists often assume that knowing what the law is means identifying valid laws. Laws are valid, according to this view, because they have certain features—call them “law-making properties”—and these properties are (or at least they may be) formal in the sense that they establish that certain laws are valid “independent[ly] of their content.”¹²

But what does it mean to identify valid legal norms? One way of understanding this is that we can identify the materials or sources that form the building blocks of legal norms. It is possible to understand what Hart says in the Postscript to The Concept of Law as the adoption of this view. He says there that Dworkin's theory needs “something very [much] like a rule of recognition identifying the authoritative sources of law.”¹⁴

In the first edition of The Concept of Law, Hart indeed usually used the term “validity” to refer to something stronger. He says there that “[w]herever … a rule of recognition is accepted, both private persons and officials are provided with authoritative criteria for identifying primary rules of obligation.”¹⁵

I shall deal later with this suggestion, but let us first examine a weaker thesis that has three components: (a) it is possible to distinguish between legal validity and legal content; (b) it is possible to distinguish between the identification of the legal materials from which we identify valid legal norms and the identification of valid legal norms, and as positivists we are interested in the latter; and (c) whereas knowing the content of legal norms requires (or may require) recourse to evaluative considerations, the identification of valid legal norms does not. The question, then, is whether we can find conceptual space for legal validity between identifying the building blocks of legal norms (identification of materials), and what the law requires in particular circumstances (knowledge of its content). One possibility is that norms are somehow self-identifying. Under this view, once we identify (by hypothesis, without reliance on evaluative considerations) the relevant legal materials, we can also identify valid legal norms—“immediately,” as it were. If this suggestion is true, the identification of valid legal norms will not be reducible to the identification of legal materials, but it will be something that is not similar to their content. But this suggestion will not do. Legal norms are not neatly demarcated within the legal materials. Consider, for example, the norm according to which for a contract to come into force in common law, the offer and acceptance must be accompanied by consideration. (I use the singular “norm” on purpose, because regardless of all the specific rules of consideration, there is a sense in which they all form one complex norm.) How does one come to know that this is a valid norm of, say, Australian law? To demarcate the doctrine of consideration from other aspects of the law we cannot just open the legal materials and find it there. Even if we find some summary of this norm, say, in a particular case, it is a summary that does not cover all situations that are nonetheless covered by a legal norm. Moreover, the judge who provided this summary had herself to look at myriad materials in order to extract from previous cases the norm that was found in them. A similar point is true of statute-based norms as well: to identify a valid legal norm that imposes a prohibition (a primary rule, in Hart's terminology) it is hardly ever enough simply to read one section of the law; to identify a single legal norm, one needs to read certain sections of the constitution, to understand certain general principles of the criminal law, to know the meaning of certain technical terms used in the section (something often found in other sections). All this requires, even if at a rudimentary level, a “theory of consideration,” a “theory of murder,” a “theory of larceny,” and so on.

In most cases found in modern jurisdictions, however, identifying the norms requires attention to both statutes and case law. And it is here that the need for “theory” is most clearly seen. As an example, consider the plain-sounding words: “Congress shall make no law … abridging the freedom of speech.” In order to know what these words “really” mean in American law, what the content of the legal norm is (as opposed to the legal text), one needs to have some “theory” of freedom of speech in American law. By saying “a theory of freedom of speech” I do not mean what academic lawyers usually mean by this term, namely a justificatory account as to why freedom of speech is important or which kinds of speech should be protected. The “theory” I have in mind need not be anything that grand. What is required, however, is a certain understanding of what counts as speech, what the conditions are that allow for certain limitation of it, and so on. It may be nothing more than a classificatory account, a “map” or a “flow chart” for dealing with cases involving speech. But to identify that requires considering what the law requires (how it applies) to particular circumstances, thereby breaking the boundary we wished to maintain between validity and content.

Perhaps, then, we could say that the identification of what the norm requires is distinct from the identification of the cases to which it applies. But this distinction cannot be sustained. One way of understanding this suggestion would be that to identify a valid legal norm does not require us to know its meaning. It is enough that we know that “contracts require consideration” in Australian law, even if we do not know what that sentence means. But surely to know the law (to know “what the law is”) is to know more than just words. Otherwise, how could I tell that the proposition whose content is “contracts require consideration” is true of Australian law but is not true of German law, whereas the proposition whose content is “contracts do not require consideration” is true of German law and not true of Australian law? I must know what contract means in German and Australian law, and what the requirements are for their coming into force. To know what the valid legal norm is, I must know what it prescribes, proscribes, empowers, and so on, and I cannot know that without knowing at least something about its application to particular instances.

More important, even if we could find some conceptual space for legal validity or for “knowing what the law is” that falls short of knowing what the law requires in particular circumstances, it is not clear why this conceptual possibility is of any theoretical interest. On this account, legal validity turns out to be a term of art that does not correspond to anything that laypeople who engage with the law ever come across or (if they did) would care about. They will never try to identify valid laws. What they will try to find is what the law requires, and this is something that by hypothesis validity will not tell them. Nor will they need to identify valid norms prior to and distinctly from identifying what the law requires. But then, if we believe with Raz that legal philosophy should aim to illuminate the important aspects of the practice, then legal positivism may succeed in maintaining the distinction between validity and content, but fail in a much more fundamental sense, because it sets its lights on a concept that is unimportant for understanding law.

In any case, this is not likely to be what legal positivists understand by legal validity, because when we consider what they say about legal validity, especially with regard to the relationship between validity and guidance, we see that what they mean by valid legal norms is something quite similar to what laypeople mean when they talk about knowing what the law requires in particular situations. Many positivists have stressed that guidance of conduct is one law's most fundamental features,¹⁸

One final suggestion is that we might be able to drive a wedge between knowing what a legal norm requires in the abstract and how we should act according to law in particular circumstances (which is distinct from how we should act all things considered, something that will be discussed below), because it may be that there are circumstances in which more than one legal norm applies to a case. Knowing how we should act in those circumstances will not be something that we could know just by identifying valid legal norms, and may even be something that requires taking evaluative considerations into account, but according to this interpretation, this goes beyond the purview of legal positivism. Not all positivists accept that conflict between valid legal norms is even logically possible; Hans Kelsen, for one, denied that in some of his writings. And although Hart accepted the possibility of conflict between norms, he specifically referred to this situation as something taken care of by the complex rule of recognition of modern legal systems (and thus as something for which we may not need evaluative considerations).¹⁹

Let us summarize this point: legal positivists have rested their case on the concept of validity but have done very little to explain what it is to identify valid legal norms that is distinct from identifying how the norm should be applied or what (if anything) it is to know “what the law is” that is less than knowing how the law requires us to act. As I have tried to show, once we look closely into this distinction, we see that it is very hard to show that the notion of legal validity can be an account of a concept that is illuminating about legal practice (through the notion of guidance) yet falls short of knowing what the law requires us to do in particular situations. As I have shown, at the very least, identifying “what the law is,” which legal positivists insist can be done without recourse to evaluative considerations, means being able to tell what the law prima facie requires. I suggest calling this the content of legal norms.

I believe that identifying anything other than that makes legal positivism at best a wholly uninteresting thesis. But even if I am wrong, and there is some interesting sense of validity that falls short of what I called content, many positivists have accepted that legal positivism is a thesis about content. To repeat earlier quotations, Hart legal positivism is a thesis about the identification of “primary rules of obligation,” that is to say, the identification of prescriptions as to how one should act according to law, and Raz and Green have been even more explicit in saying that it is possible to know the “content” of legal norms without recourse to evaluative considerations.

THE ARGUMENT ITSELF

With the stakes clearly on the table we can turn now to the Argument itself. The claim I want to make is that knowing the content of legal norms often, perhaps always, requires evaluative considerations. At this stage I do not distinguish, on purpose, between moral and non-moral evaluative considerations. As the quotations by Raz and Marmor above illustrate, at least some positivists deny that one is ever required to rely on any evaluative considerations in order to know what the law is, and this is the view I wish to challenge in this section. In the following section I will consider the weaker possibility, according to which only moral evaluative considerations are never required for knowing what the law is.

I will begin with a simple thought experiment. Imagine that you come to country whose laws you do not know and that you have reason to suspect are in some significant respects different from the laws of your own country of origin, even though you do not know which ones and in what way they are different. You now wish to know what that country's laws are. To be sure, there may be lots of different ways to get to know that, but if legal positivism is true, then it is possible to learn that by recourse to non-evaluative considerations alone, which is what you set yourself to do. There is, however, one important constraint: you cannot look into textbooks or treatises that tell you the law of that country. These materials might give an account of the content of law but they are not (or at least not necessarily) authoritative and, more important for my purposes, they “hide” the evaluative considerations that I want to bring out. Again, the analogy with legal theory is illuminating. Suppose I want to know what the important features of law are. By hypothesis this is something that requires evaluative considerations. Suppose now the account Hart expounded in The Concept of Law is true. In that case I may read the book and learn the nature of law supposedly without recourse to evaluative arguments. Have I shown that the claim that legal theory requires evaluation is false? Surely not, because—and I trust this was the point made by all the theorists—Hart himself had to rely on evaluative considerations in order to identify the important features of law.

Going back to the thought experiment, to make your project manageable you decide to limit yourself to the subject of accident law. You are allowed to look at typical legal materials (statutes, cases) and consult books about the rules of language in case you need them, but you are not allowed to consider any evaluative considerations. You start looking at the legal materials, and as is usual, they are a mess. The law reports, for instance, include lots of information: older and newer cases, facts about the accidents, facts about the trials, facts about what the parties have argued in court, citation of earlier cases and of academic literature, and all-too-common musings of the judges. Then there is also some legislation on this subject, but it deals only with some aspects of this problem and not with others. Slowly, you manage to make your way through these texts and get a clearer understanding of the legal norms that govern car accidents (which, it so happens, are very different from the ones you had known from your home country). What does this process of understanding include? It involves turning the mass of materials into a more orderly set of legal norms. My claim is that this requires the same kind of evaluative considerations of importance that are required from the legal theorist who looks at a legal system trying to find which aspects of it are important. Identifying the content of law requires distinguishing between important and unimportant cases, distinguishing between important and unimportant elements of the cases, understanding how disparate cases affect each other and how the legislation and the court decisions mesh together. One need not endorse any claim to global coherence of the law to admit that it is a fact that in order to know the accident law of a particular jurisdiction one usually has to know more than just its accident law, but learning that—all part of what we would normally consider knowing what that state's accident law is—requires more than just understanding what individual cases have decided. Taking seriously the ban on evaluative considerations, you would at best end up with a long list of facts and decisions of cases and some sections from statutes. (I say “at best” for reasons that will be clarified below.) The content of law, quite simply what the law is, would have remained obscure.

This is true not just when we try to establish a complete norm, even when all we want is to know what the “rule of case X v. Y” is. Every case involves many facts; every legal decision involves long passages. Distinguishing the important from the unimportant facts, distinguishing between pages of rhetoric and pages that affect what the law requires, does not seem all that different from the kind of task of the legal theorist. If positivists concede that evaluative considerations are required for legal theory, then the same considerations seem to be involved in knowing the content of law.

For those unimpressed with the analogy, let us continue with the thought experiment. Suppose that after beginning to learn the topic, you find it fascinating and decide to write a guide about the law of that country. Your aims are modest: no historical development, no philosophical justification, no proposals for reform, just the law. Yet even here, knowing the law requires more than just being able to state how particular past cases had been decided. In writing the book you'd need to think hard about the way statutes, cases, and customs blend together to give the legal norms of that country the content that they have. This is not just a matter of presentation; it is a matter of understanding what the law requires, of being able to apply it to individual situations. As anyone who has ever had to learn “just the law” from a textbook knows, probably the most significant difference between good and bad textbooks is that the latter give a summary ordered more or less chronologically of legal facts (certain events, court rulings, enacted statutes) but do not give a sense of what the legal norms are. This is because the former do not give an account of the legal norms but of certain legal facts. To identify and present legal norms requires deep understanding and synthesis of the disparate legal materials. Such synthesis requires judgments of importance, and such judgments of importance are evaluative.

IS THERE A WAY OUT?

Even if this argument is not a refutation of legal positivism, I believe it looks like something that legal positivists should worry about. But are there any escape routes? First, it might be argued that the argument is familiar, especially from the work of Dworkin, and as such has been adequately answered. But the Argument is different: one of Dworkin's arguments was that “[j]urisprudence is the general part of adjudication, silent prologue to any decision at law.”²⁰

Second, Dworkin argues for a necessary connection between legal truth and true morality. Thus, to take one example, his account of what the law requires in freedom of speech cases is connected with what he thinks morality—before or outside the law—requires. I do not make any such claim. As I stressed above (and as will be clarified further below), I use “theory” here in a sense that does not commit me to what justice, efficiency (or whatever) would require. It is nonetheless a “theory,” because it requires generalization and abstraction from particular cases and distinction between trivial and important aspects of decided cases to articulate what a legal norm requires. In other words, my point is that the reasons for thinking that knowing the law requires evaluative considerations are similar to the reasons for thinking legal theory is evaluative, not that the former is continuous with the latter.

Dworkin has made another argument, namely that people (and especially judges) have a duty to try consciously to make law as coherent as possible. He believes his account provides a good account of current legal practice, but even if it turns out that it is not, then Dworkin's more important claim is that it would be good if we adopted such a “conception” of law. This is a normative argument in a stronger sense than the one I put forward, and it includes claims like the impossibility of metaethics, the existence of interpretive concepts, and other heavy-duty arguments. Therefore, even if we reject Dworkin's claims (and for the record, I believe some of these claims are false), it does not follow that the Argument is false.

The positivist now has to face the challenge head-on. I will consider three ways for the positivist to answer the Argument: to show that there is a difference between law and the theory of law that undermines the analogy, to abandon the claim that legal theory is evaluative, or to abandon the claim that law is non-evaluative and yet maintain that this does not undermine legal positivism.

There are various arguments that may be understood as versions of the first strategy. John Gardner may have suggested one: he argued that legal positivism “tells us how the legal validity of any norm in any legal system falls to be determined … but leaves open whether and when and why any of us should ever bother to have or follow any valid legal norms.”²¹

Still, within the first strategy of undermining the analogy, it might be argued that the analogy with legal theory is flawed, because in order to know the content of law one need only look at historical facts, and in particular, facts about the pronouncements of certain bodies (legislature, courts, etc.). But this suggestion is unsatisfactory on two counts. First, if this is true, then why can we not just look at social facts in order to construct a correct theory of law? Why do we need to make evaluative considerations (in addition, perhaps, to observing the judgments of importance about the practice made by practitioners themselves)? Second, and regardless of the analogy, as I argued above, this will not suffice. The historical facts on their own do not order themselves into legal norms.

To see how broad the scope of this point is, we can consider the situation in which the highest court in a certain jurisdiction hands down a legal decision on a specific point of law upon which, let us assume, there was no settled law prior to that decision. Let us assume this case has been handed down in a legal system that treats the decisions of its highest court as authoritative precedents. In that case, then, under our assumptions, a new legal norm has been created by the case, and this norm will be decisive with regard to all future legally identical cases. But to be able to apply it to legally identical cases, we must first know the content of this newly created norm. Knowing this is essential for knowing which future cases will be identical to it in the relevant legal sense and which cases are not legally identical and therefore are not governed by the new norm. (For two cases to be two cases they have to differ at least in one of their spatio-temporal features—this follows from Leibniz's law—and all real-life “identical” cases are, of course, different in much more than that.) To determine that, the decision-maker will have to consider what the important aspects of the first case were and thus determine the rule of that case. Nothing in the first case itself will determine its own scope, and the judge in the second case could not rely on history to tell her what the decision in the first case was, for such history is nonexistent. Instead, in order to determine what the law is, the judge in the second case will have to rely on judgments of importance.

Suppose the case involved a traffic accident. The color of the cars involved in the accident may have been a very important consideration when they had been bought, but it is not usually important at all to questions of liability. But knowing that it is not important for liability requires identifying those features of the case that are important for identifying the applicable legal norm. And if we tried hard enough, we could come up with an example in which the color of the car would be relevant to questions of liability. To be sure, the expert judge may identify the important features of the first case immediately and with little need to consult casebooks or legal treatises, thus looking as though no evaluative considerations have been involved, but it would be a mistake to think that because of the speed of the conclusion no evaluative considerations had been involved. Even in this case, the judge is implicitly distilling the “important” aspects both from the facts of the case in front of her and from the previous case. Indeed, this may be exactly what makes this judge an expert. And even if later cases will not need to rely on evaluative considerations in order to know the legal norm determined in the first case (itself a highly questionable assumption), it is only because the judge in the second case has already done the evaluative work for them.

A different way of undermining the analogy would be to argue that making the sort of judgments described above indeed requires reliance on evaluative considerations, but that this is exactly is what agents do when there is no law on a particular question; when there is law, all that is required is the ability to read. Quite obviously the creation of new norms (by judges or by legislature) does require evaluative considerations, but given that all the cases described above are examples of norm creation, then legal positivism remains intact. This seems to me a rather desperate move: if we accept this, we must conclude that there is never pre-existing law that judges apply and that every decision involves the creation of new law. Such an answer conflicts with the understanding of most participants in legal practice and therefore is problematic if legal positivism aims to be an account of law that explicates and elaborates the understanding of legal practice as understood by participants themselves.²²

I turn now to the second route to answering the Argument. Although not specifically addressing the problem discussed in this paper, Brian Leiter has advanced an argument that may look like an adequate response to the Argument. He has argued that it is a mistake to think that the theory of law is evaluative; he concedes that it is a “banal truth” that all theorizing involves certain “epistemic values” like comprehensiveness, simplicity, consilience and so on and in fact that all understanding is “theory laden.” But from all this it does not follow that the resulting theory is not descriptive, because ultimately the theory of law is supposed to provide a “descriptively adequate”²³

Leiter also suggests that at the level of theory construction, one may need only to describe certain values that are part of the system without having to endorse them (thus adopting Hart's purported view described in Section I.B above). And the same idea might be tried at the level of the content of law: in order to say what the law is, we need only to be able to describe (without endorsing) the values of others. But, first, this answer is not available to the many positivists who denied this move and have argued that this is not the only sense of evaluation that the theorist has to undertake in order to give a successful account of the nature of law. Although they have argued that the theorist may describe without endorsing the moral considerations of participants in the practice, they have argued that the theorist has to make evaluative judgments of importance on her own. In other words, if this argument is successful, then it shows that legal theory is not necessarily evaluative as well. As such, this argument will prove too much as far as most positivists are concerned. Second, at least as far as judges or other participants in the practice of law are concerned, this solution is not available, because for them considerations of importance are constitutive of what the law is. Put differently, the Argument is not an epistemic argument.

The third solution, the opposite to the last one, is to bite the bullet and concede that knowing the content of law and the theory of law both require evaluative considerations. Can a positivist accept that? A positivist can perhaps do so, because we can distinguish between two kinds of evaluative considerations, moral and non-moral ones, and as long as only non-moral evaluative considerations go into the determination of the content of legal norms, the result is perhaps something that still captures the central tenet of legal positivism. Legal positivism, so this argument might go, emerged as an attempt to make clear the distinction between law and morality. Therefore the fact that determining the content of law requires non-moral evaluative considerations is no cause for concern for the positivist as long as it is not shown that moral evaluative considerations are required for knowing the content of law. Because showing that non-moral evaluative considerations are required for knowing the content of law does not entail that moral evaluative considerations are required for knowing the content of law, the Argument is not, by itself, something that legal positivists should worry about.

SOME MORE TROUBLE?

Until now the Argument sought to show that if one concedes that evaluative considerations are relevant for the theory of law, then they are relevant for knowing the law as well. We are now considering the response that we can distinguish between moral and non-moral evaluations and that legal positivism rejects only the claim that moral evaluative considerations are required for distinguishing between important and unimportant aspects of cases in order to have a “theory” of particular norms; put more simply, the response is that we do not need moral considerations in order to know the content of law. This is already a departure from the pronouncements of some legal positivists, who, as we have seen, denied that determining the content of law requires any evaluative considerations. Nonetheless, it is worth considering this possibility, because legal positivism is often associated with a thesis about the separation (or, as some prefer, separability) of law and moral considerations.

I will now take the conclusion of the Argument one step further and argue that if knowing the content of law requires evaluative considerations, then a similar argument suggests that it requires moral considerations as well. The reason is not that taking non-moral evaluative considerations necessarily forces one to take moral ones into account as well but that because of what law is, when one is engaged in the determination of law's content, this distinction cannot be sustained and one will have to rely on moral considerations as well. If this is true, the positivist faces the following dilemma: if what she argues is that people should rely on non-moral evaluative considerations but not on moral evaluative ones, then legal positivism loses its claim to be an accurate account of the nature of law as it is, for now it is clear that it is a normative thesis.²⁵

The first horn of the dilemma is, I think, quite plainly true, and I have no further comment on it. Therefore, if legal positivism is to be vindicated, this has to be done by blocking the second horn. Succinctly put, the claim here is that although it is true that there is nothing about theorizing in general that requires one to take moral considerations when making evaluations, matters are different when determining the content of legal norms. Legal norms, after all, are concerned with practical matters and at least try to provide agents with reasons for action. Moreover, the content of those reasons for action is often quite similar (unlike, say, the reasons for action provided by a do-it-yourself bookcase manual). It is therefore highly unlikely that moral considerations will not be part of the evaluative considerations that go into the determination of the content of legal norms. In terms of the earlier discussion, given the subject matter of law, the determination of which aspects of a state's legal history will be considered important will depend in part on moral considerations.

Can the positivist answer this charge? Let us first consider which answers will most obviously not do: to say that agents in fact do not base their determination of the content of law on moral considerations (despite the fact that they can) is an empirical claim, and defending it requires evidence. None, as far as I know, is available. In any case, legal positivism does not purport to be an empirical claim about how people happen to determine the content of law.

The second possibility is that in order to know the content of law, judges need not rely on moral considerations, even if in fact they occasionally do. If this argument is successful, it may be enough to maintain a conceptual separation between the content of legal norms and morality. There are two different possibilities here: one is that in order to know the content of law, agents never need to rely on evaluative considerations; the other is that in order to know the content of law, it is not the case that agents always need to consider moral evaluative considerations. But anyone who accepted the Argument up to this point is unlikely to adopt either of these views, for the Argument has established that evaluative considerations are part of what is required for determining and knowing the content of law. To say now that one must rely on non-moral evaluative considerations but need not rely on moral considerations suggests that the moral considerations are redundant or epiphenomenal, and I cannot think of a good argument leading to this conclusion. It will not do to give any reductive or naturalistic account of ethics. The fact that we can “translate” moral propositions to propositions that consist only of natural facts does not change the fact that those considerations affect people's reasoning, including the way they understand legal texts, and thus affect the determination of these texts' content. In fact, a naturalistic account of morality is more likely to suggest that moral considerations are not very different from other natural facts, and therefore the possibility that their causal powers are somehow so remarkably different from those of other facts seems even less plausible.

The third answer is that judges cannot base their determination of the content of law on moral considerations. But this view is also hard to accept. There is no logical or nomological impediment to their relying on moral considerations in determining the content of law. True, some arguments in favor of exclusive positivism seem at first to suggest some conceptual limit to the Argument, but I believe it is not hard to see that this is in fact not true, even if the arguments for exclusive positivism are sound (something that, as is well known, is a matter of some dispute, but on which I will remain largely agnostic here). The arguments for exclusive positivism have two conclusions: one is that morality cannot be “incorporated” into the law; and the other is that the criteria of legal validity are exclusively non-evaluative (or at least exclusively non-morally evaluative). As for the first conclusion, there is no conflict between the arguments, because the Argument does not purport to establish that morality is part of the law. The second conclusion may seem to be in greater tension, because the Argument shows that the content of legal norms is partly determined by evaluative considerations. But the moral considerations that one must consider according to the Argument are not the ultimate moral reasons for action that one ought to follow, so the supposed problem with inclusive criteria of legality, namely, that they fail to replace the reasons for action one has and thus stand in conflict with law's necessary claim to authority, does not arise. The moral considerations that a judge invokes in determining the content of law “color” the way in which an agent reads the requirements of a legal norm (in a way I hope to clarify below); they do not function as independent prescriptions for action and are not the reasons for action that an agent ought to follow regardless of the law.

Most of the other arguments in favor of exclusive positivism are a variation on Raz's theme and can be dealt with rather briefly here. Marmor has argued that the conventionality of law is inconsistent with inclusive positivism because “[m]oral and other practical reasons … cannot constitute reasons for acting according to reasons,”²⁶

I need not enter here into the details of Shapiro's argument, partly because by his own admission, his argument has a lot in common with Raz's argument for exclusive positivism.²⁸

If I have successfully covered all the relevant possibilities, we see that the claim that moral considerations are at least sometimes relevant for the determination of what the law is does not get a satisfactory answer. But this view may be given a more positive argument: to affirm the claim, all I need is one example in which moral considerations are relevant for determining what the law is. I will provide one such example, but if it is good, it can easily be multiplied indefinitely. In fact, I believe that on the basis of this example it can be shown that moral considerations figure quite extensively in the determination of the content of legal decisions.

Suppose we are asked to determine the rule or ratio decidendi of a certain case. The case deals with a civil claim based on negligent misrepresentation by an investment broker; suppose it was determined that in such cases the defendant has to compensate the plaintiff for the damages that she incurred. In determining that, we conclude that the fact that the negligent defendant was a broker is irrelevant (to say nothing about the company she was working in, the day of the week in which the event occurred, or the defendant's height) but the fact that she was a professional giving advice that she knew or was supposed to know was going to be relied upon is important. In order to make these determinations, we have to take (implicitly, in all likelihood) certain evaluative considerations and make some evaluative judgments, some of which may be labeled “epistemic” or “theoretical,” but also others that are squarely within the moral domain: the relevant considerations may include the meaning and importance of truthfulness and honesty to this case, or the importance of the free flow of correct information as a means for maintaining the smooth functioning of the market, which will affect our understanding of the legal case and the determination of the content of the legal norm that, let us assume, was established in the first case.

Suppose now that we are presented with a new case and are asked to determine whether the rule laid down in the first case is applicable to this case. At the first stage, the question is not whether the new rule should be extended or revised but simply whether it is applicable in the present instance. In order to know that, we need to know the content of the norm laid down in the first case. (Notice that even if we conclude that the case is inapplicable in the present situation, we can conclude that only after we have determined the scope of the norm laid down in it.) The new case involves negligent misrepresentation by a lawyer to his client about the planning rights in a certain piece of land that the client considered buying, and the question is whether the defendant should be ordered to pay compensation. There are many differences between the two cases, among them the time and space of the misrepresentation, the identity of the plaintiffs, and the profession of the defendants. Because the first case was the one that established the new rule, the subsequent rule is not grounded in a long history—there is only this case and other judicial decisions that by hypothesis are not directly applicable to the case. There are two possible interpretations of what the court does now: according to one (which I find less plausible as an account of existing legal practice in the legal systems with which I am familiar), all that matters is the decision whose scope the new court is now trying to determine, and the other cases are taken to be irrelevant. Let us now go back to the hypothetical example, and consider it more closely. Recall that one of the differences between the two cases was that in the first, the defendant was an investment broker, and in the second, a lawyer. Does this difference matter? The answer, of course, depends on judgments like the relative importance of these professions, the different deterrent effects that imposing liability (and its scope) may have on the two professions, and so on. It may well be that in the end we will conclude that none of these factors makes the slightest difference, but that is a moral judgment too.

Let us assume now an additional difference between the two cases, one that I have not mentioned until now: in the first case the defendant is a woman and in second case it is man. Let us imagine two scenarios: in one we decide that the rule laid down in the first case is applicable to the current case because we conclude that all the differences between the cases are legally irrelevant as none of them is part of the important features of the first case. (This conclusion may be wholly unconscious; it might be that we simply disregard all those factors.) In the other scenario, we believe that the moral capacities or cognitive capabilities of men and women are different (we may learned this from reading Aristotle, Kant, Nietzsche, or other famous moral philosophers who thought so), and as a result we believe that the different gender of the defendants in the two cases is a relevant factor that affects the scope of the norm in the first case. In the two scenarios our understanding of what the law is, as determined by what we think is the scope of the legal norm established in the first case, will be different, and the difference in the content of law will not be attributable to any “source.”

The other interpretation of the situation yields a similar conclusion but via a somewhat different route: here, again, the question is whether the court's decision in the first case is applicable to the new situation, and we assume that none of the other cases is directly applicable to the situation at hand. The difference is that this time we take the other cases to be of some relevance for determining the scope of the case at hand. The other cases explain or justify the legal decision in some weaker way. This may suggest that in such a case we could determine the scope of the norm in the first case by relying only on the information conveyed in those “sources.” But in fact, such examples pose an even greater difficulty to legal positivism, because determining what this weaker justificatory framework establishes is not something that is found in any of the earlier court decisions and it is not something that merely reading the facts and holdings of all past decisions will yield. Here, because the other cases are not applicable to the case, their (cumulative) effect on this particular case is not something that can be determined by mere non-evaluative examination of the history of earlier cases.

Two challenges are worth mentioning here. They have both been mentioned above, but I repeat them at this point in order to make my point clearer. First, it might be argued that these are not cases of rule-following but the creation of a new norm, and that as such, they cannot undermine legal positivism. This objection is unconvincing. To see why we need to distinguish between two possible ways of understanding this objection: one is that whenever evaluative considerations are taken into account and are not followed, then the judge creates a new norm. This, of course, will be victory by definition; instead of legal positivism telling us something interesting about law, on this interpretation legal positivism turns out to be a tautology. If, instead, the claim is that the examples suggested are not cases of rule-following but that there are other cases of rule-following, the positivist will have to explain what would count as an example of rule-following and what distinguishes such cases from the case discussed above. Furthermore, he would have to consider on what basis cases such as those above, which would be considered cases of rule-following by legal practitioners, are in fact cases of norm creation. It might be argued in response that practitioners may be misguided about the distinction between following and creating a norm, but this may be too much for legal positivism to hold. The problem with this response is that by putting out her finger to stop one hole from leaking, the positivist inevitably lets water leak from the other side, for if legal positivism is inconsistent with legal practice on such a fundamental matter, it loses its claim to be a descriptive account of what law is. In other words, such an admission might lead us to think not that practitioners are fundamentally misguided about their practice but that legal positivism is simply not a good account of that practice.

The second challenge is that it is not at all clear that moral evaluative considerations are required for knowing the content of law, for one may consider false moral considerations (like the one about the moral inferiority of women). But this is an unsuccessful objection for several reasons. First, as I have shown above, legal positivists have argued that evaluative considerations never figure in the determination of the content of law. So it is enough for me to show that in some cases (i.e., those in which it so happens that true moral considerations figure in the determination of the content of law) correct moral considerations have been required in order to determine the content of law in order to undermine the positivist claim. Because the Argument does not try to establish a necessary connection between law and morality, this objection is unsuccessful. But even this answer already concedes too much, and once again the analogy with the evaluative aspect of legal theory is helpful. We do not know if contemporary legal philosophers capture the important features of law. (At least by popular interest in their discussions, even among legal academics, it would seem that they have not!) That is to say, it may be that their judgments of importance are mistaken. And yet none of the philosophers who argued that legal theory necessarily involves evaluative considerations seemed troubled by this possibility. Third, and perhaps most important, moral considerations are taken into account even when their conclusion is that a certain distinction is not morally relevant for determining the scope of a legal norm. Thus, to disregard height, skin color, gender, and so on when determining the scope of a norm is to make the implicit moral judgment that these are irrelevant consideration for determining the content of a legal norm. If this is true, then in order for the objection to be correct, it must be that whenever judges rely on moral considerations in order to know the content of law, they always get morality wrong.

WHAT EMERGES?

I hope the arguments provided in the last section give reason for realizing that, at least sometimes, moral considerations will be taken into account in determining what the law is, not just what it should be. I would like to add some tentative suggestions about the way in which these conclusions may be relevant to understanding our practice of law and in particular to the notion of guidance, which is so central to it. I hope these remarks will further clarify the arguments of the last sections.

What lies beneath the Argument is a distinction between what may be called “normative background” and “normative foreground.” The normative foreground consists of normative considerations that are explicitly and consciously taken into account when deciding on a certain action. The normative background is the set of evaluative considerations through which we “see” the world. Those evaluative considerations are at play even when the law does not contain any reference to morality. Part of this set of evaluative considerations is a set of moral considerations. I must stress that the distinction between the normative background and the normative foreground is quite diffuse, and what for one person or at one time is part of one's normative foreground is at another part of the normative background. In practical matters these moral or practical considerations will affect the way we understand things we find in the world, including the way we understand the content of legal norms.

When people approach a legal question, they approach it with a certain background, which for much of the time is unconscious. Some of this background is occasionally brought to the fore in deep legal controversies. Some of this normative background is not strictly of the form of moral norms about everyday behavior, but it is normative background nonetheless. Consider the debates on constitutional and statutory interpretation: textualism, original intentions, the living Constitution, and the rest. These are all normative positions often backed by normative considerations about meaning, authorship, responsibility, utility, and so on. Controversy about them brings these normative considerations into the light, but they are there even when they are unchallenged and unnoticed. This example highlights two additional aspects of what I call normative background: that it consists of both moral and non-moral evaluative considerations, and that when people differ in their normative background it affects their understanding of what the law requires in a more-or-less global way. A devoted textualist will read a legal text in a somewhat different way from someone who gives special weight to authorial intentions, even if on many issues this will not affect the disposition of particular cases.

If we want to place these observations within familiar jurisprudential concepts, the best “location” for them seems to be the rule of recognition, although incorporating the required changes into this familiar concept might seem at first to burden it beyond what it could reasonably be expected to carry. As a matter of fact, I argue that the rule of recognition will have to undergo some significant changes, so significant that it might be misleading to use the term “rule of recognition” in this context. Yet I will try to explain below why, whether or not we retain the name, it would be unwise to abandon the insight we got from Hart's notion of the rule of recognition.

In order to accommodate the Argument, there are three significant changes that the notion of the rule of recognition will have to undergo. First, although not a mistake, it is perhaps misleading to think of the rule of recognition as a rule. One's normative background can be framed as rules, but it is a rule in a way the rules of English are for a native English speaker. One acts according to them without thinking,³⁰

A person's normative background is not the same as her rule of recognition, of course. The former is much broader and touches on many matters that have nothing to do with the law. But one change I propose to the rule of recognition is that it is best understood along the same lines, namely as an attitude about which set of materials is relevant for deciding legal questions, coupled with a set of attitudes regarding the way these materials should be understood. Understanding the rule of recognition in this way reveals a complex relationship between the rule of recognition and norms of the legal system. To begin with, it is a mistake to think of the foundational legal texts of a certain state (e.g., its constitution) as making up a state's rule of recognition, for those texts need to be recognized by a certain prior attitude by some members of society as the foundational legal texts of that state. No matter what a constitution proclaims itself to be, without the existence of a certain attitude that takes it to be the supreme law of a certain jurisdiction, it is just a piece of paper. However, the relationship is not unidirectional, because certain changes in the content of certain legal norms can bring about a change in people's attitudes toward legal texts (or toward what counts as legal texts). Thus a statutory or constitutional change can prescribe a certain method of interpretation and in this way can bring about change to a nation's rule of recognition. But such a change can be brought about through judicial obiter dicta or even the work of academics: an influential judge or scholar can change the rule of recognition if she manages to bring about change in the attitudes prevalent among judges (even those in another state) as to the right way to understand and interpret judicial texts. Yet in all such instances, the rule of recognition changes only after the attitudes toward the understanding of legal materials and the determination of the content of legal norms have changed.

The second change to Hart's rule of recognition, one already alluded to above, is that the rule of recognition is not merely what identifies certain legally relevant materials, and this means that the rule of recognition is much broader than Hart seems to have thought. Hart's account of the content of the rule of recognition is never very detailed and it does not include much more than the identification of those bodies with law-creating power, together with some rules of priority for resolving possible conflicts between different rules identified by the rule of recognition.³¹

But am I not confusing matters by “sucking” all these evaluative considerations into the rule of recognition? Labels are not important, so if one wishes to distinguish between the rule of recognition in the more limited Hartian sense and in the broader sense proposed here and restrict the term “rule of recognition” to the former, I will not quarrel with that. But it should be understood that an account of law cannot settle for the narrower conception of the rule of recognition, because the broader conception fulfills the same function that the narrower conception of the rule of recognition is supposed to have, namely, the determination of what the law is. Two societies with the same history and the same rule of recognition in the Hartian sense may still have different laws, that is, different normative legal requirements, if their members do not have the same normative background. (Of course, this depends on how broadly “the same history” is understood, for this will affect the scope of possible different normative backgrounds.) In fact, we see this vividly even within one legal system when judges who know the law and its history equally well still differ occasionally in their understanding of the requirements of the law. This sort of disagreement is much more common than the disagreement over the relevant legal materials, which is the sort of disagreement Hart's account could explain.

This change from Hart's account counts as yet another reason not to call the attitude toward the legal system a “rule,” because within a single legal system there is not one rule of recognition. Rather, there are many different rules of recognition. Holmes's bad man has a rule of recognition just as much as a justice of the Supreme Court does. The bad man, for example, might be less reluctant than a judge to consider prescriptions that appear in law books but are not enforced as non-laws. Even among “officials,” even all members of a state's highest court, it is a mistake to think they all follow the same rule of recognition. We can make sense of their recurring differences by taking into account their different normative backgrounds and how these affect their understanding of what the law requires. And it is exactly because the rule of recognition is general that it affects not just the interpretation of a single statute but one's understanding of what the law requires across the board. If Justice Scalia were to abandon his textualism, we would expect this to affect his opinions on many legal matters.

This is an advantage to the suggested account, because it gives a straightforward way of understanding what Dworkin called “theoretical disagreements” among judges. Quite possibly the explanation is even better than Dworkin's, because it can account for the fact that the differences among judges are quite often “patterned.” When an imaginary judge with infinite powers comes to decide a case, she can bring to every single decision the whole of the law. Real-life judges, however, never decide cases like that. To a great extent, they rely on the more specific history of the cases within one narrow field of the present case. And the best interpretation of past decisions in one area will not necessarily be related to a general normative account that is partly independent of the law. Hence we would expect the views of real-life judges to differ in a non-patterned way: a judge who tends to express “conservative” views in one case is not less likely to express “liberal” views in another. This does not happen often, and the account offered here can explain why.

I want to consider now several possible objections to this suggested account. In the first place, I wish to clarify that I do not argue that different people with different normative backgrounds cannot understand each other, which is plainly false. It is a mistake to think that one's normative background is wholly impenetrable to oneself or to others. In fact, much of the work in philosophy can be seen as attempts to examine and explore people's—or at least the philosopher's own—normative background. The work of some reflective judges is in the same vein.

But I might be challenged that although this account allows for mutual understanding, it does not (contrary to Dworkin's account) allow for genuine disagreements between judges, because it turns out that if judges have different rules of recognition, they are not really talking about the same thing. Instead of disagreeing with each other, it turns out that people talk past each other. This challenge is true in the sense that it shows that some disagreements among lawyers are not “genuine” in the Dworkinian sense; however, they are genuine in any other sense, but the account is no worse for that, because we have no reason to adopt Dworkin's view on what makes disagreement genuine. Dworkin presupposes that participants in legal debates talk about the same thing and then offers his rather idiosyncratic account of interpretive concepts to explain that. But we need not presuppose that with him. In fact, what we would normally call genuine disagreements have one feature that is conspicuously absent from many of the disagreements that caught Dworkin's attention; namely, they tend to be resolved. They are resolved because after the debate, one side discovers the error of his thinking (a false premise, a conclusion that does not follow from the premises, statistically insignificant evidence used as a basis for a certain inductive conclusion, premises that beg the question, etc.). Even enduring debates in natural science (for example between the gradualists and punctualists in evolutionary theory) remain open simply because we currently lack the evidence to decide one way or the other and the current available evidence can (arguably) reasonably be interpreted in more than one way. In contrast, disagreement that is not resolved over a long period of time—indeed, one that participants enter into believing they are not likely to succeed in changing the other side's views³²

This conclusion might seem strange at first: If this is the case, should judges not have realized this already and ended the debate? There are at least two ways of explaining the persistence of their disagreement. One is the prosaic but nonetheless genuine reason that the judges are in a job that “forces” them to go on with their debate. I doubt if Justices Brennan and Scalia really thought they had any chance of convincing the other or believed that they shared all the premises of their arguments leading to their conclusions. Yet they kept on writing their bitterly conflicting decisions because this was part of their job. The second reason is more significant: judges may genuinely disagree on what would count as a good state of affairs for the country to be in (for instance, with regard to economic equality), and this will be part of their normative background that will affect their understanding of what the law says on certain matters. In this sense, indeed, their disagreement is genuine, for what the judges try to do is convince each to change his or her rule of recognition. True, it follows that the decision in particular cases then turns out not to be genuine, but as an account of judicial disputes, the suggested account seems more accurate than the alternative: a judge arguing on the merits of a different theory of constitutional interpretation may concede that, given the other judge's rule of recognition, a certain conclusion in a particular legal case would follow. The disagreeing judge has no point in trying to convince the other to reach her conclusion from the other judge's starting point. Instead the effort is to convince the other that that starting point is wrongheaded.

The next objection is that in that case it is impossible to argue that there are many different rules of recognition in existence in a single legal system without losing much of the explanatory power of Hart's original idea, for Hart's theory is premised on the existence of a unified rule of recognition. Law cannot provide an authoritative guide to conduct if different people have competing rules of recognition; and without guidance, law cannot fulfill what is perhaps its only discernable general function. This function seems so fundamental to law that when a legal system loses its ability to guide conduct according to its normative requirements, in an important sense we no longer have law. Call this feature “guidability.” The claim now is that the suggested account not only does not explain law's guidability; it actually renders guidability impossible.

I believe this objection is mistaken, but it is an important one nonetheless, because it brings out what is at once an obvious observation about existing legal practice and something that is difficult to explain within the Hartian account. Hart's rule of recognition suggests that when certain procedural requirements are maintained, it is the law that allows and explains law's guidability. The problem is that for law to be able to guide, there have to exist some conditions that enable people to be guided by law. The account suggested here emphasizes that one of those conditions is some degree of pre-legal agreement on how to understand the texts from which one learns the content of the law. This is consistent with Hart's account but it is something that he has not elaborated on. A legal system that has a clear and simple Hartian rule of recognition will not guide behavior if those subject to the law will not be able to identify what the legal norms require. This point is trivially true if, say, the texts are written in some code (or foreign language) and cannot be understood unless we have some key for deciphering that code. And I suggest that the same is true where the “code” is a set of evaluative considerations (both moral and non-moral).

To see that such an idea is not inimical to the notion of guidability, consider the novel concept of cultural defense in criminal law: Proponents of this defense argue that defendants in criminal trials should be excused of their putative criminal acts when, due to their cultural backgrounds, they could not understand or comply with the requirements of the law. The best justification for such a defense is guidability; the idea is that in some instances the normative background of some defendants is so different from that of the rest of society that the person in question could not have been guided by the law, and therefore to punish her for something she did would be unfair.

What is true in the case of particular criminal norms may also be true more generally at the level of legal systems. For law to guide, its content must be something that those subject to it can understand. If I am right in my claim that part of one's normative background is essential for determining the content of the law, it follows that in order for law to guide, it must be the case that there is some agreement on one's normative background or at least that people with different normative background are given adequate explanation regarding the law's content. Because this is possible, it explains why differences in the rule of recognition make it possible for different people to be guided by law even if their understanding of what the law requires is occasionally different. It is important not to forget that I do not suggest that all the law is reducible to the normative background; the materials that make up particular legal norms still have the most important role in determining the content of legal norms, and this can mean that in certain circumstances even people with widely diverging normative background will agree on the content of certain legal norms. So even under the suggested account, guidability by law is still a real possibility.

Another aspect in which my account differs from Hart's is in the role of the rule of recognition as the foundation of the legal system—the sense in which it is some sort of “basic norm.” Hart's account of the rule of recognition not only serves as a means for identifying the primary rules but also provides a structural framework for understanding the foundations of a legal system. A legal system is recognized by one rule of recognition or one set of rules of recognition shared by all officials in that legal system. Hart's rule of recognition achieves this by being a kind of necessary headstone for the legal system. Once it is put in place, the legal system can (so to speak) take care of itself: the Constitution validates statutes, which validate bylaws, and all those validate individual law-changing activities (such as contracts) of private individuals. And though at various points Hart speaks of the internal aspect of rules generally, only in the case of the rule of recognition is it essential that some members of society (the officials) consider it from the internal point of view. My account, on the other hand, stresses the role of the rule of recognition in determining the content of every legal norm.

Combine the last two points, and another objection that emerges is that my account makes it impossible to tell when we have a legal system in existence or whether we have one or more legal systems within a certain area. But it is important not to misunderstand my claim. I do not suggest that, say, American citizens (and even more so, American judges) have completely different rules of recognition. Their rules of recognition overlap to a great extent. This partial overlap explains why all American judges accept certain events that occur in Congress as highly relevant for their job. But the overlap is not complete, and this explains some (though by no means all) of the differences in the ways different judges understand the content of legal norms.

In fact, this point enables us to see a solution to a problem that always nagged at Hart's account. In order to identify the legal norms, we need to turn to the officials who accept the rule of recognition, but the people who count as those officials are supposed to be recognized by certain rules of the legal system, which in turn are to be recognized by the officials' acceptance of the rule of recognition, and so on forever. What seems like an intractable problem for Hart can be relatively easily solved by the suggested account. The recognition of certain texts as establishing the legal system rests on the acceptance not by the officials but by members of society of certain documents and certain people as the legitimate law establishing texts of the state. That may be all that some people's “rule of recognition” may consist in. (They may, of course, know some other primary norms but they will know them without knowing any way of validating them; they may simply have heard them from someone they take to be an authority on such matters, such as a parent, teacher, or lawyer.) This may be exactly the kind of unquestioned normative background that is enough to get the machinery of the law going. Once this is in place, we may turn to other, more institutional aspects of the legal phenomena in order to decide whether two different rules of recognition belong to a single legal system. There are strong institutional reasons for considering all members of the English judiciary, despite their having somewhat different rules of recognition, as relating their rule of recognition to a single legal system.

Still, there might be instances in which this account might lead to the conclusion that, despite appearances, there is no law—or not a single legal system—within a certain geographical area that by other standards seems to have only a single legal system. This will happen if different courts nominally belonging to the same legal system (e.g., belonging to the same state) rely on the same materials but read them in utterly different ways. At some point we will say that there is no one legal system in that area or perhaps that there is no legal system in that area at all. This suggests a gradual rather than a binary account of the existence or inexistence of law or a legal system, and as such this is, I believe, a strength of the suggested account, as the non-revolutionary, gradual breakdown of a legal system as a result of diverging attitudes of members of a single society to their law seems like a real possibility. If this is true, then it exemplifies another weakness of Hart's account, for by Hart's lights, his account of the union of primary and secondary rules is a set of conditions that are all necessary and together sufficient for the existence of a legal system.³³

A related worry may be about the extent to which the account can explain the identity of legal systems through time; the normative background of nineteenth-century judges was very different from that of judges a century later. Yet we do not think that the nineteenth-century English judges who supposedly advanced the cause of laissez-faire by the pivotal role they assigned to freedom of contract belonged to a different legal system from those in the following generations who somewhat curtailed it. Yet it might be thought that, according to the account offered here, nineteenth-century English law is part of one legal system and twentieth-century English law part of another. Hart's account escapes this criticism because, arguably, the rule of recognition in Hart's sense has not changed in England between the nineteenth and the twentieth centuries. (I presuppose here that this difference can be traced back to certain normative background attitudes about liberty, equality, etc.) But to begin with, Hart's account is afflicted with a similar problem: suppose England abolished what is by now only a ceremonial act anyway—the requirement of the Queen's “enacting” laws in Parliament after getting “the advice and consent” of the Parliament. Would we say that England had changed its legal system? More significantly, the fact that English law now recognizes the jurisdiction of the European Court of Justice must have come through change of the rule of recognition in Hart's sense, and we do not need to conclude from this that the English have adopted a new legal system. Even though there is a historical sense in which the legal system before the change and after it is “the same,” there is also a sense, which it is important that an account of law would be able to grasp, of change. The suggested account only expands the list of potential causes of such transition. Note also that although the historical notion of a legal system is somewhat changed by the suggested account, the contemporaneous notion is kept intact: we can talk about one legal system as long as it maintains a certain required degree of unity in people's rules of recognition, along with some institutional marks of unity.

The last objection I will consider, and in my view the most significant one, concerns legal mistakes: if every legal system has numerous rules of recognition or attitudes, then we cannot explain or even logically allow for the possibility of legal mistakes. Every person has his or her legal truth, and none of those legal truths is better than the other. Yet it seems an incontrovertible fact about law that sometimes people get it wrong.

Let me begin by distinguishing three kinds of legal mistake. In the first case, a person may disregard or forget the relevance of a certain rule to the situation at hand. This kind of mistake can be easily distinguished from the two other kinds because in this type of error, once the mistake is pointed out, the person will in most circumstances acknowledge it and change her view. The second kind of mistake is not really a mistake but it is close enough to mention here: the vagueness of most words results in instances in which people disagree about the applicability of certain words (or concepts) to them. In such a case, two people may disagree whether a defendant's behavior was negligent. If a first-instance court decides a case one way, and its decision is overruled on appeal only because of difference in drawing the borderline between negligence and non-negligence, the lower court's opinion is a “mistake” only in the sense that the higher court's opinion is deemed for reasons of hierarchy and unity of the legal system to be the correct one. Note that this is true whether the cause of vagueness is metaphysical or epistemic; even if there are metaphysically correct answers to vague questions, as long as we cannot know them, the lower-court judge's decision can be considered mistaken only because the higher court's decision is considered conclusive on the matter. In the third kind of error, we would ascribe the disagreement to different normative backgrounds. In reality it would be very difficult to distinguish between cases of the second and third kind. It might even be that all cases that look as though they are only cases of difference of opinion about borderline cases are better explained as resulting from different normative backgrounds, that is, that it is the different normative background that explains why two people differ about the application of certain vague concepts.

I am not sure whether this is true, but this question can be left open, because my explanation of the third kind of mistake is quite similar to my explanation of the second kind (with one important qualification about moral anti-realism that I mention in the next section). The explanation is that a certain view is deemed correct in the law because it is accepted as such by the relevant legal practitioners. And most practitioners accept (as a matter of fact) the legal rule according to which the pronouncements of the majority of a state's highest court represent the law, although an influential high court will probably succeed in doing more than that by imprinting its rule of recognition on lower courts and other members of society by elaborately assuming and defending its normative background, and in this way convincing others that its decisions are by and large correct. This is one of the reasons why high courts' judgments tend to be long and rhetorical.

But can a state's highest court get the law wrong? The answer depends on what we mean by this question. In one sense this is possible; this would be the case when, based on the legal materials in existence before a state's highest-court decision had been made, there had been one decision that was warranted according to the views of most members of the legal community, and the court then reached a different one. An argument to that effect can be made (and in pure form will only be made) based on the normative background that underpinned the decision the court had made. But in another sense, the highest court cannot make a legal mistake; what the highest court decides determines the law. Thus legal errors by the highest court of the kind described above are “self-correcting” in the sense that by being made, such decisions change the legal materials in existence and in this way become correct. Even if the court later decides to depart from its earlier decision because its earlier decision “got the law wrong,” even if the decision “was wrong the day it was decided,” this will amount to another change in the law. This is most clearly true in legal systems that recognize precedent, because in these legal systems, lower courts are bound by the rule set forth by the “erroneous” decision in the interim period before the highest court decides differently. Even in legal systems where precedent is not formally considered a source of law, actual practice is usually quite similar to that of legal systems where it is.

This last point may seem problematic, because it suggests that high court judges cannot be guided by law for there is no standard by which to check the correctness of their decision. And if there is no such standard of correctness, then high court judges also lack a standard by which to guide themselves. But we (and high court judges) do not think there are no constraints on their decision. So there must be something wrong with the suggested account. There is some truth to this objection, but it is a truth that is worth preserving. If by correctness we mean some standard the judge can consult in order to guarantee that her decision is the right one, that is, to guarantee that her decision accords with what the law requires, then such a standard does not exist. But it would be wrong to think that this implies that the high court judge has nothing to look for in deciding cases. The combination of legal materials and normative background will make certain solutions seem plainly wrong, some answers possible, and in many situations, one answer seem the one required by law.

The decision is then handed down in public to be assessed, and as I said, if the decision is considered by the community to have been legally wrong, then there is a sense in which the court erred. (Matters here are complex, of course, because the distinction between legal correctness and all-things-considered correctness is often not clearly distinguished, especially by laypeople, and is also denied by some lawyers as well.) But ultimately the judge has to decide what she believes is demanded by law, and in this, as numerous judges have acknowledged, the judge, especially the high court judge, is alone.³⁴

WHAT DOES THE ARGUMENT MEAN FOR POSITIVISM?

The last section is not intended to be a theory of law. Many issues that received a paragraph probably require a chapter; many other questions have not been touched at all. For instance, one of the aspects of Hart's rule of recognition is that it is a social rule in a sense of regularity of behavior, whereas my account emphasized the “mental” aspect of the different rules of recognition. To what extent must we have such regularity of behavior in order to have a legal system? Do we give more weight to some people's behavior (e.g., judges or other officials) than that of others? How does this understanding of the foundation of law fit the existence of sanctions as a mechanism of compliance with the law? All these are important questions, but for the sake of the point I was trying to make, namely, that to understand the content of law we must adopt a broader understanding of the rule of recognition, I hope what is offered here will suffice. The question with which I would like to conclude is the extent to which this account is consistent with legal positivism.

Legal positivism is a label, and labels are not particularly important. Yet considering this question might help clarify what is at stake. On the one hand, my claim that we need moral and other evaluative considerations in order to determine the content of legal norms may seem like the rejection of legal positivism. This is true especially if legal positivism is taken to be the thesis according to which law and morality are necessarily separate or separable or even the narrower thesis that it is possible to know the content of legal norms without recourse to evaluative considerations.

But positivists and their detractors have often lumped together three different questions about the relationship between law and morality: normativity (or authority), validity, and content. Of the three, the Argument focuses on only the latter two. Some philosophers (for instance, Dworkin) seem to argue that the three are closely tied; at the most general level, Dworkin's argument with positivism is exactly the rejection of the possibility of knowing whether a putative norm is legally valid without considering its content, and I take his silent prologue argument, mentioned above, as collapsing the distinction between content and normativity as well.

But I believe it is important to keep these matters separate. With regard to the content of norms, the proposed account is quite clear in arguing for a connection between the content of norms and evaluative (including moral) considerations. However, it leaves one possibility open. My account says nothing about the origins of the evaluative considerations that make up part of our normative background. In particular, it says nothing about the claim that the set of evaluative considerations is no more than a set of social rules. In that case, the resulting theory looks much closer to positivism (or some versions of it) than to any anti-positivist theory, for it vindicates one fundamental positivist thesis that the law is a matter of social fact. This would be positivism, but one arrived at in an indirect way; law would then not be related to morality, if the latter is understood as a set of norms that exist independently of social attitudes, because morality in that sense does not exist. If this means that all evaluative considerations are not genuinely evaluative, then legal positivism is vindicated. Trivial or obvious as this point may seem, it is worth mentioning, because I believe that some prominent legal positivists (e.g., Hart and Kelsen) were at least to a certain degree drawn to their positivism by their metaethical skepticism.

The question of normativity is more complicated. The question that legal positivists have struggled with is whether it is possible to explain law's normativity without invoking moral considerations. The Argument is consistent with this possibility, and in fact I believe this is a possibility worth pursuing. This is the reason why my account keeps the term “rule of recognition” despite the significant changes proposed, and I hope that by doing this I maintain more than just a label. One reason Hart introduced the notion of a rule of recognition (within a broader account of social rules) was that he wanted to explain how law's normativity could be accounted for without resorting to the normativity of morality. Many believe that Hart's account, perhaps refashioned using more sophisticated accounts of conventions, joint activities, or plans, can do the trick. Others, even among legal positivists, believe that this road inevitably leads to a dead end. This is not the place to examine this question. I believe that even if the Argument is successful, it does not (necessarily) conflict with a non-moral account of law's normativity. Opting for such an account may seem like a strange tack to take for someone who believes that knowing the content of law requires evaluative (and moral) considerations, but in fact it is one embraced by many contemporary (inclusive) positivists. True, if the Argument is successful, then the particular argument they suggested is unsuccessful, but it may still be that a different account would fare better.

In that case, the question whether knowing the content of law requires evaluative considerations becomes secondary. True, claims of conceptual separation between law and morality have been something of a battle cry for legal positivists for quite some time and are probably the most important aspect of the history of this idea. But in recent years quite a few positivists have seen no conflict between their positivism and a belief in the existence of necessary connections between law and morality. And if, as Jules Coleman has argued, the central aspect of legal positivism today is that law is conventional,³⁵