As Jules Coleman remarks,Footnote 1 Joseph Raz's account of authority represents one of his most significant contributions to legal and political philosophy. Among other things, this account seeks both to identify the role that the directives of a legitimate authority should play in the practical reasoning of its subjects and to determine when a purported authority is in fact legitimate.Footnote 2 This has implications for the law because, Raz argues, the law necessarily claims to possess authority. His account of authority, then, purports to tell us (among other things) what must be true in order for the law's claim to be correct.
Raz further argues that given that the law necessarily claims authority, and given the account of authority he has provided, exclusive legal positivism is the only tenable theory of law. I shall call this argument “the argument from authority.”Footnote 3 In it, Raz purports to derive important conclusions about the nature of law from his account of authority (together with the fact that the law necessarily claims authority).
Despite being extremely influential, both Raz's account of authority and the argument from authority have been criticized on multiple grounds. Numerous commentators argue that Raz mischaracterizes the role that an authority's directives should play in the reasoning of its subjects.Footnote 4 His account of when a purported authority is legitimate has also come under fire.Footnote 5 Turning to the argument from authority, there is fierce debate about whether the law necessarily claims authority.Footnote 6 And Coleman argues that even if the law necessarily claims authority and even if Raz's account of authority is correct, this does not establish the truth of exclusive legal positivism.Footnote 7
In this article, I offer a version of the last type of criticism. I assume, for the sake of the argument, that Raz's account of authority is correct. I also assume that the law necessarily claims authority. However, I challenge Raz's contention that his account of authority, together with the fact that the law necessarily claims authority, provides the basis for a sound argument in favor of exclusive legal positivism.
While Coleman also offers this type of criticism, he seeks to refute the argument from authority by showing that exclusive legal positivism is not the only theory of law that can satisfy the requirements of Raz's account of authority. My objection goes further. I argue that even if Raz's account of authority is correct and even if exclusive legal positivism is the only theory of law that can satisfy the requirements of that account, it still does not follow that exclusive legal positivism is the only tenable theory of law. This is because even if the law necessarily claims authority, it need not be capable of satisfying the requirements for possessing authority laid down by Raz's account of authority.
Section I of this article provides an overview of the argument from authority. In Section II, I explain how Raz's claim that the law must be capable of possessing authority depends on his further claim that legal officials cannot be systematically confused about the concept of authority. Section III seeks to understand this further claim in light of Raz's criterial approach to conceptual explanation (which I label “Raz's criterialism”). I suggest that Raz's criterialism provides support for the claim that legal officials cannot be systematically confused about the concept of authority. However, to overcome various objections that arise on this interpretation, Raz needs to (and does) recognize the existence of multiple concepts of authority. This, I argue, is fatal to the argument from authority. In Section IV, I consider an alternative interpretation of Raz's claim that legal officials cannot be systematically confused about the concept of authority. I argue that this interpretation, which does not draw on his criterialism, also fails to vindicate that claim. I conclude that the argument from authority is unsound.
I. THE ARGUMENT FROM AUTHORITY
Raz's account of authority is sufficiently well-known that I will not summarize it here. However, I will set out the argument from authority, since the details of that argument are perhaps less well known and will be crucial in the discussion that follows.
The argument from authority contains four core contentions:
(1) The law necessarily claims to possess authority.
(2) Therefore, the law must be capable of possessing authority.
(3) If the law is to be capable of possessing authority, two prerequisites must be satisfied:
(a) legal directives must reflect, or at least be presented as reflecting, someone's view of how the law's subjects ought to behave; and
(b) it must be possible to identify the existence and content of legal directives without relying on considerations on which those directives purport to adjudicate.
(4) Only exclusive legal positivism is compatible with the two prerequisites.
Let us consider each of these contentions in turn.
Raz begins by suggesting that every legal system claims to possess authority: “though a legal system may not have legitimate authority, or though its legitimate authority may not be as extensive as it claims, every legal system claims that it possesses legitimate authority.”Footnote 8 How do we know this? Raz suggests that we can see what claims the law makes by examining the views expressed by its spokespeople (i.e., legal officials).Footnote 9 That the law claims to possess authority is demonstrated:
by the fact that legal institutions are officially designated as “authorities,” by the fact that they regard themselves as having the right to impose obligations on their subjects, by their claims that their subjects owe them allegiance, and that their subjects ought to obey the law as it requires to be obeyed (i.e. in all cases except those in which some legal doctrine justifies breach of duty).Footnote 10
According to Raz, these are not contingent facts about particular legal systems; rather, they are necessarily true of all legal systems.Footnote 11 It is, therefore, a necessary feature of law that it claims authority.
Raz claims that it follows from the fact that the law necessarily claims authority that it must be capable of possessing authority: “To claim authority it must be capable of having it.”Footnote 12 This leaves open the possibility that some (perhaps all) legal systems lack legitimate authority: to be capable of possessing authority is not to possess it. However, it does mean that legal directives must satisfy two prerequisites that any directive must satisfy in order to be capable of being authoritatively binding (henceforth, “the two prerequisites”). The two prerequisites are:
First, a directive can be authoritatively binding only if it is, or is at least presented as, someone's view of how its subjects ought to behave. Second, it must be possible to identify the directive as being issued by the alleged authority without relying on reasons or considerations on which [the] directive purports to adjudicate.Footnote 13
The two prerequisites are derived from Raz's account of authority. Let me illustrate by reference to the second prerequisite. Why must it be possible to identify the existence and content of an authoritative directive without reference to the considerations on which it purports to adjudicate? Because otherwise neither the normal justification thesis nor the preemption thesis could be satisfied. The authority's subjects could not do better by trying to follow its directives than by considering their dependent reasons for themselves if they needed to consider their dependent reasons in order to identify the existence or content of the authority's directives. Nor is a directive capable of replacing our dependent reasons in our practical reasoning unless we can identify the existence and content of that directive without considering those reasons.Footnote 14
Raz argues that source-based law satisfies the two prerequisites. What is source-based law? “A law is source-based if its existence and content can be identified by reference to social facts alone, without resort to any evaluative argument.”Footnote 15 Source-based law satisfies the first prerequisite, since the sources of law (such as legislation and precedent) express—or at least are presented as expressing—someone's judgment about what the law's subjects should do. It also satisfies the second prerequisite, because we do not need to consider our dependent reasons in order to identify the existence and content of source-based law. All we need to do is consider the relevant social facts. Since exclusive legal positivism claims that all law is source-based, it is compatible with the two prerequisites.
However, Raz argues, rival theories of law—such as inclusive legal positivism and Ronald Dworkin's interpretivism—are not compatible with the two prerequisites. This is because they deny that law is entirely source-based, holding instead that (at least in some possible instances) moral reasoning is needed to identify the existence and content of legal directives. This is incompatible with the first prerequisite because “[m]oral argument can establish what legal institutions should have said or should have held but not what they did say or hold.”Footnote 16 Treating moral considerations as relevant to the identification of legal directives is therefore inconsistent with the requirement that legal directives reflect someone's (at least purported) judgment about how the law's subjects should act. Holding that moral reasoning is required to identify the existence and content of legal directives is also incompatible with the second prerequisite. It means that in order to identify the existence and content of legal directives, we must take account of the very considerations on which those directives purport to adjudicate.Footnote 17
II. THE SUPPORTING ARGUMENT
Since I am assuming that the law necessarily claims authority, I must accept the first contention in the argument from authority. However, why should we accept the second contention? How does it follow from the fact that the law necessarily claims authority that it must be capable of possessing authority? Raz tells us that we can see what the law claims by looking at the views of legal officials. However, might legal officials be insincere, claiming authority even though they know that the law is incapable of possessing it? Or might legal officials be sincere but confused, believing that the law possesses authority when in fact it is incapable of doing so?
If it is possible that legal officials are insincere or confused in this way, then we cannot infer that the law must be capable of possessing authority from the fact that it necessarily claims authority. Yet if we cannot draw this inference, the argument from authority collapses. If the law need not be capable of possessing authority, then it need not satisfy the two prerequisites. This would mean that even if exclusive legal positivism is the only theory of law that is compatible with the two prerequisites, this does not constitute a reason for accepting exclusive legal positivism.
Raz is aware of this potential objection, and he offers a response that I will call “the supporting argument.” He dismisses the possibility that the law is normally insincere or conceptually confused in its claim to authority: “Since the claim is made by legal officials wherever a legal system is in force, the possibility that it is normally insincere or based on a conceptual mistake is ruled out.”Footnote 18
I will leave the possibility of insincerity to one side, and focus on the possibility of conceptual error. With regard to that possibility, Raz elaborates as follows:
Why cannot legal officials and institutions be conceptually confused? One answer is that while they can occasionally they cannot be systematically confused. For given the centrality of legal institutions in our structures of authority, their claims and conceptions are formed by and contribute to our concept of authority. It is what it is in part as a result of the claims and conceptions of legal institutions.Footnote 19
Since the claims made by legal officials help constitute our concept of authority, those claims cannot systematically diverge from the correct account of that concept. In other words, legal officials cannot be systematically confused about the concept of authority.
What counts as “systematic confusion”? Raz does not explain what he means by this phrase, but we can gain some understanding of it by considering the role it plays in the supporting argument. That argument is meant to license the inference that the law must be capable of possessing authority, since legal officials claim that it possesses authority and, the supporting argument purports to show, they cannot be systematically confused about the concept of authority. If only a few officials are confused about the concept, that inference can still be drawn. The same is true if all legal officials are confused about aspects of the concept that are unimportant when determining whether the law is capable of possessing authority. However, if confusion about the concept of authority is widespread among legal officials and concerns aspects of the concept that are important when determining whether the law is capable of possessing authority, then the fact that officials claim that the law possesses authority does not license the inference that it is capable of possessing authority.Footnote 20 The supporting argument must exclude the possibility of confusion of this type, and so it is confusion of this type that I shall regard as “systematic” confusion.
I will seek to render this characterization of systematic confusion more precise as the discussion proceeds. However, I concede at the outset that this characterization may not be intuitive. Intuitively we may not regard legal officials as systematically confused about the concept of authority unless they are confused about most aspects of the concept. However, to undermine the supporting argument, it is not necessary to show that officials are confused about most aspects of the concept. It is enough if they are confused about those aspects that are important when determining whether the law is capable of possessing authority. I say more about this below.
III. IS THE SUPPORTING ARGUMENT SOUND? THE CRITERIALIST READING
A. The Nature of Authority and the Concept of Authority
I started Section II by asking whether legal officials might be confused, believing that the law possesses authority when in fact it is incapable of doing so. I then suggested that Raz's response is to deny the possibility of systematic confusion among legal officials about the concept of authority. However, why think that any confusion on the part of officials must concern the concept of authority?
One way in which legal officials could believe that the law possesses authority when in fact it is incapable of doing so is if they were mistaken about the nature of authority—in particular, if they were mistaken about the requirements for possessing authority laid down by the correct account of the nature of authority. However, for Raz, the correct account of the nature of authority is part of a true explanation of the concept of authority. This follows from his general approach to conceptual explanation, according to which an explanation of a concept is in part an explanation “of that which it is a concept of.”Footnote 21 It follows that confusion about the nature of authority is a form of conceptual confusion.
This helps explain why Raz is concerned to show that legal officials cannot be systematically confused about the concept of authority. However, might officials be confused not about what the requirements for possessing authority are but rather about whether the law is capable of satisfying those requirements? This would involve confusion about the law, not the concept of authority, but it would still preclude us from inferring that the law must be capable of possessing authority from the fact that officials claim that it possesses authority.Footnote 22
Raz does not discuss this possibility, and I shall not pursue it further. Rather, I shall argue that the supporting argument fails even if we confine our attention to the possibility of systematic confusion about the concept of authority.
Before proceeding, we should note an important implication of Raz's view that the correct account of the nature of authority is part of a true explanation of the concept of authority. It follows that Raz's three central theses about the nature of authority—the dependence, normal justification, and preemption theses—are conceptual claims, in that they provide part of a true explanation of the concept of authority. (I shall refer to these three theses as “the key theses.”) Raz is clear about this. For example, in The Morality of Freedom, after setting out the key theses, he states: “Three theses were presented as part of an explanation of the concept of authority.”Footnote 23
It is true that the normal justification thesis is a normative thesis about when someone possesses morally legitimate authority.Footnote 24 According to Raz, however, this does not preclude it from being part of an explanation of the concept of authority, because “there is an interdependence between conceptual and normative argument.”Footnote 25 The correct account of the nature of (morally legitimate) authority must be normative, and on Raz's approach that account forms part of the explanation of the concept of authority.Footnote 26
Nor does treating the key theses as part of an explanation of the concept of authority commit Raz to claiming that someone who is confused about those theses lacks an understanding of that concept. He distinguishes between a basic understanding of a concept and complete mastery of it.Footnote 27 Depending on how one understands this distinction,Footnote 28 it may allow Raz to maintain that someone who is confused about the key theses may nevertheless have a basic understanding of the concept of authority (though he or she lacks complete mastery of the concept).
However, whether legal officials may be confused about the key theses is crucial when assessing the supporting argument. Let me explain. Raz must show that officials cannot be systematically confused about the concept of authority—that is, that there cannot be widespread confusion among legal officials about aspects of the concept that are relevant to determining whether the law is capable of possessing authority. Which aspects are these? To answer this question, we need some idea of what it is for the law to be capable of possessing authority. According to Raz, the law is not capable of possessing authority unless the two prerequisites are satisfied. We have seen that the two prerequisites are derived from the key theses, and that the key theses are theses about the concept of authority. Thus confusion about the key theses is the sort of conceptual confusion that if widespread among legal officials, would undermine the supporting argument.Footnote 29 For this reason, in assessing whether legal officials can be systematically confused about the concept of authority, I shall consider whether there can be widespread confusion among officials about the key theses.
B. Raz's Criterialism
We have seen that Raz regards an explanation of a concept as in part an explanation of that of which it is a concept. He also suggests a particular approach to conceptual explanation, which he refers to as a “criterial” approach and which I will refer to as “Raz's criterialism.” In this subsection, I offer an interpretation of the supporting argument that draws on Raz's criterialism.
What is Raz's criterialism? According to Raz, a criterial explanation of a concept:
(1) states a rule setting out conditions for the (correct) use of a concept; and (2) is a true explanation by virtue of the fact that it is a correct statement of the conditions for the correct use of the concept actually used by those who use it.Footnote 30
When one uses a concept, one relies on a rule setting out criteria (or conditions) for the use of that concept. The correct criteria are those actually used by the people whose concept it is. Nevertheless, any one of those people may be mistaken about what the correct criteria are, since the correct criteria are those that are generally believed to be correct, not those that any particular individual believes to be correct.Footnote 31
Raz does not refer to his criterialism when presenting the supporting argument.Footnote 32 Nevertheless, it may shed light on that argument. Raz asserts that legal officials can be occasionally but not systematically confused about the concept of authority. His criterialism may explain why this is so. Occasional confusion can arise because individual legal officials may be mistaken about which criteria for applying the concept of authority are generally believed to be correct. However, there cannot be systematic confusion about the correct criteria because, according to Raz's criterialism, the criteria that are generally believed to be correct are correct.
If this is the basis on which Raz denies that there can be systematic confusion among legal officials about the concept of authority, then we must interpret strictly his claim that the correct criteria for applying a concept are the criteria that are generally believed to be correct. For example, we cannot ascribe to Raz the view that conceptual analysis involves considering not only our beliefs about the concept but also what is logically entailed (or otherwise implied) by those beliefs.Footnote 33 On that view, there could be systematic confusion about the concept of authority because there could be widespread confusion about what is logically entailed by our generally shared beliefs about the concept.
Interpreting the supporting argument in light of Raz's criterialism may explain why legal officials can be occasionally but not systematically confused about the concept of authority. However, this seems to raise other problems. In what follows, I mention three such problems and then suggest how Raz might resolve them.
First, we have seen that Raz's criterialism entails that the correct criteria for applying the concept of authority are the criteria that are generally believed to be correct. However, Raz's account of authority is unlikely to be familiar to many people, even in Anglo-American legal systems, let alone in other legal systems. His account may contain certain truisms about authority about which few people are likely to be confused. However, I have suggested that when assessing the supporting argument, we should focus on the possibility of widespread confusion about the key theses, and these theses are not truisms. Indeed, Raz acknowledges that one would not be justified in assuming that people know of the key theses.Footnote 34 It would seem to follow that the key theses are not generally believed to be correct. If Raz's criterialism is correct, it further follows that the key theses cannot feature among the correct criteria for applying the concept of authority. In sum, it seems that Raz's criterialism establishes a test that the key theses cannot pass.
Second, Raz suggests that users of a concept may be ignorant of or mistaken about many aspects of that concept and that an explanation of a concept may be correct even if it is not generally available to the users of the concept:
there is no implication in the points above that an explanation of a concept to be correct must be one generally available to those who have the concept. There are many aspects of a concept that its users may not be aware of, and many mistakes about it they may make.Footnote 35
This might explain how Raz's account of authority can be correct even though many people are unfamiliar with the key theses. However, it sits uneasily with his criterialism. According to Raz's criterialism, the correct criteria for applying a concept are those that are generally believed to be correct. This allows for the possibility that some individuals are mistaken about which criteria are generally believed to be correct. However, it is not immediately apparent how it allows for the possibility that people in general are mistaken about the correct criteria or that the correct explanation of a concept is not generally available to the users of that concept.
Third, if the correct explanation of the concept of authority may not be generally available to those who have the concept, then it seems that legal officials can be systematically confused about that concept. In particular, if officials may generally be ignorant of or mistaken about the key theses, then it seems that we cannot infer that the law is capable of possessing authority from the fact that officials claim that it possesses authority.Footnote 36
None of these problems is fatal to Raz's argument, however. To see this, we must distinguish between what we take the criteria for applying a concept to be and the criteria we actually use in practice. I will refer to the former as our “manifest” beliefs and the latter as our “operative” beliefs about the correct criteria.Footnote 37 Thus our manifest beliefs are our consciously held beliefs about the correct criteria for applying the concept, whereas our operative beliefs are revealed in the way we apply that concept in practice.Footnote 38
Let me give an example to illustrate this distinction. We might take the criterion for applying the concept of bachelorhood to be “an unmarried man.” This would be our manifest or consciously held belief about the correct criterion for applying the concept. However, it might not be the criterion we use in practice, as can be seen from the fact that we might not apply the concept of bachelorhood to the Pope, even though he is an unmarried man.Footnote 39 In other words, there may be a divergence between our operative beliefs about the concept of bachelorhood (i.e., the criteria that govern our use of the concept) and our manifest beliefs about the concept (i.e., the criteria we consciously believe to be correct).Footnote 40
When Raz states that the correct criteria for applying a concept are those that are generally believed to be correct, he should be interpreted as referring to our operative beliefs. His criterialism focuses on the way in which we use concepts, since a true explanation of a concept “is a correct statement of the conditions for the correct use of the concept actually used by those who use it.”Footnote 41 The criteria that we actually use when applying a concept constitute our operative beliefs. Thus Raz's criterialism contends that the correct criteria for applying a concept are the criteria we generally use in practice—that is, the operative beliefs we generally share.Footnote 42
Raz should also be interpreted as claiming that our manifest and operative beliefs can diverge. He states that the correct criteria—by which he means, I have suggested, the criteria we use in practice—may be opaque to us, in the sense that we may have only partial knowledge of the rules we follow for the use of the concept.Footnote 43 In other words, our manifest beliefs may diverge from our operative beliefs.
This provides the key to addressing the problems canvassed above. When Raz says that we could be ignorant of or mistaken about many aspects of the concept of authority, he should be interpreted as talking about our manifest beliefs. Since our manifest beliefs can diverge from our operative beliefs, this is compatible with claiming that the correct criteria for applying the concept of authority are our generally shared operative beliefs. Even if we generally share a practice of using certain criteria for applying the concept of authority, we may have no consciously held beliefs about some aspects of that practice, and what consciously held beliefs we do have may conflict with other aspects of the practice. It is in this sense that the correct explanation of the concept may not be generally available to its users.
Similarly, when I suggested that most legal officials are unfamiliar with Raz's account of authority, I was concerned with their manifest beliefs. The point was that Raz's account does not feature in most officials’ consciously held beliefs about the concept of authority. This leaves open the possibility that his account reflects most officials’ operative beliefs. That could be true whether they are aware of his account or not. Thus Raz's account of authority may be compatible with his criterialism, after all.
Moreover, in assessing the supporting argument, the crucial question is whether legal officials can be systematically confused in their operative, not manifest, beliefs about the concept of authority. We saw that the supporting argument seeks to license the inference that the law must be capable of possessing authority, since legal officials claim that it possesses authority and, the supporting argument purports to show, they are not conceptually confused when making that claim. When officials claim that the law possesses authority, they are using the concept of authority rather than expressing consciously held beliefs about the criteria for applying that concept. One's use of a concept is responsive to one's operative beliefs, and so the claim by officials that the law possesses authority is responsive to their operative—not manifest—beliefs about the concept of authority. Thus the supporting argument needs to show that there cannot be systematic confusion in officials’ operative beliefs about that concept, but it can allow for systematic confusion in their manifest beliefs.
Raz does, however, have to show that the key theses feature among officials’ operative beliefs about the concept of authority. How might he go about showing this? Any attempt to answer this question is complicated by the fact that officials may not be consciously aware of their operative beliefs. However, it would seem to be enough for Raz to show that the way in which officials use the concept of authority is best explained by reference to the key theses. With regard to the normal justification thesis, this might require showing that officials typically treat X as an authority in circumstances where they believe that following its directives would better enable its subjects to conform to their dependent reasons (or where they would believe this if the normal justification thesis was explained to them).Footnote 44 With regard to the dependence thesis, it might require showing that officials typically act as if they expect an authority's directives to be based in part on reasons that apply to its subjects anyway. Finally, with regard to the preemption thesis, it might require showing that once officials recognize X as an authority, they typically require that its subjects treat its directives as replacing at least some of the subjects’ dependent reasons. (For example, they disapprove of subjects who treat X's directives as merely particularly weighty reasons to be balanced against the subject's other reasons.) Whether Raz can show this is something I consider in the next two subsections.
In this subsection, I have ascribed the criterialist reading to Raz. However, it might be doubted whether this is what he has in mind. When presenting the supporting argument, he does not explicitly distinguish between the criteria for applying the concept of authority that we consciously believe to be correct and the criteria we use in practice. He does not even explicitly refer to his criterialism; he talks about “the centrality of legal institutions in our structures of authority,” not about criterial explanations of concepts.Footnote 45 Nevertheless, I think it is appropriate to interpret the supporting argument in light of Raz's criterialism, since it is likely that his views concerning the possibility of systematic confusion about the concept of authority are integrated with his more general views about conceptual explanation and error.
Even if the criterialist reading is not what Raz has in mind (a possibility I explore in Section IV), it is still important to consider whether the supporting argument is sound when given that reading. As we have seen, the initial signs are promising. The criterialist reading appears to show that officials’ operative beliefs about the concept of authority cannot be systematically confused and that this is sufficient to vindicate the supporting argument while allowing Raz to acknowledge that officials’ manifest beliefs about the concept of authority may be systematically confused.
C. Recognizing Multiple Concepts of Authority
To assess properly the criterialist reading of the supporting argument, however, we must note that Raz recognizes the existence of multiple concepts of authority. He refers not only to “our” concept of authority but also to a series of “ancestors” of our concept: “our concept is a descendant of earlier concepts.”Footnote 46 Furthermore, there is an umbrella concept of authority that covers both our concept and all of its ancestors, although, Raz adds, the umbrella concept is not the “real” concept of authority since the only way to identify it is “as the concept that applies to all instances of what I called ‘our’ concept of authority and those of its ancestors (rather than by its ahistorical features).”Footnote 47 Finally, he states that we should allow for the possibility that there is no single concept of authority that is “ours” but, rather, several concepts, all of which are descended from the same ancestors.Footnote 48 In other words, there may be multiple “communities of speakers” within our society, each such community having a different concept of authority.Footnote 49
In the next subsection, I argue that Raz's recognition of multiple concepts of authority undermines the argument from authority and renders his criterialism less attractive than many of its contemporary rivals. Before doing so, however, it is worth noting the advantages for Raz of recognizing multiple concepts of authority. Doing so preserves his criterialism from two serious objections. It also defuses a potentially fatal objection to the supporting argument.
Let us start with the first objection to Raz's criterialism. Raz claims that the correct criteria for applying the concept of authority are those that are generally believed to be correct. He is therefore committed to claiming that there are criteria for applying the concept of authority that are generally believed to be correct. Indeed, he is committed to claiming that all the correct criteria for applying that concept are generally believed to be correct, since this is what makes them the correct criteria.
This requirement is not satisfied with regard to our manifest beliefs. Consider the normal justification thesis. We have seen that Raz treats this thesis as part of a true explanation of the concept of authority. However, there is currently an important debate between his supporters who embrace the normal justification thesis and many of his critics who reject it.Footnote 50 While Raz could explain the presence of an occasional dissenter on the basis that they are mistaken about which criteria are generally accepted, there are enough people on each side of this debate that neither view can plausibly be said to be generally accepted.Footnote 51 At the level of manifest beliefs, then, not all of the correct criteria for applying the concept of authority are generally believed to be correct.Footnote 52
One reason for interpreting Raz's criterialism in the way I have is that it enables him to concede that at the level of manifest beliefs, there may be widespread disagreement about the criteria for applying the concept of authority. This leaves open the possibility that at the level of operative beliefs, there is general agreement about those criteria. It may be that people generally use the same criteria for applying the concept of authority but are mistaken about what those criteria are, leading to a divergence in manifest beliefs. This would be all Raz's criterialism needs: as I have interpreted it, it requires general agreement in operative, not manifest, beliefs.
However, the objection continues, it is not enough that there may be general agreement in operative beliefs. Raz needs to show that such agreement does in fact exist. For example, he needs to show that despite the different manifest beliefs people hold about the normal justification thesis, that thesis generally features among their operative beliefs.Footnote 53 I suggested above that to show this, he needs to show that we typically treat X as an authority in circumstances where we believe that following its directives would better enable its subjects to conform to their dependent reasons (or where we would believe this if the normal justification thesis were explained to us). Yet Raz does not engage in the sort of (philosophically informed) study of our practices that would be needed to show this.
Moreover, even if it is true of our society that people generally use the same criteria for applying the concept of authority, there are—or at least could be—other societies of which this is not true. For example, there could be societies in which there is widespread disagreement (even at the level of operative beliefs) concerning the normal justification thesis. We can imagine a society in which people are split roughly fifty-fifty between those who use the concept of authority in accordance with the normal justification thesis and those who use it in accordance with a competing test of when a purported authority is legitimate.Footnote 54 How can this possibility be reconciled with Raz's criterialism, which holds that the correct criteria for applying the concept of authority are those that are generally believed to be correct?
This objection is not fatal to Raz's criterialism, however. One way in which he can respond is by contending that in any society in which there are widespread differences in operative beliefs about the concept of authority, there are multiple concepts of authority corresponding to the different operative beliefs that are widely held within that society. (Similarly, he can contend that widespread differences in operative beliefs between different societies show that those societies have different concepts of authority.) Raz can then maintain that for any particular concept of authority, the correct criteria for applying the concept are the operative beliefs that are generally shared among those who hold the concept. In this way, he can reconcile his claim that the correct criteria are those that are generally believed to be correct with the possibility of widespread differences in operative beliefs (either within a particular society or across different societies).
The second objection to Raz's criterialism is independent of the first. Even if there are generally shared operative beliefs about the concept of authority, it seems that those beliefs can be mistaken. If so, Raz is wrong to claim that the correct criteria for applying the concept of authority are those that are generally believed to be correct.
One of Raz's own examples can be used to illustrate this objection.Footnote 55 Imagine that our ancestors once believed that the criteria for applying the concept of authority are such that an authority must have a divine source. This may have been not only their manifest belief but also their operative belief—they may have applied the concept only to those purported authorities that they regarded as having a divine source. However, like Raz, I regard this belief as mistaken. The legitimacy of a purported authority does not depend on it having a divine source. It seems, therefore, that Raz is wrong to claim that the correct criteria for applying the concept of authority are whatever criteria are generally believed to be correct. Operative beliefs could be generally shared within a community and yet be mistaken.
Again, one way for Raz to respond to this objection is to recognize the existence of multiple concepts of authority. Indeed, he at least contemplates making precisely this move. He states: “if ever the concept of authority was such that it had to derive from divine authority, then recognition of the impossibility of divine authority may have encouraged change in the concept.”Footnote 56 On this view, those who believed that authority must derive from a divine source may have been right about their concept, but if so, we changed concepts when we abandoned that belief (so that, according to our new concept, authority does not have to derive from a divine source). Thus, in cases where his critics suggest that our ancestors were mistaken in their operative beliefs about the concept of authority, Raz can respond that they were right about their concept of authority, but we subsequently adopted a different concept of authority. Similarly, a claim that the operative beliefs that we now share are mistaken can be interpreted as a call for us to abandon our current concept of authority in favor of a new one.
Finally, Raz's recognition of multiple concepts of authority also provides a response to an objection to the criterialist reading of the supporting argument. Let me first present the objection. Raz's criterialism suggests that at the level of operative beliefs, we cannot be systematically confused about the criteria for applying the concept of authority. This is because the operative beliefs that we generally share constitute the correct criteria. However, who counts as “we”? One possible answer is that it is the operative beliefs generally shared within society as a whole that constitute the correct criteria. Yet legal officials form only a small subgroup of our society. Thus, as far as Raz's criterialism is concerned, it appears that officials’ operative beliefs about the concept of authority can be systematically confused, even if there cannot be systematic confusion within society as a whole. Yet the supporting argument is meant to exclude the possibility that legal officials are systematically confused about the concept of authority. It seems, therefore, that the supporting argument cannot be based on Raz's criterialism in the way suggested by the criterialist reading.
It may be unlikely that legal officials are systematically confused about the concept of authority if society as a whole cannot be. However, the mere possibility would be fatal to Raz's argument. Exclusive legal positivism purports to identify a necessary truth about law—it is meant to be true of all possible legal systems that the law is source-based.Footnote 57 The argument from authority can vindicate this claim only if it shows that all possible legal systems are capable of possessing authority, and we cannot infer that they are if there are possible legal systems in which officials are systematically confused about the concept of authority.
Nor can the objection be overcome by appealing to the idea of a division of linguistic labor. It might be argued that the rest of society defers to legal officials regarding how the concept of authority is to be used, perhaps because of “the centrality of legal institutions in our structures of authority.”Footnote 58 If so, the operative beliefs of legal officials will not diverge from the operative beliefs of the rest of society, since the latter will track the former. However, the existence of a division of linguistic labor is a contingent fact about the practices of a particular language community,Footnote 59 and so cannot be used to show that there is no possible legal system in which officials are systematically confused about the concept of authority.
Raz could instead deal with this objection in the following way. If legal officials use different criteria for applying the concept of authority from the rest of society, this does not show that they are confused about the concept of authority they share with the rest of society. Rather, it shows that they have a different concept of authority from the rest of society. Moreover, they cannot be systematically confused about their concept of authority (at least not at the level of operative beliefs, because the operative beliefs they generally share constitute the correct criteria for applying that concept). Thus it seems that the supporting argument can be based on Raz's criterialism, after all.
D. The Price of Recognizing Multiple Concepts of Authority
We have seen that recognizing multiple concepts of authority provides Raz with a response to two objections to his criterialism and to an objection to the criterialist reading of the supporting argument. However, it also poses both a direct and an indirect threat to the argument from authority.
I shall start with the direct threat. Once we recognize the existence of multiple concepts of authority, there is no reason to think that all legal systems must be capable of satisfying the requirements for possessing authority laid down by our concept of authority (i.e., the concept regarding which Raz's account is correct). It may be that legal officials cannot be systematically confused about their concept of authority, for the reasons given above, but their concept may differ from ours. If so, when they claim that their legal system possesses authority, they are using their concept of authority, not ours. Their claim may license the inference that their legal system must be capable of satisfying the requirements for possessing authority laid down by their concept of authority. However, it gives us no reason to think that their legal system must be capable of satisfying the requirements laid down by our concept.
This is fatal to the argument from authority. Raz needs it to be the case that all legal systems are capable of satisfying the requirements for possessing authority laid down by our concept of authority. We saw in Section I that it is crucial to the argument from authority that we can infer from the fact that the law must be capable of possessing authority that it must satisfy the two prerequisites. On our concept of authority, this inference is sound: the two prerequisites are derived from the key theses which form part of our concept of authority, and those prerequisites must be satisfied in order for legal directives to be capable of possessing authority. However, the key theses may not be correct according to other concepts of authority, in which case the two prerequisites cannot be derived from those concepts. On those concepts of authority, we cannot infer from the fact that the law must be capable of possessing authority that it must satisfy the two prerequisites.
Let me put the point another way. Once Raz recognizes multiple concepts of authority, there is no reason to think that his account of authority—which, we are now assuming, is the correct account of one particular concept of authority—can be used to understand better the nature of law. Even if his account is the correct account of our concept of authority (and remember that Raz queries whether there is any one concept of authority that is “ours”), why think that the law—not just in our legal system, but everywhere—must conform to the requirements of that account?
Raz might respond that there is sufficient overlap between all of the concepts of authority recognized by his criterialism for the argument from authority to go through. We saw in Section I that either of the two prerequisites can be used to show that exclusive legal positivism is correct.Footnote 60 We also saw that the first prerequisite is derived from the dependence thesis, whereas the second prerequisite can be derived from either the normal justification thesis or the preemption thesis. Thus Raz only needs it to be the case that for each of the concepts of authority recognized by his criterialism, at least one of the key theses is true. From that thesis, he can derive one of the two prerequisites and hence show that exclusive legal positivism is correct.
However, why think that this requirement is satisfied? Raz states that there must be similarities between the various concepts of authority in order for them all to be concepts of authority.Footnote 61 However, it is doubtful whether those similarities are such that all concepts of authority accept at least one of the key theses. We have seen that Raz treats the key theses as providing only part of an explanation of our concept of authority. The required similarities between concepts of authority could, therefore, be found in the other parts of that explanation.
Further, we have seen that many people are ignorant of the key theses and that there is widespread disagreement about the merits of the normal justification and preemption theses among those who are aware of them. Although Raz could argue that much of the controversy and ignorance surrounding these theses exists only at the level of manifest beliefs, he offers no reason to think that one of these theses must be true according to every concept of authority recognized by his criterialism.Footnote 62 Indeed, I suggested in the previous subsection that one reason for Raz to recognize multiple concepts of authority is that there is—or at least could be—widespread disagreement, even at the level of operative beliefs, about the normal justification and preemption theses. This suggests that there are concepts of authority according to which neither thesis is correct.
What about the dependence thesis? It is true that Raz's critics generally target the preemption and normal justification theses rather than the dependence thesis. However, Raz regards all three theses as closely interrelated.Footnote 63 Thus it is unlikely that the dependence thesis is true according to a particular concept of authority if neither of the other key theses is. In sum, it is likely that there are concepts of authority recognized by Raz's criterialism that reject all of the key theses. If there is a possible legal system in which officials hold such a concept, then the argument from authority is undermined.
There is an alternative response one might make to the objection to the argument from authority presented above. One might argue that the objection wrongly presupposes that officials use their concept of authority when they claim that the law possesses authority. This presupposition might be warranted if Raz claimed that legal officials explicitly assert that their legal system possesses authority, since officials would make that assertion using their own concepts. However, the response continues, this is not Raz's view. He bases his contention that the law necessarily claims authority not on explicit assertions by officials that the law possesses authority but rather on what we might call “the authority facts”—the fact that officials regard themselves as having the right to impose obligations on their subjects, the fact that they treat their subjects as being obligated to obey the law, and so on.Footnote 64 In other words, it is Raz—not legal officials—who employs a concept of authority when he infers from the authority facts that officials attribute the possession of (legitimate) authority to the law. Moreover, the concept of authority Raz employs is ours, not the officials’. It does not matter, therefore, that officials may hold a different concept of authority from ours, because it is not their concept that is in play.
It is true that Raz does not base his contention that the law necessarily claims authority solely on explicit assertions by officials that the law possesses authority. However, his contention is based in part on such assertions, since among the authority facts is “the fact that legal institutions are officially designated as ‘authorities.’”Footnote 65 In designating legal institutions as authorities, officials use their concept of authority, not ours.
Even if we confine our attention to authority facts other than the one mentioned in the previous paragraph, the response still fails. Raz would need to show that the property that officials attribute to the law is the property picked out by our concept of authority (i.e., the property described by the key theses). Only then can we infer that the law must be capable of satisfying the two prerequisites. However, given Raz's recognition of multiple concepts of authority, it seems unlikely that officials in all legal systems attribute to the law the property picked out by our concept of authority. This impression is confirmed when we consider the authority facts more carefully. While those facts may show that officials attribute to the law the property picked out by some concept of authority, they fall well short of showing that officials attribute to the law the property picked out by our concept. For example, the fact that officials claim that their subjects owe them allegiance does not show that they implicitly regard the law as satisfying the normal justification thesis. Similarly, the fact that officials regard themselves as having the right to impose obligations on their subjects does not show that they implicitly treat those obligations as preemptive in nature.Footnote 66 Thus we cannot infer from these authority facts that officials ascribe to the law what we call authority as opposed to what someone else—using a different concept of authority from ours—calls authority.Footnote 67
I have argued that Raz's recognition of multiple concepts of authority is fatal to the argument from authority. In addition, there is a less direct way in which recognizing multiple concepts of authority undermines that argument. On the criterialist reading, the supporting argument is sound only if Raz's criterialism is sound, since Raz's criterialism is used to explain why legal officials cannot be systematically confused about the concept of authority. If Raz's criterialism should be rejected, so should the supporting argument.
While recognizing multiple concepts of authority saves Raz's criterialism from the objections considered in the previous subsection, I shall argue that it renders Raz's criterialism less attractive than many of its contemporary rivals. I shall seek to illustrate this point by comparing Raz's criterialism with Nicos Stavropoulos's semantic realism.Footnote 68 However, my discussion shall be brief and exploratory. This is partly because of constraints of space and partly because I have already argued that—irrespective of any problems with Raz's criterialism—his recognition of multiple concepts of authority is fatal to the argument from authority.
In comparing Raz's criterialism with semantic realism, it is tempting to begin with the powerful arguments as to why the latter is preferable with regard to proper names and natural kind terms.Footnote 69 However, it is controversial whether those arguments apply with equal force to other terms such as “authority.” Thus, even if it can be shown that semantic realism is the preferable account of proper names and natural kind terms, it is open to Raz to deny that it is the preferable account of concepts such as authority.Footnote 70 Moreover, for the supporting argument to succeed, Raz only needs his criterialism to be true of concepts such as authority. For these reasons, when comparing Raz's criterialism with semantic realism, I shall focus primarily on concepts such as authority.
In making this comparison, we are concerned with which theory is preferable, and there is no uncontroversial account of what makes one philosophical theory preferable to another. However, it is widely accepted that, everything else being equal, a simple theory is preferable to a complex one. What simplicity consists in is also contested. However, one plausible measure of simplicity is the following: the more additional hypotheses that have to be added to the central claims of a theory in order to account for phenomena that the theory would otherwise be unable to explain, the less simple the theory.Footnote 71 On this measure, Raz's criterialism appears more complex than semantic realism. In addition to the central thesis of his criterialism (that “a true explanation” of a concept “is a correct statement of the conditions for the correct use of the concept actually used by those who use it”),Footnote 72 Raz needs an additional hypothesis (that there are multiple concepts of authority) to explain what appears to be widespread disagreement and error concerning the criteria for applying the concept of authority.Footnote 73
By contrast, semantic realism does not need an additional hypothesis to explain widespread disagreement and error; the explanation of these phenomena falls out of the central claims of this theory. The catchcry of semantic realism is that meaning “ain't in the head,”Footnote 74 since the meaning of our words depends to a large extent on how things are in the external world. One consequence of this is that what is “in the head” may be mistaken.
It might be objected that semantic realism can only explain disagreement and error in our manifest beliefs about authority. It entails that what is “in the head” may be mistaken, but this seems to correspond to our manifest, not operative, beliefs. Moreover, we have seen that Raz's criterialism can also explain widespread disagreement in our manifest beliefs without needing to posit the existence of multiple concepts of authority; this additional hypothesis is needed only to explain widespread disagreement in our operative beliefs.
However, when the semantic realist says that meaning “ain't in the head,” she is not just saying that the extension of our terms is not fixed by our consciously held beliefs. In the case of a natural kind term like “water,” the extension of the term is not fixed by the criteria for applying the term that we use in practice either. Rather, it is fixed by the nature of the object that the term picks out. To take Putnam's famous example, if we discover that “water” on Twin Earth has a different chemical composition from water on Earth (despite the fact that they share the superficial properties of being transparent, odorless, drinkable liquids), we are likely to cease using our word “water” to refer to the substance on Twin Earth. We might even revise the criteria we employ to determine when to use the word “water” by focusing more on the chemical composition of purported samples of water and less on their superficial properties.
Stavropoulos claims that something similar is true of legal concepts (and, mutatis mutandis, he would extend that claim to the concept of authority). He contends that lawyers’ use of a legal concept is governed by theoretical considerations—that is, “by substantive claims concerning the property that collects together what counts as paradigmatic applications of the relevant concept.”Footnote 75 The best explanation of legal practice, then, is that the extension of a legal concept is fixed by reference to the correct understanding of that property (which the substantive claims seek to identify). That understanding need not match the substantive claims that any particular lawyer makes about the concept, nor need it match the way in which lawyers generally use the concept. Once we look beyond paradigmatic applications of the concept, lawyers may not generally use the concept in the same way. Moreover, the correct understanding of the relevant property may reveal even some of the paradigmatic applications to be mistaken. It is not only our consciously held beliefs about a concept that are always potentially open to revision but also the way in which we use that concept in practice.Footnote 76
Admittedly, simplicity is only one virtue of a philosophical theory. Another widely accepted virtue is explanatory power. Again, how explanatory power is to be understood is controversial, but—very roughly—it is concerned with how much of the relevant phenomena a theory can explain and the importance of the phenomena it can explain.Footnote 77 The complexity involved in recognizing multiple concepts of authority might be a price worth paying if Raz's criterialism had greater explanatory power than rival theories. However, I shall suggest that there is at least one respect in which it has less explanatory power than semantic realism.
Once Raz recognizes the existence of multiple concepts of authority, he needs to explain how successful communication can and does occur between people who hold different concepts of authority. Two people who talk about “authority” but who hold different concepts of authority appear to be talking at cross-purposes. If A is asserting that X does or does not instantiate A's concept of authority, and B is asserting that X does or does not instantiate B's concept of authority, then they are talking about different things and so communicate neither agreement nor disagreement. Given the plethora of different concepts of authority that Raz recognizes, this would represent a significant impediment to successful communication (especially if there is no single concept of authority that is “ours,” which Raz suggests is quite possible). Yet this impediment does not seem to exist in reality.Footnote 78
This problem cannot be overcome by pointing out that we share the umbrella concept of authority. When two people who hold different (non-umbrella) concepts of authority talk about “authority,” it is not normally the umbrella concept they are invoking.Footnote 79 Moreover, while Raz suggests that Jesus and Genghis Khan disagreed about many issues even if there were few concepts they shared, he himself distinguishes this from the question of whether two people can successfully communicate their disagreement in circumstances where they do not share the relevant concepts.Footnote 80 Yet once Raz recognizes the existence of multiple concepts of authority, he needs to be able to answer that question in the affirmative or else accept that we may often fail to communicate successfully when we talk about “authority.”
We have seen that Raz states that there are similarities between the various concepts of authority. It might be argued that these similarities ensure that the holders of different concepts of authority are able to communicate successfully.Footnote 81 However, if communication between holders of different concepts of authority is to be explained on this basis, some pressing questions need to be answered. First, how does the fact that two people's concepts of authority are similar in some—even many—respects enable them to communicate successfully, given that the concepts also differ in certain respects? Second, and relatedly, what is the relevant standard of similarity? Any two concepts are similar in some respects, yet this fact would not enable two people to communicate successfully where, say, one of them was talking about riverbanks and the other about offices of a financial institution.Footnote 82 So how does this fact enable holders of different concepts of authority to communicate successfully? The obvious answer is that there is a greater degree of similarity between the two concepts in the latter case than in the former. However, further questions arise. What degree of similarity is needed? How does that degree of similarity enable successful communication? And do the various concepts of authority possess the required degree of similarity? Even if these challenges can be met, doing so is likely to exacerbate the problem discussed above—namely, that Raz's criterialism fares poorly on the measure of simplicity.
Alternatively, it might be argued that provided people realize that they hold different concepts of authority, they can overcome any potential difficulties in communication by agreeing that for the purposes of their discussion they will all use the same concept of authority (let us call it the “chosen” concept).Footnote 83 However, this solution could work only among very well-informed discussants. Not only would they need to be aware that they held different concepts of authority, but they would need to have sufficient understanding of both their own concept and the chosen concept to be able successfully to translate statements employing their own concept into statements employing the chosen concept.
This is important because it is not enough for Raz to explain how communication is possible between holders of different concepts of authority. He also needs to explain actual instances of apparently successful communication between people who, according to his criterialism, hold different concepts of authority. The fact that the solution suggested in the previous paragraph could work only among very well-informed discussants suggests that it cannot explain many actual instances of communication. More generally, there is an absence of evidence that people overcome obstacles to successful communication in the way being suggested. Indeed, to the casual observer, most discussions of authority proceed in ignorance of the (alleged) fact that there are multiple concepts of authority.Footnote 84
This is, of course, only a partial comparison of Raz's criterialism with semantic realism. There may be other phenomena that semantic realism cannot explain, or can explain only via the introduction of additional hypotheses, but which Raz's criterialism can explain without the need for additional hypotheses.Footnote 85 Nevertheless, the preceding discussion gives us some reason to think that semantic realism represents a significant advance on Raz's criterialism. Semantic realism enables us to say that two groups of people who fundamentally disagree about the key theses disagree about the correct account of the same concept of authority (and so can communicate their disagreement successfully). It also enables us to say that even if our ancestors believed that authority must have a divine source, they nevertheless shared with us the concept of authority (though we obtained a better understanding of that concept when we realized that this belief was mistaken). However, in doing so, semantic realism suggests that we may be systematically confused about the concept of authority (not only in our consciously held beliefs about the concept, but also in the criteria—if any—that we generally employ when using the concept). Assuming that the same may be true of legal officials, semantic realism requires us to reject the supporting argument.
IV. IS THE SUPPORTING ARGUMENT SOUND? THE MINIMALIST READING
In Section III, we saw that Raz's criterialism has the potential to vindicate the supporting argument because it entails that officials’ operative beliefs about the concept of authority cannot be systematically confused. However, to meet various objections, Raz needs to (and does) recognize the existence of multiple concepts of authority. Doing so gives rise to new problems. First, it becomes implausible to suggest that all legal systems must be capable of satisfying the requirements for possessing authority laid down by our concept of authority. Second, it contributes to the impression that Raz's criterialism—which provides the basis for the supporting argument—is more complex and has less explanatory power than rival theories such as semantic realism.
At this point, one might question whether the criterialist reading offers the most charitable interpretation of the supporting argument. If interpreting that argument in light of Raz's criterialism gives rise to these problems, perhaps we should seek an alternative interpretation. This thought is supported by the fact, already mentioned, that Raz does not explicitly refer to his criterialism when presenting the supporting argument. In the present section, therefore, I consider what I call the “minimalist” reading of the supporting argument, which seeks to understand that argument on its own terms without drawing on Raz's criterialism.Footnote 86
Recall that Raz contends that the claim that the law possesses authority cannot be based on a conceptual mistake because “the claim is made by legal officials wherever a legal system is in force.” He further contends that legal officials cannot be systematically confused about the concept of authority because of “the centrality of legal institutions in our structures of authority,” which means that “their claims and conceptions are formed by and contribute to our concept of authority.”Footnote 87
Let us start with the first of these two contentions. Is that contention meant to be sufficient by itself to show that the law must be capable of possessing authority? This seems to be how Wil Waluchow reads it. He interprets Raz as inferring that the law must be capable of possessing authority from the fact that all legal systems claim authority, since “It would be wildly implausible to suppose that all legal systems are conceptually confused about their capacity to command authority, claiming to have that capacity when they are the kinds of things, like trees and planets, of which this is logically impossible.”Footnote 88 This seems to treat the fact that the law necessarily claims authority as a reason in itself for thinking that the claim cannot be based on a conceptual mistake (and therefore that the law must be capable of possessing authority).
I do not think that this is what Raz has in mind, in part because the argument Waluchow attributes to him appears obviously flawed. Assume, for the sake of the argument, that it cannot be true of all possible legal systems that the claim to authority is the result of conceptual confusion. All that follows is that there is at least one possible legal system in which that claim is not the result of conceptual confusion. It does not follow that there is no possible legal system in which that claim is the result of conceptual confusion. Yet Raz needs to show that there is no such legal system in order to show that all possible legal systems are capable of possessing authority.
It may be, however, that Waluchow interprets Raz as assuming that if any legal system is incapable of possessing authority, then all legal systems must be incapable of doing so, because the incapacity must be due to an essential property of all legal systems (e.g., that legal systems, like trees and planets, are incapable of issuing directives). On this assumption, it cannot be the case that the claim to authority is the result of conceptual confusion in some legal systems but not others. Either all legal officials are confused (because no legal system is capable of possessing authority) or none are (because all legal systems are capable of possessing authority). The former is implausible, so the latter must be the case.
However, Raz does not argue that if any legal system is incapable of possessing authority, then all legal systems must be incapable of doing so. Nor can he simply assume that this is the case. At this stage in the argument from authority, it is still an open question whether a legal system can require one to engage in moral reasoning to determine the content of the law. (Raz must first establish that all legal systems are capable of possessing authority and that this means that they must satisfy the two prerequisites before he can conclude that no legal system can require one to engage in moral reasoning to determine the content of the law.) If a legal system did require one to engage in such reasoning, it would—on Raz's account—be incapable of possessing authority, but this would be due to a property specific to that legal system. Its officials would be confused when they claimed that the law possesses authority, but this would not entail that the officials of all other legal systems are similarly confused.
For these reasons, I do not interpret the first contention as asserting that the fact that the law necessarily claims authority is sufficient to show that the claim cannot be based on a conceptual mistake and hence that the law must be capable of possessing authority. Rather, Raz recognizes that further argument is needed, and the second contention seeks to provide that argument. So let us turn to the second contention.
Raz argues that legal officials cannot be systematically confused about the concept of authority, because their claims and conceptions contribute to that concept. For this argument to be sound, officials’ claims and conceptions must play a large enough role in constituting the concept of authority as to ensure that officials cannot be systematically confused about the concept. If we leave Raz's criterialism to one side, he gives us little reason to think that this is so. He does state that officials’ claims and conceptions contribute to the concept of authority because of “the centrality of legal institutions in our structures of authority.” However, what does this phrase mean? If the claim is that legal institutions represent a socially important instantiation of (purportedly legitimate) authority, then this is insufficient to show that officials’ claims play the requisite role with regard to the concept of authority. It may suggest that their claims about authority are particularly influential in our society, since it is legal officials who speak on behalf of legal institutions. However, this falls well short of showing that their claims play such a large role in constituting the concept of authority as to ensure that officials cannot be systematically confused about that concept.
Moreover, even if legal institutions represent a socially important instantiation of (purportedly legitimate) authority, this is a contingent fact about particular societies. (In some societies, legal—as opposed to, say, religious or familial—institutions may play a relatively minor role.) If so, this fact cannot be used to show that it is true of every legal system that officials’ claims and conceptions contribute to the concept of authority.
Perhaps we can overcome these problems by drawing on Raz's thesis that explaining concepts like authority involves explaining our self-understanding. He states:
the way a culture understands its own practices and institutions is not separate from what they are . . . they—the practices and the institutions—are in part shaped by the way they are understood by the people whose practices and institutions they are. They are manifestations of the self-understanding of their culture.Footnote 89
Given the role that the concept of authority plays in the practices and institutions of our culture, it may be that the content of that concept is not separate from our understanding of it.Footnote 90 Moreover, officials’ understanding of authority may be especially important given that they issue and apply legal directives, which arguably represent a particularly important type of authoritative directive in our society.Footnote 91
Again, however, if legal directives represent a particularly important type of authoritative directive, this is a contingent fact about particular societies. Perhaps this does not matter. Perhaps all Raz needs is that explaining the concept of authority involves explaining our—not just or especially officials’—self-understanding. However, if explaining the concept of authority involves explaining our self-understanding, then the concept of authority in question cannot be the one possessed by other societies that have different self-understandings. In other words, the thesis that explaining the concept of authority involves explaining a society's self-understanding entails that there are multiple concepts of authority, since different societies have different self-understandings and hence different concepts.Footnote 92 Yet we saw in Subsection III.D that recognizing multiple concepts of authority is fatal to the argument from authority because there is no reason to think that all legal systems must be capable of satisfying the requirements for possessing authority laid down by our concept.
Finally, recall that one of the initial objections to the criterialist reading of the supporting argument was that Raz's claim that legal officials cannot be systematically confused about the concept of authority sits uneasily with the assumption that his account of authority is correct. Given that many legal officials are unfamiliar with Raz's account (as he himself concedes), and given that this account represents part of the correct explanation of the concept of authority, how can there not be systematic confusion among officials about that concept? This question is as pressing on the minimalist reading of the supporting argument as it is on the criterialist reading.
Indeed, it is more pressing. On the criterialist reading, Raz has a response to this objection—namely, that legal officials may not be consciously aware of his account but may nevertheless use the concept of authority in a way that accords with that account. Once we abandon the criterialist reading, he can no longer offer that response, since it draws on the claim, central to the criterialist reading, that the correct criteria for applying a concept are our generally shared operative beliefs. (Admittedly, I argued that Raz's response breaks down once he recognizes multiple concepts of authority. However, on the criterialist reading, he at least has an initial response to the objection; on the minimalist reading, he does not.)
V. CONCLUSION
I have argued that on both the criterialist and minimalist readings, the supporting argument fails. On the criterialist reading, Raz's recognition of multiple concepts of authority leaves us with no reason to think that all legal systems must be capable of satisfying the requirements for possessing authority laid down by our concept of authority. Moreover, Raz's criterialism—on which the supporting argument is based, according to the criterialist reading—appears less attractive than rival theories such as semantic realism.
Alternatively, if we leave Raz's criterialism to one side, the brief remarks that constitute the supporting argument provide little reason to believe that legal officials cannot be systematically confused about the concept of authority. Indeed, there is reason to believe that they can be systematically confused about that concept, since Raz's account of authority (which he claims is part of a correct explanation of the concept) is unfamiliar to many legal officials.
If legal officials can be systematically confused about the concept of authority, then we cannot infer from the fact that the law necessarily claims authority that it must be capable of possessing authority. If the law need not be capable of possessing authority, then it need not satisfy the two prerequisites. And if the law need not satisfy the two prerequisites, then the argument from authority collapses, since it no longer matters (for the purpose of better understanding the nature of law) that only exclusive legal positivism is consistent with the two prerequisites.
It follows that the case for exclusive legal positivism is not as strong as it may have appeared. Admittedly, the argument from authority is not the only argument Raz offers in favor of exclusive legal positivism,Footnote 93 and other exclusive legal positivists offer further arguments in favor of their position.Footnote 94 Nevertheless, the argument from authority is by far the most influential argument in favor of exclusive legal positivism. If I am right, that argument is unsound.
