I. Introduction

In November of 2019 Thomas Bustamante hosted a colloquium on my book, Judging Positivism (henceforth JP), at a conference held by the Brazilian Association for Political Philosophy and Constitutional Law at the University of São Paulo. The papers presented were eloquent and insightful. I remain grateful for the time and effort put forth by each participant. I would like to extend my gratitude to Thomas Bustamante for organizing the colloquium and for his excellent editorial work. In what follows, I offer a reply to three of my critics: André Coelho, Jorge Fabra-Zamora, and Thomas Bustamante. Their probing contributions are included in this issue of Isonomía.

In JP, I argue that Raz has changed his position over time and that these changes have produced inconsistencies and incoherencies in his later works. The key claim that sets the argument in motion is that Raz’s account of legality in Practical Reasons and Norms (^{Raz, 1999}, henceforth PRN) is grounded in, and dependent on, the claim that judges have a duty to apply the law. While problems remain for Raz if he distances himself from PRN, it is nevertheless very useful to return to this early work: we can see how his theses about the nature of law and the nature of adjudication, which are now considered wholly separate, were once mutually reinforcing. By returning to this early work, the worries I track in JP are brought into focus. Problems arise for Raz, I argue, when he introduces moral elements into his positivistic account of law.

For instance, in chapter three of my book, I argue that Raz’s morally robust theory of adjudication fits uneasily with his positivistic theses about the nature of law. Raz, I argue, champions a thesis about legal rights that is distinctively Holmesian. André Coelho focuses his commentary on this line of argument. He claims that I have erred in my interpretation of PRN: Raz’s account of law and adjudication, Coelho argues, is best read as offering us a hypothetical account of law from the perspective of authority. In my reply, I revisit PRN and, in addition, I explore some of the assumptions about the nature of philosophy more generally. Coelho’s eloquent piece prompts a return to first principles. While I disagree with Coelho, significant common ground emerges in the course of the analysis.

My second critic, Thomas Bustamante, focuses on my critique of Razian authority. My central claim in JP is that Raz’s account is marred by a permanent tension between his two central theses, the normal justification thesis (NJT) and the pre-emption thesis (PT). Bustamante wonders whether this tension can be eased, at least in part, by turning to the idea of jurisdiction (as Raz does in The Morality of Freedom) (^{Raz, 1986}, henceforth MF). Bustamante introduces two thought provoking examples in order to explore this possibility. Bustamante’s central test case is both timely and complex: he asks readers to contemplate the impeachment of Brazilian President Dilma Rousseff. Bustamante notes that many commentators, including most academic lawyers and some judges, agree that the impeachment was illegal, but they also agree that the decision should not be revisited. I explore the possibility that this case study can help Raz, but I ultimately resist Bustamante’s conclusions. The best arguments available to Raz have serious consequences for his position: Raz becomes either a Benthamite positivist or a traditional natural lawyer. Bustamante’s discussion is enlightening, but I doubt Raz’s taxonomy adds anything to his astute analysis.

Finally, I explore the detailed critique of JP offered by Jorge Fabora-Zamora. He presents readers with the positivists’ response, par excellence. From the perspective of positivists, my analysis in JP is confused. I do not place great weight on the distinctions and argumentative moves that they cherish. In my reply, I seek to illustrate that doing so would make little sense since I reject the positivists’ methodological starting point. That is, I reject the possibility of non-normative conceptual analysis. I argue that even if we begin with a question about the existence conditions for legality (as Fabra-Zamora does), we cannot answer this question without offering moral and political arguments. Moreover, adopting the “detached” perspective cannot save Raz because the questions themselves require political answers. I begin my reply by returning, once again, to PRN. I reject the possibility, championed by Fabra-Zamora, that PRN can be interpreted in a manner consistent with the moral vision of adjudication offered in Ethics in the Public Domain (^{Raz, 1994}, henceforth EPD).

II. Choosing a Flashlight

André Coelho’s critique takes flight with a metaphor: “I see legal theories as the blind men of the Indian parable, conceptualizing of what an elephant is by having each of them touching a different part of its giant, polymorphous body” (^{Coelho, 2021}, p. 169). A theory can only illuminate part of the object of interest – it always and inevitably casts long shadows on other features of the entity under investigation. Coelho contends that I have failed to see Raz’s theory in this light, and as a result, I have asked too much of him. In JP, I assume that Raz is attempting to explain the entirety of the elephant, but according to Coelho, Raz offers readers a theory that is “all legs” (^{Coelho, 2021}, p. 172). What does it mean to say that Raz is offering an account that is “all legs”?

Coelho argues that Raz’s theory is best understood as a hypothetical perspective of authority. Raz is not talking about “real, concrete social practices,” nor is he talking about “normative, idealized ones” (^{Coelho, 2021}, p. 172). Instead, “We are talking of highly stylized practices that are real in the authority-based world, which means that, compared to the real, concrete practices of our world, they capture at most the aspects of these practices that are authority-based” (^{Coelho, 2021}, p. 173). In JP, I attribute to Raz an inconsistency between, on the one hand, his early view that judges have a duty to apply the law, and, on the other hand, his mature view that proper judicial reasoning is a species of moral reasoning. If, however, PRN is viewed as a hypothetical account of what law would look like if we lived in an authority-based world, the inconsistency disappears: we would not expect a hypothetical account to be consistent with his moral recommendations about what judges should do.

At this juncture, it is useful to note that Coelho’s critique contains a degree of ambiguity. At times, he seems to suggest that Raz presents his account in PRN as merely hypothetical. But at other times, Coelho claims that Raz’s account is limited because philosophy itself is a limited enterprise. He argues that Raz is not the only legal philosopher who is offering so-called leg-based accounts: “Hart’s convention-based, Dworkin’s integrity-based and Finnis’s flourishing-based conceptions of law are also leg-based theories of elephants” (^{Coelho, 2021}, p. 172). My failure, from this perspective, amounts to a failure to grasp the limitations inherent in philosophical reflection. It is possible that Coelho is making both claims, but these two lines of argumentation are, nevertheless, distinct.

In what follows, I will not engage in full with Coelho’s claims about the nature of philosophy, yet some brief comments are useful to set up my reply. I agree with Coelho that no single theory will illuminate the whole, but I am reluctant to agree with him about the vision of philosophy on offer. Coelho views philosophy as a very limited enterprise: the concepts which we construct to make sense of things are the stuff of philosophy and it is only the sociologist, but not the philosopher, who has access to the empirical world (^{Coelho, 2021, p. 171}). Philosophy, he insists, is self-referential and self-contained – we only have our subjective intuitions to anchor it (^{Coelho, 2021}, p. 171).¹ According to Coelho, we cannot hope to explain the whole (and must settle for the legs) because our ideas are always mediated by the concepts we construct (^{Coelho, 2021}, p. 171).

It is, of course, not a fact that philosophy is limited in the manner that Coelho supposes. The nature of philosophy is itself a matter of philosophical debate. If, however, we assume that he is right about the limits of the discipline, questions can be raised about the point of philosophical debate. Is it the case that we can only aim at a “leg-based” theory? Should we presuppose that all accounts are equally illuminating, albeit in this limited fashion? In addition, I am not certain that Coelho adheres to the limitations he articulates. After all, how can we identify the legs of the elephant if we do not have some access to the whole? Or, put differently, how do we determine if there is a gap between theory and practice (as both Coelho and I assume) if we are not also assuming that practice can be understood as existing outside of our theoretical conceptions of it?

Note that my critique of PRN in JP does rely upon a clear conception of the whole. Instead, I simply illustrate that Raz’s account fails to accommodate a significant element of legal practice, namely, common law reasoning. Tellingly, both Coelho and I claim that there is a significant gap between Raz’s account in PRN and the legal world as we experience it. The question becomes: what are we to make of this gap? Coelho insists the gap is the result of the adoption of the “hypothetical” perspective of authority. Conversely, I argue that the existence of this gap signals a failed attempt, on Raz’s part, to describe what judges do.

Perhaps Coelho is friendlier to Raz because he sees a grain of truth in Raz’s theses. Consider the following passage, wherein Coelho finds a place for Raz’s two different (and, in my opinion, inconsistent) views about adjudication:

In the case of a legal rule that, applied to a certain case, would give rise to great injustice, it would make sense to say both that the judge as judge, assuming the legal point of view, should apply that rule nevertheless; and that the judge, as a moral reasoner, should not, all things considered, apply that rule (^{Coelho, 2021}, p. 175; my italics).

Both of the views canvassed here are found in Raz’s writings. In PRN, Raz tells us that judges make decisions from the “legal point of view” (^{Raz, 1999}, p. 139); in EPD, the claim is that judges are “moral reasoners” (^{Raz, 1994}, p. 338). Coelho takes what he sees as two partial truths and weaves them together and offers an account that better captures the behaviour of judges. While I prefer Coelho’s account to either version of adjudication Raz gives us, notice that there are methodological implications that follow from the passage above: Coelho is committed to the view that some accounts are better than others and we can figure this out by looking at how the views in question map onto practice. There is more common ground between us than Coelho supposes. Our methodological assumptions appear to be very similar, if not identical.

Once it is apparent that both Coelho and I are looking at the practice of adjudication, and we both identify a gap between PRN and legal practice as we grasp it, the interpretive question I identified at the outset comes into view. In PRN, is Raz offering a descriptive theory of adjudication that is vulnerable to empirical counterexamples as I contend? Or is PRN meant to give readers a stylized, hypothetical account as Coelho claims? That is, is Raz trying to understand legal authority in the world, or is he offering a vision of law that is meant to represent the hypothetical perspective of legal authority if the world were structured by his vision of authority? In the process of making a case for the first option, I will also address Coelho’s claim that my reading of Raz’s work fails to “distinguish a leg-based theory of elephants from a normative theory of elephants” (^{Coelho, 2021}, p. 172). According to Coelho, the confusion is rooted in my failure to grasp that PRN is a discussion of the “practice of identifying rights is described as it would be in an authority-based world” (^{Coelho, 2021}, p. 176). Again, if PRN is an attempt to explain the legal world, the confusion disappears.

In JP I argue that, at the centre of Raz’s account in PRN is the claim that judges have a duty to apply the law (^{Martin, 2014}, pp. 10-16). To be clear, I agree with Coelho that the word “duty” is not defined by Raz as a moral duty. In PRN, Raz believes he can simply describe the practice without making a further commitment. In other words, the theory Raz offers readers is not a moral theory, it is a “description” or “observation” about this duty.² I also agree with Coelho that Raz places a lot of emphasis on the idea of the “legal point of view,” but this does not (in my opinion) capture the perspective of the system as if judges operated with Raz’s vision of legal authority (as Coelho argues). Rather, Raz (wrongly) thinks this is what judges do in most instances.

Consider some key pieces of evidence that support my interpretation. After elucidating his thesis about the work of judges, Raz notes that Common Law jurisdictions where judges can “overrule established precedent” can be regarded as an “overwhelming” objection to his account (^{Raz, 1999}, p. 140). Raz insists that he can respond to this objection because the rules that govern the behaviour of judges bind them. Notice that this is a very different claim from the one Raz had been making up until this point in the discussion : judges, Raz repeatedly tells us, are supposed to apply the very same norms that citizens are supposed to treat as reasons for action (^{Raz, 1999}, p. 139). While I address Raz’s full response at great length in JP, the point to note is that Common Law jurisdictions would not stand out as a potential counterexample to his theory unless he had been insisting that judges have a duty to apply the law in our shared world, and not merely in some hypothetical one. In my view, Raz is very confident that he has explained the nature of adjudication (and, as a result, the nature of law). It is this confidence that helps to explain the shape his argument takes.

Consider an astute observation Coelho makes, meant to support his claim that PRN is best read as a “hypothetical” account. Coelho argues that if Raz is offering a positivistic theory of adjudication that is based on the duty to apply the law, as I contend in JP, then we would surely find a more fulsome discussion of the activity of judging. For instance, Coelho notes that if Raz is offering a positivist account of adjudication, then one would expect to see references to “formalism” and “textualism” (^{Coelho, 2021}, p. 175).In my view, the reason for this omission can be traced back to Raz’s belief that he has unlocked legalities’ secrets with a single descriptive thesis about judicial behaviour.

“Textualism” and “formalism” are types of interpretive methods or approaches commonly championed by those committed to normative positivism. If Raz included them in his discussion, he would be admitting that judges have interpretative options. He would then need to argue that textualism and formalism are better than the alternatives, and this would require him to make normative arguments. In PRN, he does not think he needs a normative theory about what judges ought to do. While he insists judges are not “computing machines,” he assumes that they have a duty to apply the law and, in most cases, this duty is realized (^{Raz, 1999}, p. 139). In short, Raz wrongly assumes there are no options available to judges.

Tellingly, the only option Raz takes seriously is hypothetical: “hypothetical systems of absolute discretion” are introduced in order to cast light on the nature of legality. These imaginary discretion-based systems are very different from legal systems because “they are always to make that decision which they think is best on the basis of all the valid reasons” (^{Raz, 1999}, p. 138). Conversely, legal systems “consist of norms which the courts are bound to apply regardless of their view of their merit” (^{Raz, 1999}, p. 139). The perspective of the judge in a legal system is “partial” (I agree with Coelho on this point) but it is not hypothetical.

Again, we learn about the duty to apply the law, which is a defining characteristic of legal systems, by imagining what would happen in a system that lacked this duty. Unlike in these imaginary systems, legal systems do provide guidance to individuals precisely because they are under a duty to apply the law (^{Raz, 1999}, p. 138). Judges in legal systems maintain the authoritative status of a pre-existing body of norms, and these are the norms that guide the behaviour of law’s subjects. Raz makes it clear that the evaluation of the behaviour of law’s subjects is “based on the very same norms which guide behaviour” (^{Raz, 1999}, p. 139). Consequently, the choice (according to Raz) is between realizing the duty to apply the law, thereby securing rights and providing guidance to law’s subjects, or adjudicating on a case-by-base (discretionary) basis. Because the second option is fictional, we are prompted by Raz to conclude that he is right.

Notice the argumentative strategy at play here. In the imagined world of systems of absolute discretion, we cannot explain how law guides conduct. The law is not the source of our legal rights; nor can we even identify legal norms in the hopes of being provided with guidance (^{Raz, 1999}, p. 139). In short, systems of absolute discretion are imaginary because they do not work. If the system we live in work, even to a degree, this is evidence that law is governing as it should – it is evidence that the judges are carrying out their duty as Raz conceives of it. But this is, of course, a false dichotomy. Thus, I agree with Coelho that the account is implausible as an account of the world. But if Raz intended his readers to interpret his account of law and adjudication as hypothetical then it is quite peculiar that he would turn to a hypothetical example to shore up a hypothetical account of law.

This leads me to the penultimate criticism that I will address here. Coelho believes my argument would only work if Raz had defended a moral theory in PRN (^{Coelho, 2021}, p. 173). The assumption at play here is a very common one: it is often assumed that normative theories of adjudication and descriptive theories of law inhabit separate spheres. In JP, I reject the assumption that theories of law and theories of adjudication are wholly distinct. I remain deeply skeptical about the viability of the sharp form of this distinction; I am also skeptical about the possibility of non-normative legal philosophy (^{Martin, 2014}; ^{Martin, 2020b}). Tellingly, Raz does not always abide by the distinction that he champions.

Consider the following, lengthy, passage from Between Authority and Interpretation. Raz offers what I take to be a plausible account of both law and adjudication:

The crucial point is that our interest in legal authority lies in how it establishes the moral authority of the law, or of parts of it. We are interested in the authority of law, if any, in order to establish whether we have an obligation to respect and obey it. Moreover, the grounds for the authority of the law help to determine how it ought to be interpreted. Judges, perhaps more than anyone else, follow the law because they believe they are morally required to do so. There can be no other way in which they can justify imprisoning people, interfering with their property, jobs, family relations, and so on, decisions that are the daily faire of judicial life. It may be worth repeating that none of these implies that there is no room for more narrowly focused legal reasoning about whether any institution meets the purely legal conditions for the possession of authority. My claim is only that such an inquiry is of interest because it is embedded in a wider inquiry into the moral legitimacy of that institution’s power” (^{Raz, 2009}, pp. 332-333).

In this passage, the question of law’s authority is linked both to the behaviour of citizens and to the work of judges. Raz has not even attempted to dissect his concept of authority into normative and non-normative segments. It is hard to see how such a division of labour would work in this instance. Not only does Raz fail to respect the sharp distinctions that structure positivist thought, he also places the narrow legal questions in the context of a larger whole. The whole is itself defined by moral considerations, including a concern with the possibility of offering a justification of the use of force by judges.³ Raz’s later views reflect a different conception of jurisprudence, and one that I believe is closer to the mark.

Once we see that Raz fails to align his thoughts with the distinctions he frequently relies upon, it is less surprising, perhaps, that my work would question the viability of these distinctions. After we recognize that the distinction between a moral theory of adjudication and non-normative analysis of law is not a matter of fact, but a contestable assumption, my approach to Raz’s works becomes intelligible. My discussion of rights (which Coelho is critical of ) aims at working out the implications of Raz’s claims from the perspective of the wider whole. It is this discussion to which I will now turn.

In PRN, Raz argues that we have rights because judges have a duty to apply the law (^{Raz, 1999}, p. 138). If judges are now (in Raz’s mature position) understood as moral reasoners rather than law appliers, readers can anticipate a shift in Raz’s conception of legal rights. After all, if Raz holds onto the idea that law is best understood as a pre-existing set of positive norms and, in addition, he also insists that judges are best understood as moral reasoners, then Raz must revise his account of legal rights accordingly. Raz does not disappoint. Once he tells us that judges do and should use their discretion to make good moral decisions, his account of rights shifts as expected. Given that judges are often involved in activities that include “changing and revising legal arrangements” it follows that “quite often people do not have expectations regarding their legal rights” (Raz, 1994, p. 338). Consequently, people “know, or should know, that the law is gappy or that it is liable to change on the issue in dispute” (^{Raz, 1994}, p. 338). We no longer look to the set of pre-existing norms in order to determine the content of our rights. Raz, I argue in JP, begins to look a lot like O.W. Holmes, at least at this juncture (^{Martin, 2014}, p. 55).

Contrary to Coelho’s contention, I do not confuse the “normative” with the “descriptive.” Rather, I track the way in which Raz’s shift in thinking about judging shapes his theses about the nature of law and authority. Again, one of my aims in JP is to illustrate that these two domains – a theory of law and a theory of adjudication – interact. One’s theory of law has implications for one’s theory of adjudication and vice versa. The upshot is that a non-normative account of law is not an option (^{Martin 2014}; ^{Martin 2020b}; ^{Martin 2021}).

In sum, it is my view that PRN is meant to illuminate the entire elephant, which (I argue in JP) it fails to do (^{Martin, 2014}, chapter 1). Even if we hold onto a limited portion of Raz’s account – i.e. that judges occasionally apply the law as Raz suggests – the thesis is very limited and is now re-cast in a normative form. Furthermore, it is unclear that this very limited version of normative positivism needs any of the theoretical theses that Raz offers readers in PRN. At best, PRN teaches us what our world does not look like. The Razian flashlight, in my humble opinion, is broken.

III. Razian Authority: Take Two

Thomas Bustamante’s thought-provoking contribution focuses on Raz’s account of authority, which is the focus of chapter four of JP. Bustamante agrees with my conclusion in chapter three of my book, namely that the pre-emption thesis is only viable in its normative form. Bustamante states this point in a novel and powerful way: Raz “must overcome the burden of demonstrating that all alternative explanations of the character and force of legal reasons are conceptually wrong” (^{Bustamante, 2021a}, p. 185). I agree that this is something Raz cannot do, and as a result, the pre-emption thesis is best understood as a normative thesis.⁴ Bustamante’s exploration of Razian authority proceeds with this amendment in place. (It is a friendly amendment from my perspective, but I suspect Raz’s supporters will not see it as such.)

Bustamante believes, however, that I may very well have overstated my case against Raz’s account of authority found in MF. He is not convinced that the preemption thesis (PT) and the normal justification thesis (NJT) are fundamentally antagonistic, which is an argument I put forward in chapter four of JP. The reason I am skeptical about the compatibility of these two central theses is as follows: the PT requires a pre-commitment to obedience whereas the NJT requires us to evaluate the content of the norm in order to determine if it is binding on us. The NJT demands precisely what the PT prohibits. This is a difficult problem to overcome because Raz clearly states that only norms that possess legitimate authority are pre-emptive “reasons for action” (^{Raz, 1986}, p. 46). This means that we must sacrifice their pre-emptive status in order to attempt to determine if they have such status. Raz also tells us that “for every person it has to be asked afresh, and for every one it has to be asked in a manner that admits of various qualifications” (^{Raz, 1986}, p. 74). We are continually expected to evaluate legal norms based on the NJT to see if they are pre-emptively binding on us (^{Raz, 1986}, p. 78). Consequently, the tension continually reappears. The fact that Raz’s account of authority is both “piecemeal” and agent-relative is to blame. Can Bustamante chart a way out for Raz?

Bustamante draws upon Raz’s appeal to jurisdictional considerations to help make his case (^{Bustamante, 2021a}, p. 185). He wonders whether it is possible, “for instance, to apply the normal justification thesis to conclude that there are certain jurisdictional reasons to attribute to some person or institution an exclusive power to make a judgment on certain issues” (^{Bustamante, 2021a}, p. 185). Bustamante asks us to consider the following example:

Consider the case of presidential impeachment processes. In the 2016 impeachment of President Dilma Rousseff, in Brazil, most lawyers maintained that the Senate’s decision to oust Rousseff from the presidential office was substantially wrong and unfair. Almost all decent academic lawyers in Brazil found that the impeachable offenses were non-existent and that the president committed no wrong that would justify the extreme measure of removing her from a duly elected office (^{Bustamante, 2021a}, p. 186).

He adds that almost “all decent lawyers in Brazil” agreed that the president had not committed a legal wrong (^{Bustamante, 2021a}, p. 186). It is also noteworthy that “very few people argued that these legal mistakes rendered the decision void” (^{Bustamante, 2021a}, p. 186). Does it make sense to say that the legislative decision was “legally wrong while still maintaining that judicial review would be inappropriate?”

When considering this question with Raz in mind, a few complexities are immediately apparent. First, when Raz introduces readers to his NJT, he focuses on the relationship between the norms and individual citizens. In Bustamante’s example, the focus is on the decisions of the officials and the secondary norms that govern their behaviour (which is why the idea of jurisdiction appears to be of assistance). And, instead of a legal norm, we are dealing with a decision not to revisit a prior decision. This is significant because the idea of a pre-emptive reason for action is easier to contemplate when we are dealing with legal rules rather than one-off decisions, at least if we hope to draw general conclusions about the nature of law. In addition, it is not immediately evident how one would apply the NJT to the case at hand. Would it be correct to say that the decision to impeach the president amounted to a breach of the NJT? Or is it best to ask whether the decision not to revisit the original decision is itself in accordance with NJT? Bustamante has the second option in mind. Again, this is because Bustamante is appealing to the idea of jurisdiction to see if he can address the tension between Raz’s two core theses.

Bustamante argues that most of the reasons that prohibit judicial review of the decision to impeach the president are jurisdictional reasons: “When the Brazilian Supreme Court claimed that it lacks legal powers to review the merits of this process, one could perfectly say that the reason for upholding the Senate’s trial is that the Senate made no jurisdictional mistake in the sense of Raz” (^{Bustamante, 2021a}, p. 186). He proceeds to restate this point in Razian terms: “It could be argued, for instance, that the Senate was acting within the limits of its authority and that the normal justification thesis recommends that we recognize this jurisdiction even if we are convinced that the Senate’s decision is incorrect” (^{Bustamante, 2021a}, p. 186). According to this explanation, it is the concept of jurisdiction which eases the tension between the NJT and the PT. Instead of looking towards “right reason,” which is the original standard offered by Raz when he defines the NJT, the NJT now tells us that only jurisdictional mistakes are problematic from the perspective of authority. Given that the mistake is not jurisdictional in this instance, the decision is deemed authoritative (and retains its pre-emptive status). Both the PT and NJT have been respected, at least insofar as the NJT is understood as standing for jurisdictional correctness.

My worry with this formulation is the same worry I express in JP when Raz suggests that we can look to the idea of jurisdiction to address the tension. Specifically, I argue that Raz’s reference to the idea of jurisdiction does not ease the tension between his two theses, rather this new idea wholly displaces the NJT (^{Martin, 2014}, p. 84).

Instead of asking whether the content of the norm (or decision) is in accordance with “right reason,” we now ask whether the officials have the legal authority to draft the rule or render the decision in question. These are two fundamentally different inquiries that lead to two distinct accounts of law’s normative authority. Recall that when Raz elucidates the NJT, he tells us that it requires a case-by-case analysis because the expertise of individuals is relevant to the assessment (^{Raz, 1986}, p. 74). Conversely, a jurisdictionally sound rule is sound for everyone who is subject to it. The introduction of the idea of jurisdiction eradicates the source of the tension between theses precisely because the case-by-case, agent-relative dimensions of the NJT disappear when the concept of “jurisdiction” replaces “right reason.” In other words, this shift in thought produces a wholly distinct theory of law’s authority – a point Bustamante recognizes.

Once Bustamante combines the appeal to jurisdiction (which, according to Bustamante, represents the NJT) and the normative version of Raz’s pre-emption thesis, the position on offer is indistinguishable from Bentham’s. Bustamante makes this point explicit in a footnote: “When interpreted as a normative claim, Raz’s preemption thesis is hardly distinguishable from Jeremy Bentham’s ‘motto of the good citizen,’ which tells us ‘to obey punctually [and] censure freely’” (^{Bustamante, 2021a}, p. 191, note 6).⁵ Interestingly, this is so even if we do not begin with Bustamante’s assumption that the pre-emption thesis is inherently normative. In MF, Raz makes it clear that only norms that meet the NJT have pre-emptive force (^{Raz, 1986}, p. 46), which means that we end up in the same place: if jurisdiction is the key concept that tells us if a law enjoys normative force then it follows that all jurisdictionally sound norms have pre-emptive force.⁶ In sum, the tension between the Raz’s two key theses has been eased, but at a significant cost: the NJT is unrecognizable as it has been displaced by the idea of jurisdiction, and as result, Raz’s position is now indistinguishable from Bentham’s.

There is, however, a second way to think about Bustamante’s example, which comes to light when we consider the following formation of the example:

It might be possible, for instance, to apply the normal justification thesis to conclude that there are certain jurisdictional reasons to attribute to some person or institution an exclusive power to make a judgment on certain issues. (^{Bustamante, 2021a}, p. 185)

In this instance, the idea of jurisdiction does not do all the work. Instead, Bustamante gestures towards a more fundamental principle that informs the analysis when he suggest the “exclusive power” is wielded by Parliament only in reference to “certain issues.” It is this more foundational principle that then allows us to identify the set of issues over which a person or institution will have “exclusive power.” In this version of the example, jurisdiction gives way to the more foundational principle in a set of cases, however, this means that “jurisdictionally sound” decisions that are not captured in this set may still be reviewed by the Supreme Court.

In order to determine which cases fall into this narrow category, we must turn to the normative argument that informs the Supreme Court’s decision not to review the case at hand. Bustamante explains that Brazilian officials, including the Justices, were probably worried about undermining the principle of the separation of powers:

If [a justice of the Supreme Court] believes a decision to annul the trial would produce dangerous systemic effects, leading to a serious degree of politicization of the Judiciary or a severe breach of the separation of powers, inasmuch as other judges might be encouraged to discuss the merits of legislative proposals and political judgments, perhaps the best moral and legal decision would be to rely on Raz’s conception of authority and tolerate any mistake that is not a jurisdictional mistake about the legitimacy of the political institution at stake (^{Bustamante, 2021a}, p. 187).

In this instance, we are dealing with a jurisdictionally sound act of government. The mistake in question – the mistake that deems the decision to impeach the President “illegal” – is not a jurisdictional mistake. But it is not simply the fact that it is a non-jurisdictional error that matters: the reason that the Supreme Court does not opt to review the illegal decision to impeach the president is because there would be a “severe breach of the separation of powers” that would lead to “a serious degree of politicization of the Judiciary.” The class of cases of interest are not simply those where the official acts within their proper jurisdiction, but cases where judicial review would threaten “legitimacy of the political institution at stake.” If we accept this reading of the case study, there is no longer the threat that Raz becomes indistinguishable from Bentham. However, another serious problem emerges.

Jurisdiction is no longer the centre of gravity of the analysis. After all, only a subset of jurisdictionally sound norms will be considered off-limits. Consequently, the NJT now stands for the proposition that we have a reason, independently of the law, to uphold the foundational principles, including the separation of powers. The problem with this possibility is that Raz self-consciously avoids general assessments of the legitimacy of the system – a point Bustamante notes (^{Raz, 1986}, p. 70; ^{Bustamante 2021a}, p. 187). Perhaps Raz resists this kind of argument because it leads him into territory that has been well traversed by natural lawyers.

There is an additional problem with this formulation of the NJT: the tension between the NJT and the PT remains in place. Once the emphasis is placed on the underlying reasons (and not the idea of jurisdiction qua jurisdiction) the following question arises: are officials treating the norm as a pre-emptive reason for action, or does the norm have practical sway because of the power of the underlying reasons? Recall the idea of the pre-emption thesis is that we are not supposed to revisit the reasons for the rule. Raz insists that the rule alone is meant to do the practical work (hence Raz’s account of the “mediating” nature of legal authority) (^{Raz, 1986}, pp. 58-59). Raz makes it clear that “double counting” (counting the rule and the reason behind it) is prohibited in the context of the PT (^{Raz, 1986}, p. 58).⁷ In Bustamante’s case study, it is very hard to separate the “reason behind the rule” from the rule itself. The strength of the underlying principles seems to do all the work, or at least most of the work: all officials see the risks involved if the decision to impeach President Dilma Rousseff was subject to judicial review. If this is so, then we have a situation where the NJT is displacing the PT. While the tension appears to be eradicated, this is because the PT has slipped from view.

Similar problems emerge when we think through Bustamante’s second example: tax law. According to Raz, tax law is usually in accordance with “right reason” and hence it can be justified via the NJT (^{Raz, 1986}, p. 49). But what if the particular tax scheme is morally problematic? Would we still have reason to commit to an unfair scheme and treat it as a reason for action? Bustamante suggests that, at least on some occasions, that the answer is ‘yes’:

Suppose we live under a reasonably decent society, with the protection of basic liberties and some important public services, but we are convinced that the distribution of the tax burden is suboptimal because it is unfair. Wouldn’t this be a case in which one may think that we have a preemptive reason to pay the tax, no matter how strongly we think that the scheme is unjust and must be reformed in the future? (^{Bustamante, 2021a}, p. 187).

The federal government is operating within its jurisdiction when it introduces this policy, even though (like the previous example) the policy is suboptimal. Bustamante asks, “Could not we say that the normal justification thesis generates conclusive reasons to uphold this tax scheme even if we think that it is somehow substantially mistaken?” He adds that “If the answer is yes, then Raz’s distinction between jurisdictional and non-jurisdictional mistakes still holds” (^{Bustamante, 2021a}, p. 188).

This example potentially carves out a place for the NJT and the PT. Notice that this example only requires us to revisit the reason underpinning the rule once, and then we treat tax law as a set of authoritative pre-emptive reasons for action. In other words, tax law can perform its mediating role. So far so good. What about the NJT? The question now arises: does the tax scheme get its moral authority from the idea of jurisdiction? Or the underlying assessment of the importance of tax as an institution in the society in question? If the answer is “jurisdiction” (qua jurisdiction) then we are back to the Benthamite reasoning: citizens must obey promptly and censor freely.

If, however, the answer is that the underlying reason does the work, then the NJT references the foundational principle articulated by Bustamante: our obligations turns on whether or not our society is “reasonably decent” and provides certain key benefits like “basic liberties” and “public services.” If society provides these benefits, then we have a good (moral) reason to treat tax law as a set of pre-emptive norms. This is a strong argument. The problem is that it is not at all clear that this is a Razian argument. As mentioned above, and as Bustamante concedes, Raz rejects the suggestion that we ought to make general assessments about the moral authority of the legal system (^{Raz, 1986}, p. 70; ^{Bustamante, 2021a}, p. 187). Traditional natural lawyers embrace this approach. This kind of argument is offered by John Finnis, for instance, but not by Raz (^{Finnis, 1984}, p. 120).

While I suspect a persuasive account of law’s moral authority can be built on the back of this example, I am not yet convinced that Raz’s theses add anything of value to the picture Bustamante is carefully painting. Whenever Bustamante manages to ease the tension identified at the outset, Raz becomes indistinguishable from his intellectual opponents – Bentham or the natural lawyers. The ambiguity that is at the heart of Raz’s account makes his theory appear infinitely flexible. But upon closer inspection, it is far from clear whether Raz’s theses do the needed intellectual work in concert with each other.

Before closing I wish to consider one more reply from Razians, which can be offered if we look at the examples in a slightly different way. Bustamante, in both of his examples, is drawing attention to the conventional aspects of legality: instances where we ought to follow the law even when the law is unjust. The question now becomes: can Raz accommodate this conventional dimension of law into his account of authority?

In MF, Raz argues that a reason for treating legal norms as if they have pre-emptive force is found in law’s coordination function:

An orderly community can exist only if it shares many practices, and (…) in all modern pluralistic societies a great measure of toleration of vastly differing outlooks is made possible by the fact that many of them enable the vast majority of the population to accept common standards of conduct (^{Raz, 1986}, p. 58).

The problem with this argument is that Raz appears to be leaving it up to the individual to decide whether to or not to prioritize co-ordination above the NJT. This argument is offered as one of the reasons individuals would choose to pre-commit to following legal norms (as the PT requires) and forgo returning to the reasons behind the rule (as the NJT requires). The fact that it is the individual who is left to make this decision for herself ultimately undercuts this appeal to law’s conventional dimension, which necessarily requires a broader, community-wide commitment. Notice that if I am right, Raz’s account may be criticized for being unduly protestant in its approach to this particular judgment, which is surely a moral one. Typically, this charge is levelled at Dworkin’s account. While Bustamante masterfully defends Dworkin from his critics on this point, it is far from clear that Raz can be saved (see ^{Bustamante, 2020} and ^2021c). Raz’s only way out is to abandon the agent-relative orientation of his theory, but this is the defining feature of his account of authority. While I would welcome this move, abandoning it will lead him deep into the intellectual territory of his opponents.

IV. Positivism and First Principles

Fabra-Zamora offers a probing critique of JP from the perspective of positivist doctrine. In the course of his commentary, he provides a careful exposition of the variety of views offered by scholars writing in the Hartian tradition. What becomes clear quickly, however, is that if one hopes to address all the views he references, the book would be unmanageable and likely of interest only to Hartians (who would, of course, not like the project very much). While the title of my book is broadly worded, I quickly make it clear that I am taking aim at Joseph Raz, who is one of the leading thinkers writing in the positivist tradition. Raz’s influence has been, and continues to be, quite remarkable. This means that when I employ the term “positivism,” I am usually referring to Raz’s position, either in whole or in part. When I draw broad conclusions about the entire school of thought, I aim to do so only when I believe my conclusions call into question a foundational assumption shared by all (or most) adherents. The strategy comes with a certain amount of risk, but the alternative strategies have their own risks. In what follows, I will address what I view as the two key points of disagreement between Fabra-Zamora and myself. The first point of contention is interpretive; the second, methodological.

Fabra-Zamora rightly observes that in JP I rarely discuss certain core positivist ideas, including the distinction between internal/external points of view, and the “detached” perspective (^{Fabra-Zamora, 2021}, p. 201).⁸ I do not spend a lot of time on these topics because I do not think that an appeal to these ideas can save Raz’s project.⁹ For instance, any statement can be re-stated in a detached fashion.¹⁰ This move does not tell us whether the construction of a concept of law is possible without moral and political argumentation about law. I will return to this point below.

Fabra-Zamora contends that I have made an additional error: I have altered “the positivists’ order of explanation” (^{Fabra-Zamora, 2021}, p. 201). I suspect I am guilty of this charge. But the shift is not accidental. If one rejects the composition of the chessboard, it makes little sense to play the game on the terms of its creator. My argument may very well be disorientating to positivists, but this is simply because I am operating with a different set of first principles. In other words, the dispute between Fabra-Zamora and me is based on a fundamental difference in philosophical orientation. Nevertheless, we both agree that it is important to get Raz right. I will now turn to our interpretive disagreement, which focuses on PRN.

Contrary to my claims in chapter one of JP, Fabra-Zamora maintains that PRN does not give us an account of law and order. Instead, he argues, the discussion of judging should be read in the broader context of institutions in which the discussion of law is couched. Fabra-Zamora is correct that Raz situates his discussion of the courts in PRN within a broader discussion of law applying organs and practical rationality more generally. But in my view, little turns on this. Raz makes it clear that law-applying (rather than law-creating or enforcing) institutions hold the key to understanding law. This is so regardless of whether he thinks such law-applying organs hold the key to all rule-governed bodies. Raz insists that “courts, tribunals and other judicial bodies are the most important example of primary organs” (^{Raz, 1999}, p. 136). The reason he maintains that the “presence of a primary organ is a defining feature of institutionalized systems” is twofold: it is “based not only on our common knowledge of typical cases of legal and similar systems but also on the crucial role such institutions, when present, play in regulating social relations” (^{Raz, 1999}, p. 137). Raz adds that the presence of these institutions is of “momentous importance to their utility and function in regulating social behaviour” (^{Raz, 1999}, p. 137). Raz does not tell us, of course, that legal rules are more important than, say, the rules governing the local tennis club. But it is hard to resist this conclusion given that law is the focus of PRN and not, say, the local tennis or the school board.

The relationship between law and order also appears in the distinction between legal systems and systems of absolute discretion. Raz makes it clear that we either have legal systems that guide our conduct, or a situation where we do not know what is expected of us because judges can make the decisions. Recall that the imaginary tribunals in systems of absolute discretion “are not obliged to follow any common standards and can decide whatever they think best” and, Raz adds, this means that “such a system does not provide any guidance to individuals on the behaviour which would entitle them to a decision in their favour, should a dispute arise” (^{Raz, 1999}, p. 138). It does not follow that law is the only thing that is responsible for order, but there it is clearly stated that law guides our conduct because of this judicial duty.

To see this point with greater clarity, consider, once again, the argumentative strategy Raz employs.¹¹ Raz gives us a glimpse into this fictitious, disorientating world populated by discretionary decision-makers in order to lead us to a single conclusion: this is not our world. As mentioned above, in this imagined world where judges exercise discretion, citizens do not know what the law expects of them, in part because they cannot even identify the legal norms (^{Raz, 1999}, p. 139). If we cannot identify the law (because Raz had assumed that we identify the law by looking to the norms judges are dutybound to apply) then, of course, we cannot expect judges to uphold our legal rights. Our legal rights do not exist in this imagined world. It is hard to avoid the conclusion that this would have a significant impact on the social order, at least insofar as we assume that law plays some kind of role in creating and sustaining an orderly society. Raz has denied ever holding strong views about law’s order-engendering function and there is good reason for this: the risk is that an account of law’s order-engendering role will involve value-laden assumptions that are incompatible with his positivist methodological commitments (^{Raz, 2009}, pp. 382-396). In JP, I cast doubt on the possibility that Raz can shed the narrative, even if he hopes to do so (^{Martin, 2014}, chapter 7).

Fabra-Zamora also objects to my interpretation of the judicial duty to apply the law. He contends that we can, and should, interpret Raz’s claim that judges have a duty to apply the law in a manner that is consistent with the claim that judges also have a good deal of discretion. In order to understand Raz’s position, he asks us to think of judges as referees: both make decisions according to a set of pre-existing rules that govern the exercise of their powers. Like referees, “judges can have significant discretion and still be bound by pre-existing rules” (^{Fabra-Zamora, 2021}, p. 202). In other words, he interprets PRN in a manner that is consistent with EPD. Recall that in EPD, Raz argues that judges frequently exercise discretion (and that they should do so) (^{Raz, 1994}, p. 338).

Of course, the amount of discretionary activity a judge enjoys in a given system (or according to a given theory) sits on a spectrum: it is not an all-or-nothing matter. I argue that Raz assumes, in PRN, that judges are on the far end of the spectrum. Judges apply the primary norms of the system more often than they create new law. Conversely in EPD, Raz is on the other end of the spectrum: he holds that judges exercise discretion with a notable degree of frequency. Fabra-Zamora wants to situate PRN closer to EPD, but I remain doubtful that this is correct. Consider the following evidence from PRN:

Fabra-Zamora insists that I would be better served by reading Raz’s discussion about the nature of adjudication in PRN in light of his comments about “comprehensiveness, supremacy and openness” (^{Fabra-Zamora, 2021}, p. 203). However, I do not do so precisely because of the strength of the evidence presented above.

Let me now turn to the methodological dispute between Fabra-Zamora and me. Fabra-Zamora insists that I have failed to grasp the key tenets of legal positivism, including the scope of legal positivists claim about their chosen method. He offers this formulation of my position before explaining where my error lies:

MLP2) A (conceptual) theory of law does not need to consider politico-morally-based adjudication (^{Fabra-Zamora, 2021}, p. 204).

I do not (and would never) present my position in this a-contextual fashion because it leaves too much room for ambiguity – ambiguity which Fabra-Zamora proceeds to exploit. He argues that I fail to recognize the fact that positivists allow for all sorts of claims that are normative in nature and, for this reason, the position he attributes to me above, is too sweeping. Fabra-Zamora states that “[m]ost legal positivists are happy to recognize that, setting aside the theories of law’s existence: a theory of law includes an undeniable normative dimension” (^{Fabra-Zamora, 2021}, p. 205). But this is precisely the point that I am not willing to concede: I doubt that this conceptual island is available to the positivists. I have not misunderstood the positivist’s position. I simply reject it.

Consider an example. At the outset of The Concept of Law, Hart assumes that law can be distinguished from mere coercion (^{Hart, 2012}, chapter two). Some critics have pointed out that it is far from clear that Hart can explain the difference between law and the proverbial gunman writ large without relying on a normative argument – the mere presence of a set of conventions guiding official behaviour does not take us far enough (^{Simmonds, 2011}, p. 610). Notice, also, that this is a question about the existence conditions of legality. It is not a question about what law ought to be in some idealized, utopian sense.¹⁵ To see the problem more clearly, let us translate the question into Razian terms: is it possible for a legal system to systematically fail to realize the authority it claims?¹⁶

Consider the first option. If it is possible for a legal system to fail to enjoy any of the moral authority it claims, then legality becomes the vehicle for the exercise of illegitimate authority. Recall that Raz holds the position that legitimate norms are binding, which means that no laws in such a system will be binding on law’s subjects. It appears that we have a system of power that is clothed in the veneer of legitimacy. The possibility is left open that legal systems may take the form of the gunman writ large. Surely, I would want to know why such a system would (or should) qualify as a legal system. Any argument that could be offered by way of response would be normative in nature. A similar problem emerges if Raz opts for the second path.

If Raz says “no” to the initial question and maintains that all legal systems must enjoy legal authority to some extent, we have a clear distinction between legal systems and the gunman. However, this comes at a price: Raz is not merely insisting that all law necessarily claims authority, instead, he is now committed to the view that law must realize its claim to moral authority to some degree. This means, as a matter of substance, that legal systems necessarily contain legal norms that enjoy moral legitimacy. There are methodological implications as well: the moral ideal of authority, and not the mere “claim to authority,” is at the centre of Raz’s account.¹⁷ Raz becomes an inadvertent natural lawyer.

It is this second position that Raz seems to want to champion. Raz argues that officials cannot be “systematically confused” (^{Raz, 1994}, p. 217). In other words, law must be the kind of thing that can enjoy moral legitimacy – the claims of officials cannot be wholly vacuous. However, it is difficult to know why officials cannot be mistaken (or why they cannot intentionally abuse the system). But Raz’s insistence that officials cannot be mistaken is but a distraction: whether or not law is legitimate, according to Raz’s account, turns on whether or not it reproduces the requirements of right reason. The beliefs officials hold about the nature of their role are irrelevant. If Raz is going to support his claim, he must argue for it in greater detail: how can we be sure that law reproduces the demands of right reason, at least in part? Any answer offered will lead Raz into the territory of the natural lawyer.

My point, at this juncture, is as follows: the determination of the boundary between the law and the gunman writ large is a normative matter, hence Raz should be worried. He does not have the resources to explain the distinction if he adheres to non-normative conceptual analysis to explain his account of law. In other words, his non-normative account of law cannot be kept wholly separate from his normative account of authority. The only way to avoid this conclusion is to avoid answering the question about the status of the gunman writ large altogether. This is the route Raz takes in Between Authority and Interpretation. Put differently, there is no a-political, non-normative, answer to the question so best to avoid it.¹⁸

In this book, Raz nods in the direction of Finnis and Dworkin, attributing to them the view that “law can only modestly fail in its claims” adding that he has also suggested a “view very close to this.” (^{Raz, 2009}, p. 113). The difference between their positions and his is typically understood to be Raz’s insistence that it is the claim to authority, and not its realization, that is the feature that unites all legal systems. Raz likely spies the problem that I have been discussing. When he reflects further on the nature of his own position, he retreats to the idea of a “detached” statement, before restating the issue:

Typical legal statements can be either committed or detached. Committed statements, I suggested, entail the legitimacy of the law. Detached statements do not include this implication. I too regard committed statements as primary. The difference between my view and its rival is that I believethat far from non-committed statements being relatively rare, and an extension of the discourse of law to describe political systems which are not legal strictu senso, detached statements are prevalent in legal discourse about our own or any other legal system. This makes it possible for me to say that there are legal systems in the world even if we are mistaken about which ones, if any, enjoy moral legitimacy. On the alternative view, if all legal systems lack legitimacy then all the statements to the effect that there are legal systems are simply false (^{Raz, 2009}, pp. 113-114).

Initially, Raz appears to align himself with the view that law must enjoy some degree of the legitimacy it claims. However, his assertion that “committed statements (…) entail the legitimacy of the law” is a statement about the views of the speaker and not about whether the law is actually legitimate. Recall that for Raz, whether the laws in question enjoy legitimacy turns on whether they reproduce the requirements of right reason. The views of officials are irrelevant to this determination. At best, this point is deeply ambiguous.

The second portion of this lengthy passage seems to suggest that Raz is committed to the second option, namely, that a legal system could fail to secure moral legitimacy. Not only might we miss-identify which systems enjoy legitimate authority, Raz leaves open the possibility that all systems may in fact, fail to do so. Raz seems to suggest that by adopting the detached perspective he is able to tell us that some legal systems exist in the world, “even if we are mistaken about which ones, if any, enjoy moral legitimacy.” Here Raz appears to leave open the possibility that law can completely fail enjoy the legitimacy it claims. It can be the Gunman in disguise.

Raz, however, does not commit to this view. Instead, he quickly pivots back to the alternative position and proceeds to entertain the possibility that law must enjoy a degree of legitimacy in order to qualify as a legal system. Notice that he outlines the position in a non-committal, detached, way: “On the alternative view, if all legal systems lack legitimacy then all the statements to the effect that there are legal systems are simply false.” It is possible that legal systems that do not enjoy moral legitimacy do not exist. Raz has re-stated the original problem about how (or whether) we can distinguish law from the gunman writ large in a remarkably opaque fashion.

Raz proceeds to inform us that he “will not try to resolve the dispute here.” (^{Raz, 2009}, 114). By side-stepping the issue, he avoids giving up the game, so to speak. The distinction between law and the gunman is a normative distinction and it requires an argument that would move Raz into the territory of the natural lawyer (at least if he wishes to retain his thesis that officials cannot be wholly mistaken). But if he opts for the second position – that legal systems may fail to enjoy moral authority altogether, this raises another issue. How do we know such systems are “legal” systems? Perhaps they are best understood as systems where power is exercised through the guise of law.

A second example places this point in relief. Consider a society where judges are regularly bribed. In such a society, trials are merely show trials: the accused is not given a chance to convince the judge (or the jury) of his innocence.¹⁹ If this is a systemic problem, are we still dealing with a legal system? Fabra-Zamora might interject that I have made some key errors. I have failed to respect the distinction between a theory of law and a theory of adjudication. I am offering an “internal” account about the nature of “adjudication,” and not an “external” account about “our concept of law.” Only the latter is concerned with “existence conditions.” He will likely add that I have made an additional error this time: I have failed to separate my account of the ideal of the rule from my account of the concept of law.

But this example points to another possibility: we may have reflect on the nature of adjudication in order to determine if we are dealing with a legal system. At the very least, we cannot rule out this possibility before we begin our investigation. Jeremy Waldron notoriously accuses the methodological positivists of begging the question on precisely this point:²⁰

... it begs the question to say that the concept of law must be regarded as part of that descriptive or empirical apparatus, or that we cannot perform the descriptive or empirical task without it. On the view that I shall argue for, to describe an exercise of power as an instance of law-making or law-application is already to dignify it with a certain character; it is already to make a certain assessment or evaluation of what has happened (^{Waldron, 2008}, p. 12).²¹

The problem is not merely that positivists like Raz omit key features of legality, but that in the process, they lend credence to existing systems that are using legal institutions to exert arbitrary power. Waldron adds that while Raz emphasizes the importance of courts in PRN, he places all of the emphasis on judicial “output,” omitting any discussion of process (^{Waldron, 2008}, p. 22).

If we apply Raz’s criteria in MF, or Hart’s criteria in The Concept of Law, we may end up classifying the corrupt legal system as a legal system: it may have the relevant institutional trappings that are central to Hart’s account (a union of primary and secondary rules, minimum degree of official acceptance of the conventions that govern their behaviour, and general obedience);²² its norms may meet the NJT requirement as set out by Raz (the content of the law may reproduce the demands of right reason). But it is far from clear that this is a society governed by law.

Defenders of the Hartian tradition may wish to say they can account for this by explaining the content of the rule of recognition in more detail. But if it is the case, then this opens up the possibility that the rule of recognition necessarily contains moral criteria – criteria that are not discussed by Hart. And, in addition, the content of these criteria would pertain to the nature of adjudication, and specifically to the norms that govern the behaviour of judges. If Hart made this move he would have to abandon his separability thesis (which posits that the connections between law and morality are contingent in nature). In addition, Hart would then be signaling that the divide between normative theories of adjudication and non-normative theories of law is as fragile as the divide between theories of law and theories of the ideal of the rule of law (this divide is also breached by this move).

In addition, any attempt to integrate the need for good-faith decision-making in the courtroom into Raz’s normal justification thesis will require Raz to concede too much. My example of a corrupt judiciary does not refer to de facto authorities (such a system may qualify as a de facto authority without also qualifying as a legal system). The question is whether it is best thought of as a legal system. If the answer is no, or maybe not, then the door is opened for moral criteria to inform the divide between law and not law. Raz emphatically rejects this possibility – he rejects the idea that law has an essential task in addition to rejecting the possibility that it performs a valuable one (^{Raz. 2009}).²³

It is also worth pointing out that both Raz and Fabra-Zamora assume that the judges are acting in good faith in all the discussions I canvass above. The debate, thus far, has been about the amount of discretion that judges can, do, or should have. There has been no mention during these discussions about the potential abuse of power on the part of officials, including judges. In other words, they both quietly assume that judges are living up to the moral ideal that structures their profession: i.e. that the job of the judge is to make good-faith determinations about the issue at hand, in accordance with the norms and conventions that govern their behaviour. In short, it is far from clear that the rule of law is a wholly separate concept from law, it is also far from clear that the dividing line between law and non-law can be identified without considering what happens in the courtroom.

In my view, at best, this corrupt system is a morally “deficient” legal system. But I am open to the possibility that this corrupt system is not a legal system at all. The moral failing in question takes place in the courtroom, but it is this moral failing that threatens to transform the system from a legal system into the gunman writ large. So again, the points of disconnect between Fabra-Zamora and I are not a result of my own confusion about the positivist project. I reject the positivist’s starting point. I have argued here and elsewhere that there is no island, however small, where non-normative analysis can illuminate law.