abstract. For more than forty years, jurisprudence has been dominated by the Hart-Dworkin debate. The debate starts from the premise that our legal practices generate rights and obligations that are distinctively legal, and the question at issue is how the content of these rights and obligations is determined. Positivists say that their content is determined ultimately or exclusively by social facts. Anti-positivists say that moral facts must play a part in determining their content. In this Essay, I argue that the debate rests on a mistake. Our legal practices do not generate rights and obligations that are distinctively legal. At best, they generate moral rights and obligations, some of which we label legal. I defend this view by drawing analogies with other normative practices, like making promises, posting rules, and playing games. And I try to explain why it looks like legal practices generate distinctively legal rights and obligations even though they do not. I conclude with some thoughts about the questions that jurisprudence should pursue in the wake of the Hart-Dworkin debate.

author. Professor of Law and Professor of Philosophy, University of Michigan, Ann Arbor. Thanks to Elizabeth Anderson, Jules Coleman, Chris Essert, Evan Fox-Decent, Mark Greenberg, Daniel Halberstam, Don Herzog, George Letsas, Liam Murphy, David Plunkett, Richard Primus, Peter Railton, Don Regan, Larry Sager, Alex Sarch, Steve Schaus, Sam Scheffler, Gil Seinfeld, Stephen Smith, Scott Shapiro, Seana Shiffrin, Nicos Stavropoulos, Jeremy Waldron, and Ben Zipursky for helpful comments and conversations about this Essay and early ancestors of it. Thanks also to audiences at Harvard, McGill, NYU, UCLA, and UNAM.

Introduction

For more than forty years, jurisprudence has been dominated by the Hart-Dworkin debate. The terrain of the debate has shifted several times, but it is not hard to say what is in dispute. Hart and his heirs contend that the content of the law—the set of rights, obligations, privileges, and powers in force in a legal system—is determined by social facts. Dworkin and his followers counter that moral facts play a part in determining law’s content. Some find the debate moribund, but the truth is that the last decade of the debate has been as productive as any. Even though most participants defend positions that have been familiar for twenty years or more, the arguments advanced are increasingly sophisticated. They have not resolved the debate, but they have deepened our understanding of it. Still, I am sympathetic to the prescription of those who think the debate stale: we should move on.

We should move on because we can. There is a way out of this fly-bottle.1 Indeed, as Wittgenstein might have supposed, we are trapped by our own confusion, or at least that is how it now seems to me. The position I am going to defend is not completely novel. In recent years, Mark Greenberg has developed a view that shares much in common with it.2 But Greenberg sees himself as answering the question at issue in the Hart-Dworkin debate, rather than moving beyond it. Moreover, as Jeremy Waldron recently observed,3 Dworkin himself seems to have hit on something like the view I will defend toward the end of his life, and I think he glimpsed it much earlier. But for most of his career, Dworkin was buzzing around the fly-bottle with the rest of us, developing and defending a view that participates in the confusion that I hope to free us from. I’ll return to this history later. For now, I just want to emphasize that my aim is to reject the question at the center of the Hart-Dworkin debate, rather than defend anyone’s answer to it.

Of course, an end to the Hart-Dworkin debate would not mark the end of jurisprudence. But it would allow us to reorient jurisprudence toward a different end. For far too long, the field has been preoccupied by a question that is poorly formed. The time has come to set it aside and take up a better one. But before we can set a new end for jurisprudence, we must free ourselves from the old one. To start, we should remind ourselves what the fly-bottle looks like from the inside.

I. the fly-bottle

We’ll turn to the Hart-Dworkin debate in a moment, but I don’t want to start there. Instead, I want to start with a sign that is just around the corner from my house. I rarely give the sign much thought, but it poses all sorts of puzzles. Some are historical: Who put the sign there? Who decided to put it there? Who decided it would say, “SPEED LIMIT 35”? In addition to the historical puzzles, there are sociological ones: Do people notice the sign? Does it affect their behavior? Still other puzzles are normative: Should the sign have a different number on it? Should it be a bit further down the road? Should it be there at all? And then there is the puzzle that interests me the most, which is also normative, but in a different way: Does the sign affect how people ought to behave? If it does, how and why? What are the normative upshots of the fact that the sign is where it is and says what it says?

Some answers are easy enough. To start, the sign has prudential upshots. It signals something about the speed at which it is safe to drive. Assuming the sign got there in the normal way, its text and location reflect decisions made by people with expertise in traffic control. Given the characteristics of the road and the neighborhood surrounding it, it may not be safe to drive much faster than thirty-five miles per hour. Now that might be true quite apart from the sign, in which case the sign does not create new reasons but instead signals reasons that I already have but might not recognize. But the sign might also create new reasons. Driving is in part a coordination problem. It is safest to drive roughly the same speed as everyone else. If other drivers will react to the sign by traveling about thirty-five miles per hour, prudence may require that I do the same. And prudence may have even more to say. If the police are likely to ticket people who drive in excess of speeds posted on signs like this one, then prudence may require that I keep my speed down to avoid a fine.

The sign also has moral upshots. Some are closely related to the prudential ones. To the extent that I have moral reasons not to impose excessive risks on others, the presence of the sign, and the reactions other drivers are likely to have to it, may make it the case that I am morally required to drive about thirty-five miles per hour. But there are other ways the sign might make a moral difference. Perhaps the people who put the sign there have the moral authority to decide how fast I should drive. If so, I may be morally obligated to do as they have instructed. That is, I may be morally obligated to drive no more than thirty-five miles per hour. Or perhaps I have promised my wife that I won’t get any more speeding tickets. If so, I may be morally obligated not to act in ways that would lead the police to ticket me. That might require that I drive at less than the speed posted on the sign; more likely, it requires that I not drive too much above it.

Many people assume that the sign has yet another kind of normative upshot. They say that whatever the sign requires as a matter of prudence or morality, it legally requires that I drive no more than thirty-five miles per hour.4 That’s a familiar thought, but I should note one complication with it. Some people who speak this way think that our legal practices generate a distinct domain of legal normativity, separate from other normative domains, like morality and prudence.5 To these people’s way of thinking, a complete list of the sign’s normative consequences would need to include its distinctively legal consequences alongside its moral and prudential ones. Other people, however, would deny that the legal requirements imposed by the sign are properly listed with its moral and prudential requirements. Those latter requirements, they might say, are inherently normative, while the legal requirements need only purport to be normative and, indeed, might not be. This view comes in a variety of flavors. The most common holds that when we refer to legal requirements we are referring to the law’s point of view on our moral requirements. I’ll say more about this later. For now, I just want to note that if legal requirements are not genuinely normative, they are at least quasi-normative, as they traffic in normative notions, like obligation and right. And that is enough for our purposes. Indeed, the thought that traffic signs and the legal practices they are embedded in have distinctively legal upshots—that is, normative or quasi-normative upshots that are legal but not moral or prudential—is the glass that makes the fly-bottle. And it is that thought that I am going to propose we reject.

Why is the thought that legal practices have distinctively legal upshots the glass that makes the fly-bottle? Because that thought sets the terms of the Hart-Dworkin debate. Indeed, without that thought, there would be nothing to debate. Let me show you what I mean. Suppose that I tell you that when you are on this particular road, you are legally obligated to drive no more than thirty-five miles per hour. And now suppose that you ask me why that is. There are two ways to hear your question. You might want to know why the people who set that requirement set that one, rather than a different one. That is, you might want to know why they set the speed limit at thirty-five, rather than twenty-five or forty-five. But you might be after something else. You might want to know what makes it the case that you are legally required to drive no more than thirty-five miles per hour. That is, you might want to know what facts make that fact obtain. After all, the fact that you are legally required to drive no more than thirty-five miles per hour is not a basic fact about the world as we find it, in the way that we might suppose some fundamental physical facts are. There are further facts that make the speed limit what it is, and it is reasonable to suppose that we could figure out which facts those are.6

Without doubt, some social facts are among the further facts that determine the speed limit. In other words, the speed limit is what it is at least in part because of what certain people said and did. Someone, we can be reasonably sure, decided to set the speed limit at thirty-five rather than at twenty-five or forty-five, and that decision is one of the facts that makes the speed limit what it is. Everyone in the Hart-Dworkin debate agrees about that. What they disagree about is whether all the facts that figure in fixing the legal requirement are social facts, or whether there might be (or must be) some normative facts that figure in determining the content of the law too.7

Roughly speaking, there are three positions in the debate. According to exclusive legal positivists, the content of the law is determined solely by social facts. If this view is right, then when we set out to explain why you are legally required to drive no more than thirty-five miles per hour, we may point only to facts about what people have said, done, thought, and so on. It would be a mistake for us to point to any normative facts about what people should say, do, or think. The reason it would be a mistake is that facts like that play no part in determining the content of the law. Anti-positivists hold the opposite view. They think that to fully explain why the speed limit is thirty-five, we must point to some normative facts alongside the social facts. An anti-positivist might argue, for example, that we must point to moral facts that determine the legal relevance of actions taken by different people or institutions. Something of a middle ground is occupied by inclusive legal positivists, who hold that moral facts might play a part in determining the content of the law, but only if the relevant social practices assign them that role. Inclusive legal positivism is a form of positivism because it holds that social facts are the ultimate determinants of the content of the law, and that the law might be determined by social facts alone. But it allows that people might choose to have the content of their law depend on moral facts, as they seem to do, for example, when they prohibit punishment that is cruel, or confer rights to legal protections that are equal.

The last several decades of jurisprudence have seen a pitched battle between these views. But for all their differences, they share something in common. And that should not come as a surprise, since what they share is a condition of their coming into conflict. All three views offer an answer to the same question. The question is what facts determine the content of the law. In different ways, all three views purport to provide a metaphysical account of our legal rights, obligations, privileges, and powers. In other words, they purport to tell us what makes it the case that we have the legal rights, obligations, privileges, and powers that we do.

II. the troubles

So framed, the Hart-Dworkin debate is important. The law can make a great deal of difference in our lives, and we often disagree about what it demands. An account of how and why the law requires what it does should illuminate those controversies and maybe even contribute to their resolution.8 It is no mystery, then, that the Hart-Dworkin debate has held our attention for as long as it has. But it remains devilishly difficult to resolve, and I want to quickly review some of the problems that make it so.

Let’s start by trying to untangle the metaphysics of just one legal obligation. Suppose you ask a lawyer what makes it the case that you are legally obligated to drive no more than thirty-five miles per hour on the street near my house. She might tell you that an administrative agency set the speed limit on that road and posted the sign to inform you of it. But you might ask: why does that sequence of events have any bearing on what I am legally obligated to do? And she might reply that the administrative agency was acting pursuant to a statute adopted by the state legislature and signed by the Governor. Unsatisfied, you renew your question: why do the actions of the state legislature and Governor have any bearing on what I am legally obligated to do? If she’s not annoyed with you, she might point to the state constitution, which says that the legislature is authorized to adopt statutes, subject to the Governor’s veto. Ever the contrarian, you push one more time: why does the text of the state constitution have any bearing on what I am legally obligated to do? If your lawyer remembers her jurisprudence, she will know that Hart had an answer to this question.

According to Hart, every legal system has a foundational rule—Hart called it the rule of recognition—that identifies the other rules that are part of that system.9 Those other rules establish people’s legal rights, obligations, privileges, and powers. The rule of recognition is not validated by some further rule. Instead, it is a social rule—that is, a rule whose existence and content are fixed by a social practice.10 Roughly, the right kind of practice exists when most legal officials converge on criteria for identifying law, treat their convergence as supplying a common standard, and regard themselves as obligated to comply with it.11 The content of the rule of recognition is fixed by the criteria that legal officials converge on and take the proper attitude towards. If your lawyer remembers all this, she might tell you that the practice of legal officials around here is to recognize rules made in accord with the procedures in the state constitution as legal rules.

Hart’s picture is elegant. But it has drawn lots of critics, and even some positivists worry about it. They worry because it seems to license inferences that run afoul of David Hume’s famous injunction that you cannot derive an ought from an is.12 Hart invites us to derive a normative statement (that is, a claim about what you are legally obligated to do) from descriptive statements about the social practice among legal officials around here. But if Hume is right, inferences from merely descriptive statements to normative statements are invalid. Of course, Hume might have been wrong, and some philosophers think so. But anyone who would defend a legal positivism like Hart’s must show Hume wrong or navigate around his injunction that you cannot derive an ought from an is.

Many contemporary positivists take the second tack, and there are several different strategies on offer. I want to highlight the most common strategy, both because it seems to me the most promising and because it will figure in the argument to follow. As I mentioned earlier, some people think that there is a distinctively legal domain of normativity, separate from other normative domains like morality and prudence.13 Hart held a view like this; he thought that the legal concept of obligation was normative but not moral.14 Other positivists, however, think that law employs the same concept of obligation as morality, so that claims about a person’s legal obligations are really claims about her moral obligations. But they think that these claims are qualified in an important way. To say that a person has a legal obligation is not to say that she has a moral obligation full stop. Rather, it is to say that she has a moral obligation from the law’s point of view.15 On this sort of picture, when you talk about your legal obligations, you are talking about the moral obligations the law thinks you have, which is roughly akin to talking about the moral obligations your grandmother thinks you have. A claim about what obligations your grandmother thinks you have would be a descriptive claim, not a normative one. And the same is true of claims about legal obligations, according to positivists who hold this sort of view.16 To put it in the language I used earlier, claims about legal obligations are, on this picture, quasi-normative; they appear to be normative, but they are not really.

This strategy may help positivists escape Hume, but it raises new questions. It is easy to grasp the idea that my grandmother has views about what I am morally obligated to do. But the thought that the law has a point of view on what I am morally obligated to do is more elusive. The idea cannot be that the law has views in just the same way that my grandmother does.17 But if the law’s point of view is not the point of view of any particular person, we must figure out how all the different aspects of legal practice combine to generate the law’s point of view. And it is far from obvious how that happens. Indeed, Mark Greenberg has argued that legal practices are always consistent with many possible sets of legal requirements,18 such that social facts are, by themselves, incapable of fixing the law’s point of view.

There’s much more to say here, but we’re not trying to resolve the Hart-Dworkin debate. We’re just trying to appreciate why it’s so persistent. So let’s move on to a new set of problems. We started with Hart’s answer to the question how the content of the law is constituted. And then we considered a worry that many positivists have about Hart’s picture. But we haven’t yet said anything about the worries that Dworkin raised about it, and we should spend some time there, since it is, after all, the Hart-Dworkin debate that we are trying to get a grip on.

Over the years, Dworkin lodged many objections to Hart’s positivism, but we can boil the main part of his critique down to two complaints. First, Dworkin argued that legal officials do not converge on criteria for identifying law in the way that Hart supposed.19 Those criteria, he contended, are constantly contested, and not just at the periphery.20 Take, for example, what you might think is a basic question that any rule of recognition would answer: what legal rule is generated by an act of legislation? Around here, at least, legal officials disagree. Some think that the rule generated is the rule expressed in the text, whatever the legislature might have intended in adopting it.21 Others think that the legal rule generated is the rule that the legislature intended to adopt, whether or not that rule was fully or accurately captured in the text it approved.22 Hart had said that the rule of recognition is indeterminate as to any point on which legal officials fail to converge, with the consequence that the law is indeterminate on those questions too.23 But Dworkin observed that disputes of this sort don’t lead legal officials to conclude that the law is indeterminate. Even when judges can plainly see that they have not converged on criteria for identifying the law of their community, they nevertheless insist that there is law to be applied.24

Dworkin’s second complaint was that Hart could not explain all of the ways in which legal officials disagree about what the law is.25 Sometimes, they disagree about what the law is because they disagree about some or another social fact. Hart has no trouble with that. But sometimes, Dworkin observed, legal officials disagree about what the law is, even though they agree on all the social facts.26 On Hart’s picture, these sorts of disagreements are mysterious: since social facts constitute the law, any disagreement about what the law is should rest on a disagreement about some or another social fact. This puts Hart in the awkward position of denying the possibility of disagreements that seem rather routine.27

In Law’s Empire, Dworkin advanced a different picture of how our legal rights, obligations, privileges, and powers are determined. Roughly, Dworkin argued that the content of the law is a function of the principles that best fit and justify past political decisions about the state’s use of force.28 According to Dworkin, people who agree on all the social facts can nevertheless disagree about what the law is because they have moral disagreements about which principles best fit and justify their community’s political history.29 Those who hold that the legal rule generated by an act of legislation is the one the legislature intended to adopt might think so in part because they believe that best serves the values of democracy. In contrast, those who hold that the text reigns supreme might think so in part because they think democracy better served by holding the legislature to the words it chose. The parties to a debate like this disagree about the criteria for identifying law in their community, and they know that they disagree. But they think their answers not merely better, but right.30 And that, Dworkin suggested, accounts for the fact that they regard the law as determinate, even though they disagree about what it is.

Of course, positivists are not persuaded by all this. They argue that positivism can account for the sorts of disagreements Dworkin observed, which they often doubt are genuine anyway.31 Moreover, some positivists worry that law couldn’t serve the purposes that it is supposed to serve if it worked the way Dworkin says it does (though, I might add, they don’t all agree on what those purposes are, or even whether law has purposes in the first place).32 Finally, some positivists charge that Dworkin is embarrassed by the existence of morally abhorrent laws and legal systems.33 Now, I should say—because it will be helpful to have in mind later—I think this last worry is overblown. Dworkin never claimed that there are moral constraints on the existence of particular laws and legal systems, in the way that some natural lawyers from long ago might have done.34 So he’s not embarrassed by the mere existence of evil laws or legal systems. But they do pose a problem for his view nonetheless. The problem is that evil laws seem to impose legal obligations, even though they cannot be justified by any morally attractive principles. It seems apt to say, for example, that the Fugitive Slave Act obligated federal marshals to arrest those accused of being runaway slaves, notwithstanding the grave injustice involved. Any plausible anti-positivism must explain this, or explain it away, and many find Dworkin’s attempts to do so unsatisfying.

The point is not that Dworkin’s critics are right; maybe they are, and maybe they aren’t. As I said, I’m not trying to resolve the Hart-Dworkin debate. I’m just trying to convey a quick sense of the troubles that attend any attempt to sort out the metaphysics of our distinctively legal rights, obligations, privileges, and powers. And perhaps the only conclusion I am confident of is this: there are a lot of them.

III. the way out

There are a lot of them, and there’s an awful lot more to the Hart-Dworkin debate than we just canvassed. There are books about it, and there will be many more. But there don’t have to be. There is a way out of the debate, and indeed, the way out is as simple as the way in: to escape the debate, we could simply abandon the thought that starts it up. That is, we could abandon the thought that, in addition to their moral and prudential upshots, legal practices have distinctively legal upshots. This might sound radical; later, I will try to make it seem less so. But for now, just note that if we did abandon this thought, we would still wonder what the law requires of us. The only difference would be that when we did so, we’d take ourselves to be engaged in a moral inquiry, or perhaps a prudential one, rather than a distinctively legal one. That is, we’d take ourselves to be asking what our legal practices give us moral or prudential reasons to do.35 Of course, it might be difficult to identify the moral and prudential upshots of our legal practices. But whatever challenges those inquiries might pose, they would not present any metaphysical problems of a distinctively legal sort, which philosophers of law might take it as their task to solve.

Now, it should not come as a surprise that we could stop the Hart-Dworkin debate if we denied the existence of its subject. The question is whether we should stop it. And though we might welcome the opportunity to dispense with a difficult debate, the fact that the debate is difficult does not give us good reason to deny the existence of its subject. (The metaphysics of consciousness are hard to figure, but that does not give us much, if any, reason to doubt that consciousness exists.) Still, the possibility that we could stop the Hart-Dworkin debate raises a question: did we have good reason to take it up in the first place? As we just saw, the thought that legal practices have distinctively legal upshots lands us in a mess of trouble. If we don’t have good ground for it, we should drop it, and the Hart-Dworkin debate too.

I suspect that most people just think it obvious that legal practices have distinctively legal upshots. And for good reason. We regularly draw a distinction between what we are legally required to do and what we are morally required to do, and we are live to the possibility that legal and moral requirements won’t match up. We often face legal requirements that we doubt are accompanied by moral requirements. And we often face moral requirements that we know are not backed by legal requirements. The most straightforward way to make sense of this is to suppose that legal practices generate distinctively legal rights, obligations, privileges, and powers, which can and do differ from their moral counterparts.

This is a simple, powerful picture, and if it is right, then we cannot escape the Hart-Dworkin debate or the troubles that come with it. If we have distinctively legal rights, obligations, privileges, and powers, then we cannot avoid asking metaphysical questions about their constitution. But I don’t think this picture is right. Indeed, I think there is something quite puzzling about it. The world is full of practices that are law-like, in that they aim to shape the norms that govern our lives. But we don’t take the vast majority of these practices to give rise to their own distinct domain of normativity, or even quasi-normativity, in the way that this picture supposes that legal practices do. In the next section, I want to illustrate this puzzle with a series of vignettes that have little to do with law. What I hope to show is that we can navigate increasingly complex normative practices without generating the problems that preoccupy philosophers of law. If I am right, then we should wonder whether we can navigate law without generating those problems too.

IV. house rules

We can start by thinking through the normative upshots of another sign. But this time we’re aiming for something that’s not connected to law, or at least not very directly. So let’s try this: Imagine that you have rented a house at the beach. When you arrive, you notice a sign in the foyer, which reads, “Leave your cares at the door.” Should you leave your cares at the door? Maybe. It depends on what your cares are and why you are visiting the beach. If your cares are not pressing and the point of the trip is relaxation, then perhaps you should leave your cares at the door. If, instead, you have come to the beach because it is a quiet place to work on the things you care about, then you probably should not leave your cares at the door.

Of course, all of that would be true even if there were no sign enjoining you to leave your cares at the door. Does the sign play a role in determining whether or not you should leave your cares at the door? Of course, it might play a causal role in determining whether or not you do leave your cares at the door. The reminder may put you in mind to relax. But it is hard to see how the sign bears on the question whether you should leave your cares at the door. After all, it is difficult to imagine that the owner of the house has standing to demand that you do. And even if she does, it’s not clear that she intended to make a demand through the sign; she might just have liked the way it looked. It would be a different story, of course, if your spouse told you to relax. Her demand might give you a reason. But the sign does not.

Now suppose that the sign says, “No smoking.” Should you refrain from smoking? To be sure, you have reasons not to smoke quite apart from what the sign says. Smoking is harmful, to you and others. But unlike the first sign, this one has normative consequences, at least assuming the owner put it there. To the extent you are interested in avoiding conflict with the owner, the sign gives you a prudential reason to avoid smoking. And it probably gives you moral reasons too. The sign tells you that the owner of the house does not want you to smoke, and because smoke lingers behind, she has some stake in whether you do. Indeed, the sign might even obligate you not to smoke, on the plausible assumption that the owner of a house has a right to decide whether people may smoke in it.36 Of course, we might wonder whether she has done enough to invoke that right. It may be that she should have given you notice up front if she intended to restrict your smoking, rather than leave it to a sign you would see only on arrival.

But for our purposes it does not matter whether you should avoid smoking, or, for that matter, leave your cares at the door. The point of these vignettes lies in what I have not said about them. In sorting through the normative upshots of these signs, I have not suggested that there is a distinct domain of normativity, or even quasi-normativity, unique to the rental house or its signage. I did not, for example, suggest that the first sign generated a “rental house” obligation to leave your cares at the door, separate and apart from whatever moral obligations it might have generated. And I did not suggest that, from the second sign’s point of view, you were morally forbidden to smoke, whether or not the sign actually imposed such a moral prohibition. Yet the accounts I gave of these situations do not seem impoverished for my failure to invoke this conceptual machinery. Indeed, it is hard to see how the notion of a “rental house” obligation, or the suggestion that a sign has a point of view, would shed any light on the situation. There is a sign that says you should leave your cares at the door; it does not give you any reason to do so. If instead the sign said “No smoking,” it would give you several reasons to avoid smoking, and perhaps an obligation as well. That is all there is to it, and we do not need to invoke any special kind of normativity, or quasi-normativity, unique to the rental house or its signage, in order to make sense of these situations.

Let’s complicate the story to see if anything changes. Suppose that when you enter the rental house, you see a list tacked to the wall, labeled “Rules of the House.” It reads:

1. No smoking.

2. Take the trash out when you leave.

3. Do not use the garbage disposal; it doesn’t work.

4. Check-out is 11:00 AM. If you stay longer, you will be charged for an additional night’s stay.

5. Have fun!

What are the normative upshots of the Rules of the House?37 The fifth rule is similar to the sign telling you to leave your cares at the door. It might put you in mind to have fun, but it has no bearing on whether you should. In contrast, the first four rules seem to have both prudential and moral upshots. The fact that the owner took the time to write out the rules suggests that she cares about whether you do as they direct, and all four reflect matters about which the owner has standing to make demands. The second rule is interesting; it instructs you to take the trash out when you leave, but it seems likely that the obligation this rule generates is to take the trash out by the time you leave, not when you leave. This is because it is hard to see any purpose to taking out the trash when you leave, so long as you do it beforehand and do not generate any trash after you do. Though the owner of the house presumably has standing to demand that you take out the trash, the question whether you do it as you leave or a bit before does not seem to affect her interests, so it would be odd to attribute to her a right to determine the timing.

Or so it seems to me. But the interesting question is not whether I have all this right; it is whether something is missing on account of my failure to look for normative upshots of the Rules of the House that are neither moral nor prudential. Once again, it is hard to see how positing a special sort of normativity, or even quasi-normativity, unique to the Rules of the House would shed any light on the situation. So let’s complicate the story again. Suppose now that the rules tacked on the wall are preceded by the following statement: “The property manager is authorized to enforce the Rules of the House by adding appropriate charges to your bill.” This is the first we have heard of the property manager. If he has been lurking in the background all along, we might have thought him authorized to add appropriate charges to the bill even without this statement. But if we would have had any doubt about the role of the property manager, this helps clarify matters. It indicates that the owner of the house has delegated authority to impose additional charges to the property manager, and it is hard to see anything wrong with that, at least on the facts we have.

Exactly what charges the property manager may tack on is a tricky matter. If you smoke or attempt to run the garbage disposal, it would seem appropriate to charge you the cost of repairing any damage caused. A modest fee would likely be in order if you fail to take out the trash. The fourth rule specifies that if you overstay the rental period you will be charged for an additional night. That does not seem an egregious penalty, and the rule puts you on notice of it, so it is plausible that the property manager is permitted to charge you the sum specified, subject perhaps to a de minimis exception if you run just a few minutes over. Of course, the property manager may not penalize your failure to have fun, regardless of what the rules say.

What if the property manager adds an additional charge to your bill, but not for misconduct mentioned in the Rules of the House? Does that imply that he has done something impermissible, on account of the fact that his mandate is limited to enforcing the Rules of the House? No, probably not. If you knocked over a vase and the charge is to repair or replace it, it would be obtuse to object that knocking over vases was not against the Rules of the House. The property manager would presumably be permitted to impose remedial charges of this sort in absence of the statement that he is authorized to enforce the Rules of the House,38 and it is not tempting to apply the maxim expressio unius est exclusio alterius here.

Now we have an official, of sorts, charged with enforcing the Rules of the House. And we are starting to engage the sorts of arguments that occupy lawyers. We’ve asked whether notice is required before a penalty may be imposed; whether a rule is subject to a de minimisexception; and whether explicit authorization to engage in some acts implies a lack of authorization to engage in others. But yet again, it seems we can describe and assess this situation just fine without imagining that there is a distinct domain of normativity, or quasi-normativity, created by the rules of the rental house. And it is hard to see what benefit we’d get from invoking those notions.

But let’s give it another try. Suppose you check out a few minutes late and the property manager says, “Look, I don’t think this is fair, but I’m obligated to charge you an additional night’s rent. Those are the rules of the house.” Is the property manager telling you that he has a non-moral “rental house” obligation to charge you an additional night’s rent? I find it difficult to imagine that he is, as there are much more plausible ways to hear what he says. To start, the property manager might think that he is morally obligated to do something he regards as unfair. He might think that the owner of the house has the right to set the charge and the conditions for imposing it. That is, he might see himself as more or less a bookkeeper in the matter. If this is the property manager’s view, he might add, “I’m sorry. I’m just doing my job.”

Another possibility is that the property manager means to communicate that he is obliged to add the charge, not obligated to do it. That is, he might be trying to say that he is adding the charge to the bill because he does not want to risk reproach from the owner for failing to do so. If this is the property manager’s view, he might add, “I’m sorry. I need to keep my job.” Ever since Hart’s devastating critique of Austin, philosophers of law have guarded a distinction between being obligated and obliged, and helpfully so. However, there is no reason to think that property managers are punctilious about the philosophers’ distinction, which does not seem to have filtered into common usage.39

Of course, the facts we’ve imagined are rather thin, so I can’t say for sure how we should take what the property manager says. But I do know this: the property manager can navigate the situation that he’s in without supposing that the Rules of the House give rise to their own distinct domain of normativity, or that they have a point of view on how he should behave. And because he has no need for those notions, it would be odd for us to employ them when we interpret his claim that he’s obligated to charge an additional night’s rent.

But we can try one more time, flipping this last example around. Suppose again that you check out late and that the property manager tells you that he’s obligated to charge you an additional night’s rent. This time, however, he adds that he’s not going to impose that charge because he thinks it unfair. Is the property manager invoking a special sort of normativity, unique to the rental house? Once again, I’m skeptical, since there are much more plausible ways to hear what he says. As before, the property manager might think that, given his role, he is obligated to do what the owner of the house has instructed him to do, whether or not he thinks it fair. If he’s right, then in refusing to impose the charge, he’s failing to live up to the responsibility of the role. That may be unwarranted, but it’s a failing that is easy to understand, as people all the time feel conflicts between the roles they occupy and the decisions they would make if freed from the constraints of the role. But of course, there’s another way to make sense of what the property manager says. In referring to his obligation to impose the charge, he might just mean to indicate that he feels compelled to do so, even though he plans to resist. Of course, we’d need to know more about the situation to know how best to interpret the property manager. But once again, the property manager can navigate the situation without supposing that the Rules of the House generate a special sort of normativity, or even quasi-normativity, and because he can, we can, too.

We could go on, making these examples ever more complicated and law-like. But that would get tedious, and we have no reason to think anything would change. At every step, there would be moral and prudential upshots to the social facts that constitute the situation. However, there is no reason to think that we would arrive at a point where we would have to posit a distinctive class of non-moral “rental house” obligations to make sense of the situation. There is no reason to think that we would have to attribute a point of view to a sign or a set of rules. And that should give us pause. If we don’t need recourse to these ideas to understand the ways that people engage the rules posted in a rental house, maybe we don’t need to appeal to them to understand the ways that people engage law.

V. promises

But that’s getting ahead of ourselves. Before we turn to law, we should see if we can extend the lesson learned from the rental house cases to other normative practices. We can start with promises, which are one of the most common tools we use to shape the norms that govern our lives. Once again, we can work through a series of cases to see if we generate the problems that preoccupy philosophers of law.

Suppose that I promise that I will drive you to the grocery store at noon tomorrow. Before the promise, I wasn’t obligated to do so. After the promise, I am obligated, and that’s just what I intended. That’s straightforward enough, but now suppose that the time has arrived and your needs have changed. You don’t need to go to the grocery store, you need to go to the doctor, as you are worried about a sudden shift in your vision. Am I obligated to take you? We might want to know more. (How far away is the doctor’s office? Could you get another ride? How long would that take?) But I am inclined to think that I am obligated to take you. Certainly, I would think so if the time commitment were similar to the commitment I’d already made when I promised you a ride to the grocery store. And given the urgency of the situation, I would probably think so even if it were not. Moreover, if I am obligated to drive you to the doctor, it’s plausible that this obligation is a consequence of my promise to drive you to the grocery store. Had I not made that promise, you might not have had a claim on my time.40 If that is right, then this is a case in which my promise obligates me to do something that is not what I promised to do, which is an indication that we are not fully in control of the moral significance of our promises.

Now, imagine that I promise that I’ll murder your cousin. Before the promise, I wasn’t obligated to do so. After the promise, I’m still not obligated, as one can’t obligate oneself to commit murder. Of course, you and I might not recognize that. We might be mobsters, who think that a person who gives his word is bound to do what he said he would, even if it is evil. But that is one of the many ways in which mobsters are morally misguided. A promise to murder does not generate an obligation to murder. Indeed, if it generates any obligations at all, they are likely obligations to abandon the plot or foil it before it comes to fruition.

Once again, the interesting question is not whether I have all this right. It is whether anything is missing from the analysis on account of my failure to suppose that a promise gives rise to a distinct domain of normativity, or even quasi-normativity, separate and apart from the moral and prudential consequences that the promise might trigger. And once again, the answer is no. We wouldn’t gain anything by suggesting that, from the point of view of my first promise, I am morally obligated to drive you to the grocery store. And we wouldn’t make any progress by supposing that my second promise generated a promissory obligation to murder your cousin, even though it did not trigger a moral obligation to do so.41 So far as we are concerned, the promising cases look just like the rental house cases. Once we set out the social facts and run through the moral and prudential consequences of those facts, there’s nothing left to do. We don’t need to look for additional normative, or quasi-normative, consequences of the promises.

And though we have only considered a handful of cases so far, it is starting to look like there is a general lesson to draw here. When we post rules or make promises, we are aiming to shape the norms that govern our lives. But we do not shape those norms by creating, out of whole cloth, new sorts of normativity, or even quasi-normativity, unique to those activities. Rather, we shape those norms by shifting the social facts in ways that have moral or prudential consequences. If law is continuous with practices like posting rules and making promises, we might expect that legal practices shape the norms that govern our lives in the same way.42 That is, we might expect that legal practices shape the norms that govern our lives by shifting social facts in ways that have moral and prudential consequences, not by creating, out of whole cloth, a new sort of normativity, or even quasi-normativity, unique to legal practice.

VI. playing games

Is law continuous with practices like posting rules and making promises? Here is a difference that might seem to matter: the normative practices we have examined so far aimed to subject people to isolated norms or small sets of loosely related ones. Legal practices, however, aim to subject people to large sets of norms that bear systematic relations to one another. Now we should be careful not to make too much of this difference, for you can promise to abide by a system of norms. You might, for example, promise to follow the tenets of Orthodox Judaism. But there is still a difference here, as those tenets are not established by that promise; at best, the promise establishes the isolated norm that you should adhere to those tenets. Our legal practices are much more ambitious than this. Taken together, they aim to subject people to systems of norms, and, indeed, systems that regulate our lives rather pervasively. And we might wonder whether the fact that legal practices aim to subject people to systems of norms gives us reason to think they generate their own distinct domain of normativity.

Of course, legal practices are not the only sorts of activities that aim to subject people to systems of norms. The practices that constitute games do as well. And it is worth thinking about games in some detail, since philosophers often point to games as analogues for thinking about law. We can work with chess, which in its most organized form is strikingly law-like. It has a governing body—the World Chess Federation (FIDE)—that publishes official rules (actually, they are called “laws”), arranges tournaments, qualifies arbiters, and maintains rankings. All those actions plausibly have normative consequences. That is, they plausibly affect what people have reason to do—morally, prudentially, or (since we are dealing with a game) simply for purposes of having fun, or an opportunity to manifest the kinds of excellence that games like chess make possible. But now we might wonder: separate from these sorts of normative consequences, do the FIDE rules generate a distinct domain of normativity, or quasi-normativity, such that there are FIDE reasons or obligations that are not also reasons or obligations of other sorts?43

Once again, it will help to work with an example, so let’s try this. Imagine that you are in the midst of a game of chess when you absentmindedly pick up a pawn with your left hand and put it back on the board with your right. Your opponent objects. You look confused, and she explains that FIDE’s rules provide that a move must be made with one hand only. She’s right about that; in the midst of a lengthy section that addresses nearly every question that might arise about the act of moving pieces on a chessboard, the rules say that “[e]ach move must be made with one hand only.”44 Other rules in that section provide that a player must announce his intention before adjusting any piece he does not plan to move;45 move the first piece he deliberately touches if it is one of his own;46 and castle if he deliberately touches his king and rook, and it is possible for him to do so.47 These rules, and several more that surround them, are closely related to one another. There would be little point in having just one or two of them. But, taken together, they help ensure that a player cannot judge his opponent’s reaction to a move before completing it. And they provide clarity about when a player is committed to a move, such that he is not permitted to second-guess it. So when your opponent objects that you switched hands mid-move, she is complaining that you flouted a finely wrought system of rules that improve play of the game.

But did you really flout the rules? To answer, we need to know more about the context of the game. After all, the mere fact that FIDE has published rules for playing chess does not in itself affect what anyone playing chess ought to do. To be sure, the FIDE rules are rules that one could follow to play a game of chess. But they are not the only such rules, and until we know more about the context of the match, we cannot say whether FIDE’s rules have any normative consequences for you. For all we know, the FIDE rules might be just like the sign that said, “Leave your cares at the door.” They might not bear at all on the question what you should do.

Of course, it is easy to imagine circumstances in which the FIDE rules do have normative consequences for you. If you’re playing in a FIDE-sanctioned tournament, for example, then you are presumably obligated to follow the rules, and, if you fail to do so, you are subject to the appropriate sanction. This obligation owes in part to FIDE’s promulgation of the rules, but also to your decision to participate, knowing that those rules had been selected to govern the tournament. The picture looks rather different if you are playing a casual game with a friend. Unless you hang out with folks who are very serious about chess, it’s hard to imagine that switching hands mid-move is wrong regardless of what the FIDE rules say. If your friend objects, you’ll push back, reminding her that you are playing a casual game, which is not governed by arcane rules. If she insists that you did not do what the rule requires, you might retort that that rule did not require anything of you at all.

We need to pick apart this dispute carefully. There’s a sense in which your friend is right to say that you did not do what the FIDE rule requires. You didn’t do what the rule requires in the sense that you didn’t satisfy the standard specified in the text. But, of course, that doesn’t count as an objection if you had no reason to satisfy that standard. In other words, it is open to you to maintain that FIDE’s adoption of a rule requiring players to make moves with one hand only has no normative consequences for people engaged in a casual game of chess. And if you pressed that claim, I’d think you were right. A casual game of chess seems more apt to be governed by the widely known rules of chess than it does by FIDE’s more complete set of rules, absent an antecedent agreement about what rules are in play. I said before that the FIDE rules improve play of the game, but that will only be true when the point of playing is to match wits in the fashion that the rules facilitate. If the point is simply to have fun, then appeals to the FIDE rules will almost certainly get in the way.

But once again the question is not whether I have all this right. The question is whether something is missing from my analysis on account of my failure to look for non-moral FIDE obligations to play the way the rules say. And once again, the answer is no. FIDE adopted a rule that says chess players must make each move with one hand only. That rule is part of a system that improves play of the game, at least for some purposes. When you choose to play in a FIDE-sanctioned chess tournament, you are obligated to follow it. But when you play a casual game of chess, you are not. As we have seen repeatedly, we can navigate these situations just fine without supposing that the FIDE rules give rise to a non-moral FIDE obligation to make each move with one hand only, or that the FIDE rules have a point of view on how players should handle chess pieces. That is, we can navigate these situations just fine without supposing that the FIDE rules give rise to their own distinct domain of normativity, or even quasi-normativity.

But I want to drill down a bit further here. When FIDE adopted its rules, it did not generate a new sort of normativity. FIDE’s rules govern people’s behavior only to the extent they have reasons to follow them. But in some circumstances, we might want to bracket the question whether people have such reasons and ask the narrower question how they should behave if they do. We might, for example, want to envision how a chess match would go if it were governed by the FIDE rules, without supposing that the FIDE rules are in fact the rules that govern. There is no question that we can do this sort of thing, and we often do. And so we might think that there is a kind of quasi-normativity associated with the FIDE rules. Indeed, we might even go so far as to say that the FIDE rules have a point of view as to how people should play chess: they should play chess according to the standards specified in the rules.

No doubt, we can speak this way. The point I have been trying to make is that we don’t need to speak this way, and typically we don’t. But we shouldn’t avoid this sort of talk just because it is superfluous. We should avoid it because it risks a great deal of confusion, and I want here to explain why. The FIDE rules don’t exist in a vacuum. They have a history, and it is only in light of their history that anyone has any reason to care about them. The central fact in their history is, of course, that they were adopted by FIDE. When you couple that fact with some others, it may turn out that a particular chess match is governed by the standards expressed in the FIDE rules. As I said before, this seems apt to be the case when the chess match is part of a FIDE tournament, which the players chose to participate in.

But even in those circumstances, it is possible that play is not governed by the standards expressed in the FIDE rules. Instead, play may be governed by the standards that FIDE intended to adopt, which may or may not have been fully or accurately captured in the text that its officials had in front of them when they voted. Or it may be that play is governed by the standards that would best serve the purposes that FIDE had when it adopted the text, even if those standards are slightly different than the standards reflected in the text. Or it may be that some combination of these things is true, depending on the phase of play or the context of the match. This should all feel familiar, as these possibilities are also at play in debates over statutory interpretation. And here, just as there, the question which standards govern is a question about the normative significance of a set of social facts. The question is whether the act of adopting the rules subjected people to the standards expressed in the text of the rules adopted or to some other set of standards, differently connected to that event.48

Of course, there is no reason to think that the answer to this question will always be the same. Most lawyers (but by no means all) are open to the possibility that an act of legislation subjects people to standards that might not be fully or accurately captured in the relevant statute. With games, there might be reason to take the opposite view, as participants may have only the text of the rules available to them, with no further indicia of the intentions that lay behind the endorsement of that text. But one doesn’t have to search far for instances in which people contend that the rules governing a game are not quite the rules expressed in the text that purports to provide the official rules of the game.49

I don’t want to chase further examples down the rabbit hole. But I do want to say this. We started with the thought that we could bracket the question whether anyone has reason to pay attention to the FIDE rules, so that we could focus on the narrower question how they should behave if they do. The problem with that way of proceeding is that the reasons people have to pay attention to the FIDE rules affect the standards that govern their behavior. If we say that the standards expressed in the FIDE rules constitute FIDE’s point of view as to how chess should be played, we run the risk of missing the possibility that, in adopting those rules, FIDE actually subjected people to standards other than the ones the text expresses. So not only is it superfluous to say that FIDE’s rules have a point of view as to how chess should be played, it might also be misleading as to the normative consequences of the adoption of those rules. We’d do better to ask about the normative consequences directly, rather than to imagine into being a point of view that might or might not bear on what people should do. So the point is not only that we can navigate the situations imagined just fine without supposing that the FIDE rules generate a distinctive sort of quasi-normativity, but that we should.

VII. can we leave the fly-bottle?

Enough with the law-like practices. The time has come to see if we can find our way out of the fly-bottle. As I said at the start, the Hart-Dworkin debate is premised on the idea that legal practices give rise to distinctively legal rights, obligations, privileges, and powers, which can and often do differ from their moral counterparts. The debate is over how this distinctively legal domain of normativity, or quasi-normativity, is constituted. But if we have drawn the right lessons from the normative practices we’ve examined so far, then we should be skeptical that law generates a distinctively legal domain of normativity, or even quasi-normativity. The problem is that it sure looks like law does. After all, we talk about our legal obligations; distinguish them from our moral obligations; and recognize that the two can conflict, sometimes in terrible ways. All that would seem to imply that our legal practices generate a distinctively legal domain of normativity, or at least quasi-normativity. Unless, of course, there are other explanations for our inclination to talk that way.

I think there are other explanations, and my plan for this Part is to suggest several. But let me say up front that I don’t think these are the only explanations; I single them out because they strike me as relatively important ones. Let’s start by assuming that our legal practices are continuous with the normative practices we’ve examined so far, in that they do not generate their own distinctive domain of normativity, or even quasi-normativity. What I hope to show is that even if that is true, we would still have good reasons to talk about our legal obligations; distinguish them from our moral obligations; and recognize that the two can conflict, sometimes in terrible ways. If I am right about that, then the main reason we have for thinking that there is a distinctively legal domain of normativity, or quasi-normativity, will be seriously weakened, if not entirely vitiated.

Let’s take it from the top. Why would we talk about legal obligations if there is no distinctively legal domain of normativity or quasi-normativity? Well, nearly no one doubts that our legal practices can and often do generate moral obligations.50 Moreover, the moral obligations that our legal practices generate are, in an important sense, legal obligations. This is because they are obligations that we would not have but for our legal practices. We mark moral obligations by their source all the time. If you are going to miss a friend’s party to take care of your grandmother, you might explain your absence by saying that you have a family obligation. When you do that, you are not suggesting that your family has its own distinct domain of normativity, or quasi-normativity, associated with it. Rather, you are saying that you have an obligation of the ordinary moral sort, which arises in the context of your family.

We do the same with obligations that arise from promises. We call them promissory obligations, not because we are suggesting that promises generate a distinctive domain of normativity, but because we are signaling the source of the obligations in question. Since it’s often helpful to mark the sources of our obligations, there’s every reason to think we’d do the same with law. Which of our moral obligations would we mark as legal? Well, if we were trying to signal the source, the first criteria would surely be that the obligation was generated by our legal practices. But the case of promissory obligations suggests that we might label a narrower class of obligations “legal.” After all, we don’t label all the obligations that promises generate “promissory obligations.” To use our earlier example, if I promise to murder your cousin, you might be obligated to turn me in. That’s an obligation generated by my promise, but we wouldn’t call it a promissory obligation. Rather, we seem to restrict that label to those obligations that are generated in the way that the practice characteristically generates obligations; in the case of promising, we tend to restrict the label “promissory” to obligations to do as one promised to do. We might expect then that we would reserve the label “legal” for moral obligations that were generated in ways that legal practice characteristically generates obligations. As with promises, we’d expect that obligations to foil or frustrate morally iniquitous laws would not count as legal.51 But an obligation to drive no faster than the speed limit probably would. I won’t take up the task of drawing this boundary more precisely, because it’s not obvious that anything turns on how it’s drawn, and we might draw it in different ways for different purposes. For example, instead of using the label “legal” to signal the source of an obligation, we might sometimes want to use it to signal the sort of institutions that might call us to account for violating it. That means that we might tack the label “legal” on different obligations at different times, depending on what we are trying to communicate. Still, the general point remains: even if there were no distinctively legal obligations, we would nevertheless have good reason to regard some of our obligations as legal.

But would we have any reason to distinguish our legal obligations from our moral obligations? It might not seem like we would, as legal obligations are, on this picture, just a species of moral obligations. But that is not quite right, as it might well be helpful to distinguish moral obligations that are legal from moral obligations that are merely moral. An illustration might help here. Suppose we think that you are morally obligated to contribute your fair share to community efforts to support the poor. Now, suppose that the legislature in your state has adopted a special tax, with the plan that the funds raised will be spent on poverty relief. As it happens, the tax demands too little of people as wealthy as you, with the end result that the poverty programs will not be adequately funded.

My guess is that we’d think you were morally obligated to pay the tax. You are, after all, obligated to contribute to the sorts of programs that the revenue raised will fund, and this kind of coordinated effort is probably far more effective than any action you could take on your own. Moreover, this moral obligation would be a legal obligation, in the sense just explained, as its source would lie in legal practice. But your legal obligation to pay the tax would not completely displace your moral obligation to contribute your fair share in support of the poor, since, by hypothesis, the law demands only part of your fair share. That means there’s a moral remainder: you have a moral obligation to do more in aid of the poor than the law requires. Now, it’s possible that the existence of this moral remainder is a reason to increase the tax, but perhaps not. There are sometimes good reasons for the law to demand less of people than they are morally required to do. But we need not worry about that here. The point for now is that we can say all this and more without even once supposing that there is a distinctively legal domain of normativity, or even quasi-normativity, to which our talk of legal obligations refers.

My suspicion is that we can handle many of the distinctions we draw between our legal obligations and moral obligations in the fashion just suggested.52 But that still leaves the hard cases. We don’t just distinguish legal and moral obligations; we sometimes say that they conflict, and worse yet, we sometimes say people have a legal obligation to do things that are morally repugnant. Would we talk that way if legal practices did not generate a distinctively legal domain of normativity, or at least quasi-normativity? I think that we would. The ordinary cases of conflict aren’t so hard, as our moral obligations conflict with one another all the time. Suppose that I have promised my son that I will take him to the movies on Saturday afternoon, but, when the time arrives, my mother is ill and I am the only person available to take her to the doctor. Here, I have an obligation to take my son to the movies and an obligation to take my mother to the doctor. The second would surely take priority over the first, but both obligations would be genuine, and I would need to apologize to my son for missing the movie and make it up to him at some future date. These sorts of conflicts are routine, and there is no reason to doubt that our legal practices generate them. That is, there is no reason to doubt that moral obligations whose source lies in legal practice sometimes conflict with moral obligations whose source lies elsewhere.

That all seems straightforward enough. The tricky cases are the ones in which it looks like we have a legal obligation to do something morally repugnant. If there are genuine legal obligations to do things that people could not possibly have a moral obligation to do, then we are stuck in the fly-bottle, as that would imply that there is a class of distinctively legal obligations whose metaphysics we need to untangle. But I don’t think the fact that we sometimes talk as if there are such cases demands that we conclude that there is a distinctively legal domain of normativity, or even quasi-normativity. As it turns out, there would be good reason to talk that way even if there were not, though it will take a bit of effort to explain why.

In our everyday lives, we use a set of heuristics, or working theories, to identify our moral obligations. For example, we tell ourselves that we are obligated to keep our promises, and that’s a helpful guide to our responsibilities, even though a moment’s reflection reveals it as an obvious oversimplification. The same is true with law. We have a working theory about how our legal practices generate legal obligations. We don’t agree on all the particulars, and there are marginal cases we aren’t sure about. But we have enough of a working theory to recognize when our legal practices have generated an obligation, without much investigation into the particulars. Around here, for example, we tend to think judges are obligated to enforce duly enacted statutes. Of course, there are exceptions; judges shouldn’t enforce statutes that are unconstitutional, and there may be justiciability concerns that warrant declining to enforce particular statutes in particular cases. But in the main, when we see that a statute has been duly enacted, we move quite quickly to the judgment that it would be wrong for a judge to ignore it. From time to time, however, facts arise that confound our working theory. Sometimes, the legislature passes a statute that is so morally repugnant that no judge ought to enforce it.

Now you might think that confronting cases like this, or even just anticipating them, should lead us to qualify our working theory. Instead of holding that judges are legally obligated to enforce duly enacted statutes, we should hold that judges are obligated to enforce duly enacted statutes unless they are morally repugnant. But there are reasons to resist qualifying our working theory that way. To start, if judges take themselves to be free to ignore morally repugnant statutes, then they might abuse that freedom. They might ignore statutes that are merely morally misguided, but not so morally repugnant that they ought not to be enforced. Or worse yet, they might ignore statutes they simply dislike. Another reason is that we might not want to cultivate in our judges the sense that they sit as moral arbiters of what the legislature has done.53 The upshot is that even if we think there are statutes that are so morally repugnant they ought not to be enforced, we may not want to frame the moral responsibilities of judges that way. And so our working theory may sometimes suggest that a judge has a legal obligation to enforce a statute that is so morally repugnant that it ought not to be enforced. Because there is a gap between our working theory of the role of a judge and our case-by-case judgment as to what statutes should be enforced, we are apt to say that a judge has a legal obligation to do something that, considered on the merits, she morally ought not to do. But there are good reasons to construe the role of a judge that way.

I’ll illustrate this more concretely in a moment. But I want first to draw an analogy, which might sharpen our sense of the moral phenomenon that I’m trying to capture. As a parent, I regard myself as obligated to love and support my children. If I step outside my role, and think about things in a clearheaded fashion—or maybe if I just think about some other parent and her kids—then of course I recognize that there are limits to that obligation. Some people are moral monsters, who are not proper objects of love. And some people pursue projects so repugnant that they do not deserve support of any sort, not even from their parents. But when I step back into my role as parent, it would be a serious mistake to frame my obligation to love and support my children in a more qualified fashion. It would be a mistake for me to say to myself, “I will love and support these children so long as they are not morally repugnant.” And it would be an even worse mistake to say something like that to them. My children ought to think that my love and support is unconditional, even if, in the final analysis, it is not. And I should not even entertain the thought that my children might trigger the conditions under which I should and would withdraw my love and support. The upshot is that I construe my role as parent to impose moral obligations that I do not, strictly speaking, have. But there are overwhelming moral reasons to do so, and we are all better off for it.

I think this phenomenon is at work in law too, though the reasons for resisting the most accurate characterizations of people’s obligations are different. Let me circle back to the well-worn example of the Fugitive Slave Act to see if I can show you what I mean. The Act was as repugnant as any statute could be, and we might assume that no one could be under a moral obligation to do the terrible things that it demanded. But it nevertheless strikes us as appropriate to say that those subject to the Act had a legal obligation to do what it demanded. And that makes it seem like legal obligations just could not be moral obligations. But I don’t think that’s right.

Whatever its moral faults, the Fugitive Slave Act was a duly enacted statute, and our working theory is that federal marshals are legally obligated to enforce such statutes. Now from where we sit, we can plainly see that there was no moral obligation to enforce this particular statute, which means that there could not be a legal obligation to enforce it either, as we are supposing that legal obligations just are moral obligations generated by legal practice. We could respond to this conflict by revising our working theory about the role of federal marshals to hold that they have an obligation to enforce duly enacted statutes unless they are morally repugnant. But we have reasons to resist framing the responsibilities of federal marshals that way. As before, we don’t want federal marshals to think of themselves as moral arbiters of the acts of Congress; we want them to approach their jobs with humility and deference. But even more than that, we want them to experience conflict when they come across a statute they think they ought not to enforce. We’d like federal marshals to have the thought that it is not their role to decide whether a statute should be enforced, even if we also hope that they will recognize that there are occasions for stepping outside their role and declining to enforce a statute.

If I had to sum up the point I’m trying to make, I’d say this: we often have reasons to be morally obtuse about our moral obligations. We should regard ourselves as having a moral obligation to love and support our children unconditionally, even if, on cold reflection, we realize that no one could have such an obligation. And we should sometimes regard ourselves as subject to legal obligations to do things that we could not have a moral obligation to do, even though legal obligation is a species of moral obligation. The upshot is that even if our legal practices do not give rise to a distinctively legal domain of normativity, or quasi-normativity, there would nevertheless be good reason to see ourselves as subject to legal obligations to do things that we morally ought not to do.

Put all this together, and you can start to see a path out of the fly-bottle. We’ve been trapped inside because it seemed that we had to posit a distinctively legal domain of normativity, or quasi-normativity, to make sense of the way that we talk about law. But that’s a mistake. Even if legal practices did not give rise to a distinctively legal domain of normativity, or quasi-normativity, we would still have good reason to talk about our legal obligations; distinguish them from our moral obligations; and recognize that the two can conflict, sometimes in terrible ways. That means that the case for thinking that there is a distinctively legal domain of normativity, or quasi-normativity, cannot rest on the way we talk about law.

VIII. should we leave the fly-bottle?

I cannot prove that there are no distinctively legal rights, obligations, privileges, and powers. All I can do is invite you to try on that view and appreciate its virtues. I want to highlight three here. First, if we deny that law generates a distinctively legal domain of normativity, or quasi-normativity, then we can represent law as continuous with the other normative practices we have examined, like posting rules and making promises. At the least, anyone who would hold on to the idea that law generates its own distinctive domain of normativity, or quasi-normativity, must explain why law is different from these other sorts of normative practices. And that will be difficult to do. As we saw with the rental house cases, we can quite quickly generate complicated normative practices that we do not take to give rise to their own distinctive domain of normativity, or quasi-normativity. So complexity won’t be the answer. And our investigation of games taught us that the systematic nature of law won’t be either. It’s hard to see what’s left that might set law apart.

Second, the picture I am proposing is ontologically spare. Indeed, we might think of it as a kind of eliminativism, since it denies the existence of an entity—a distinctively legal domain of normativity, or quasi-normativity—that more traditional pictures presuppose.54 Moreover, the picture is spare in a welcome way. If we deny that legal practices give rise to a distinctively legal domain of normativity, or quasi-normativity, we relieve ourselves of the burden of explaining just what that domain is and how its content is constituted.55 In other words, we relieve ourselves of the burden of defending a position within the Hart-Dworkin debate. Of course, that’s not a sufficient reason to deny that law generates distinctively legal rights, obligations, privileges, and powers. But it is a welcome consequence of concluding so on other grounds.

Third, and perhaps most important, adopting the eliminativist view allows us to take advantage of the best that positivism and anti-positivism have to offer. Joseph Raz once pronounced Hart “the heir and torch-bearer of a great tradition in the philosophy of law which is realist and unromantic in outlook” on account of the fact that “it regards the existence and content of the law as a matter of social fact whose connection with moral or any other values is contingent and precarious.”56 The eliminativist picture I have proposed is realist and unromantic in much the same way.57 It allows that any particular law, or legal system, might be wholly devoid of moral merit, or worse than that, morally repugnant. It is a picture from which we can endorse a version of the positivist slogan: “[T]he existence of the law is one thing, its moral merit and demerit quite another.”58 And it is a picture that allows us to talk clearly about the difference between the way our legal practices are and the way that they ought to be.

What sets the eliminativist picture apart from positivism is its insistence that the question what rights, obligations, privileges, and powers are generated by our legal practices is a moral question. Of course, most positivists would agree that, ultimately, the question what anyone—layperson, lawyer, or legal official—ought to do in light of our legal practices is a moral question.59 They simply insist that our legal practices generate distinctively legal rights, obligations, privileges, and powers separate from whatever moral rights, obligations, privileges, and powers they generate. But that insistence is mysterious, for all the reasons we have explored so far. And it is not necessary, as we can maintain a realist and unromantic outlook toward law without it.

If positivism is at its best when it emphasizes that law is a social practice with contingent moral merit, anti-positivism is at its best when it explains the role that morality plays in that practice. People disagree about what the law requires. But they think that the law imposes determinate requirements, notwithstanding the fact that they disagree about what they are. In holding that legal obligations are a species of moral obligation, the eliminativist position vindicates Dworkin’s suggestion that people disagree about what the law requires because they disagree about the moral significance of our legal practices. The only difference is that Dworkin took morality to play a part in determining the content of distinctively legal rights, obligations, privileges, and powers, whereas the eliminativist position denies that there is a distinctive domain of legal normativity to be determined. But that difference does not affect the explanation Dworkin gave of the role that morality plays in legal practice.

The path out of the fly-bottle should be attractive to positivists and anti-positivists alike. It honors the core commitments of both, while avoiding the troubles that plague the efforts to vindicate one over the other. We might not have to leave the fly-bottle. But it is increasingly hard to see why we would stay.

Conclusion: The End Of Jurisprudence

I promised some history—and a new end for jurisprudence, too. So let me close with some thoughts about where jurisprudence has been and a suggestion for where it ought to go. For a long time, jurisprudence has been about the question posed in the Hart-Dworkin debate—not exclusively, of course, but primarily. The premise of that debate is that our legal practices generate distinctively legal rights, obligations, privileges, and powers; the question at issue is how they are constituted. In The Concept of Law, Hart proposed an answer. When Dworkin first came on the scene, he challenged Hart’s answer, but he did not take issue with the question. In The Model of Rules, Dworkin presumed that there is a distinctively legal domain of normativity.60 He simply argued that Hart was wrong about what it contained and how it was constituted. Hart had said that the law is composed of rules, but Dworkin argued that the law contains principles as well.61 Hart had said that legal rules were picked out by a socially constituted rule of recognition, but Dworkin rejected the thought that one could distinguish a community’s legal standards from its other standards by appeal to a master rule.62

By the time Dworkin wrote The Model of Rules II, he had a very different objection to Hart, an objection that prefigures the position defended in this Essay.63 Hart had argued that social rules are constituted by social practices. If men regularly take their hats off when they enter church, and they regard that regularity as supplying a standard of behavior, and they criticize one another for deviations from that standard, then there is a rule that requires men to take off their hats when they enter church.64 But Dworkin said that Hart had the connection between the practice and the rule all wrong. He wrote:

It is true that normative judgments often assume a social practice as an essential part of the case for that judgment . . . . But the social rule theory misconceives the connection. It believes that the social practice constitutes a rule which the normative judgment accepts; in fact the social practice helps to justify a rule which the normative judgment states. The fact that a practice of removing hats in church exists justifies asserting a normative rule to that effect—not because the practice constitutes a rule which the normative judgment describes and endorses, but because the practice creates ways of giving offense and gives rise to expectations of the sort that are good grounds for asserting a duty to take off one’s hat in church or for asserting a normative rule that one must.65

Take a moment to take that in: social practices do not constitute social rules, which we first identify and then endorse or reject. Rather, they help to justify normative rules, and the challenge is to figure out what rules they help to justify.66 Dworkin drove this point home a couple paragraphs later, while imagining a debate about whether parents must take bonnets off male babies in church.

It is true that [the churchgoers] will frame their dispute . . . as a dispute over what ‘the rule’ about hats in church requires. But the reference is not to the rule that is constituted by common behavior, that is, a social rule, but the rule that is justified by common behavior, that is, a normative rule. They dispute precisely about what that rule is.67

The point in these passages is the point I’ve been pushing all along. Our social practices don’t give rise to new kinds of normativity; rather, they warrant new normative judgments of the old familiar kind. This is true even when the social practice in question involves positing rules. As Dworkin put it:

The social rule theory fails because it insists that a practice must somehow have the same content as the rule that individuals assert in its name. But if we suppose simply that a practice may justify a rule, then while the rule so justified may have the same content as the practice, it may not; it may fall short of it, or go beyond it.68

We have seen that this is true time and again with the social practices we have interrogated. The rules posted in a rental house might or might not generate obligations to do as directed. A promise might or might not generate an obligation to do as promised. Writing down rules for a game might or might not generate an obligation to play the way they say. And enacting a law might or might not generate an obligation to do as it demands. It all depends on the content and context of the practice, and the obligations that are justified by the practices can go beyond or fall short of the obligations that are posited within them. But it is the obligations that are justified by our practices that we have reason to care about; any interest we have in the obligations posited within them is, at best, derivative.

To my thinking, Dworkin’s remarks in the opening pages of The Model of Rules II constitute the most incisive contribution that anyone has made to the Hart-Dworkin debate. But almost as soon as he wrote them, he forgot them, or maybe he never quite appreciated the importance of what he’d said in the first place.69 By the time he wrote Law’s Empire, Dworkin was back in Hart’s framework, trying to work out the relationship between our legal practices and a distinctively legal domain of normativity. To be sure, he had a different answer than Hart. Dworkin argued that morality played a role in determining our distinctively legal rights, obligations, privileges, and powers. But he took for granted that there was a distinctively legal domain of normativity.

It was only at the end of his life that Dworkin came to see his mistake. In Justice for Hedgehogs, he said that he never should have accepted a “two-system” picture, on which “‘law’ and ‘morals’ describe different collections of norms.”70 Instead, he said, he should have insisted on a one-system picture, which would render law as a branch of morality. Roughly, Dworkin said, law is the branch of morality that deals with rights that people are entitled to enforce on demand in court.71 He might have added, though he didn’t, that those rights are not constituted by our legal practices, but rather are justified by them. The task of courts is to figure out what rights and obligations are warranted by our legal practices. And those rights might go beyond or fall short of the rights and obligations that are posited within them.

Dworkin did not develop the one-system picture much, and there are some questions about how best to interpret what little he said. Jeremy Waldron has ventured some thoughts about how the one-system view should be fleshed out, which are broadly consistent with what I have said here.72 But he emphasizes an essential point that I have not. The claim that legal rights are a species of moral rights does not imply that the law is morally perfect, or even that we have the legal rights it would be best for us to have. Waldron draws this point out with an analogy to promising. Some promises are morally regrettable—they ought not to have been made—but they might nevertheless have moral significance. “In the case of the man who promised to cover up an illicit affair,” Waldron suggests, “he might have an obligation to warn the erring husband before speaking truthfully to the innocent wife or he might have an obligation to avoid situations where the truth might be required of him.”73 The first sounds better to me than the second, but the point is that morality is sensitive to our history, even when it is morally compromised. Law poses this problem with a vengeance. Rare is the statute that says just what it morally ought to have said. But morally compromised legislation might nevertheless have moral significance, and courts are often called on to decide just what that significance is.74

There is more wise counsel in Waldron’s discussion of the one-system picture. But I think even he would agree that the ideas involved have received their most important expression in the work of Mark Greenberg, who has recently articulated what he calls the moral impact theory of law.75 According to Greenberg, our legal rights, obligations, privileges, and powers are the moral rights, obligations, privileges, and powers generated by our legal practices in what he calls “legally proper” ways.76 This is very close to the view that I’ve endorsed here. Greenberg also rejects the idea that our legal practices generate a distinctively legal domain of normativity, or quasi-normativity. And because he does, his view has the virtues of the eliminativist position I have outlined.

But I worry that Greenberg is not fully out of the fly-bottle, or perhaps that he is leaving one fly-bottle only to buzz into another. And before I close, I want to see if I can explain what makes me reluctant to frame the task of jurisprudence as Greenberg does. Like Hart and Dworkin before him, Greenberg takes himself to be telling us how the content of the law is constituted.77 On Greenberg’s picture, the content of the law consists in all the legal rights, obligations, privileges, and powers in force in a given jurisdiction at a given time.78 The great advance in Greenberg is that he does not think of these legal rights, obligations, privileges, and powers as distinctively legal, for they are moral, too. But he holds on to the view that our legal practices make something—the content of the law—whose metaphysics we must unravel.

I think that is a mistake. And early in his career—before he lapsed back into Hart’s framework—I think Dworkin would have too. Responding to Hart, he wrote:

I hope to persuade lawyers to lay the entire picture of existing law aside in favour of a theory of law that takes questions about legal rights as special questions about political rights, so that one may think a plaintiff has a certain legal right without supposing that any rule or principle that already ‘exists’ provides that right.79

Greenberg signs on to part of this—the most important part. Though he puts the point differently, he accepts Dworkin’s suggestion that whether a plaintiff has a legal right is a question of political morality. But Greenberg holds on to the idea that there is a body of “existing law,” which encompasses the entire set of legal rights, obligations, privileges, and powers in force in a legal system at a given time, and he aims to say just what it is that makes something a member of that set. At the end of his life, Dworkin seemed to share that ambition. As I mentioned before, Dworkin suggested that law is the branch of morality that deals with rights that people are entitled to enforce on demand in court.80 Greenberg objects to that characterization of the law, because it “rule[s] out . . . the possibility of legal obligations that the courts . . . should not enforce.”81 That strikes Greenberg as a mistake because, he says, “it is a familiar idea that the President and Congress may have legal duties that the courts should not enforce.”82 Instead, Greenberg suggests that the content of the law consists in all the moral rights, obligations, privileges, and powers generated by the actions of legal institutions in legally proper ways.83

What is at stake in this dispute? Not very much, and both Greenberg and Dworkin acknowledge that. After observing that some lawyers talk as if the Constitution confers legal rights that courts nevertheless should not enforce, Dworkin says, “[t]his is indeed an available way to describe the situation: no one would misunderstand.”84 He goes on to say that he prefers to employ a “different vocabulary,” which construes the constitutional rights that courts should not enforce as “political but not legal.”85 He adds a few words in favor of his preferred vocabulary, but once he allows that no one would misunderstand if we said the opposite, it is hard to see why we should insist on saying things his way. Greenberg, for his part, allows that “the distinction[] between legal and non-legal obligations” is “less important” on his theory than on others.86 This is because the question what obligations we classify as legal has no bearing “on what we take our genuine obligations to be.”87 But Greenberg still insists that his vocabulary is superior to Dworkin’s. He says that “an account of law should help us to explain why courts should enforce some rights and not others.”88 Dworkin’s account can’t meet that test, since on his view legal rights just are the rights that courts should enforce.

But that does not give us much reason, if any, to side with Greenberg. If we employ Dworkin’s vocabulary, then it is true that we will not be able to explain why a court should enforce a right by observing that the right in question is a legal one. We would instead have to point to the underlying considerations that make it the case that the court should enforce the right. If instead we employ Greenberg’s vocabulary, then saying that a right is legal might help explain why courts should enforce it. But that will be true only if Greenberg’s use of the label “legal” tracks the underlying considerations that warrant judicial enforcement. That is, it will be true only if (absent countervailing considerations) courts should enforce the rights generated by legal institutions in legally proper ways. So whether we follow Greenberg or Dworkin, the underlying considerations ultimately do all the explanatory work. Courts should enforce whatever obligations they should enforce, and whether we use the label “legal” to report the conclusion of that inquiry (as Dworkin suggests) or an intermediate step (as Greenberg prefers) in no way affects the substance of the decisions that courts have to make. The only thing at stake is the way that we talk about those decisions, and I see no reason to think that one vocabulary will always be superior to the other.

The idea that we must settle on a single characterization of the law of our community strikes me as a hangover from the picture of “existing law” that Dworkin long ago proposed that we reject. If we suppose that, at any given time, there exists a body of law that encompasses all the legal rights, obligations, privileges, and powers in force in a legal system, then it makes sense to ask what is a part of it and what is not. But that picture has been the source of so much confusion that I think Dworkin was right to suggest that we leave it behind.89 And there is little cost to doing so. A crude version of the picture forms part of many laypeople’s understanding of law. They suppose that our legal rights and obligations are printed in the books that line the shelves of law libraries, just waiting for lawyers to look them up. But that picture is too simpleminded to survive even the first few days of law school. And though we could make the picture more sophisticated, there is little return to doing so. The thought that there is an existing body of law that comprises all the legal rights, obligations, privileges, and powers in force in a legal system plays no role in legal practice. Lawyers do not consult the law to ascertain what legal obligations people have. Rather, they read records of their community’s legal history—statute books, case reports, and the like—and then they construct arguments about what obligations people have as a result.

To be clear, I do not object to talking about what the law requires. What I object to is the supposition that there is a single entity called the law to which all such talk refers. As I said before, I am happy to allow that some of our moral rights, obligations, privileges, and powers are helpfully labeled “legal.” But there are many helpful ways to use that label. If we want to signal the source of our rights, then Greenberg’s vocabulary might be useful. But if we want to signal which institutions have responsibility for vindicating our rights, then it might be helpful to join Dworkin in distinguishing legal rights from political rights. The ambition to characterize just one set of things as the law of our community requires that we cast lots with Greenberg, or Dworkin, or someone yet to come, because now that the issue has been mooted, there will be many proposals for how to distinguish our legal rights from our merely moral rights. But I do not see why we should feel pressure to pick just one way to characterize the legal domain. That strikes me as an invitation to another fly-bottle.

As I suggested earlier, we characterize the legal domain in different ways for different purposes. Those who take a practical interest in law might construe it differently depending on whether their concerns are primarily moral or prudential.90 And even if we restrict our attention to morality, we might find different characterizations of the law suitable for different occasions, or perhaps for people occupying different roles. It is not obvious to me, for example, that a county clerk, a district judge, and a Supreme Court Justice should all think about what the law requires in just the same way.91 When we have a project in mind, we can ask whether it matters what notion of law we employ, and which one to use if it does. But these are practical questions, not metaphysical ones.92

The time has come for jurisprudence to drop the metaphysics and take up morals. The question that jurisprudence should aim to answer is how our legal practices affect our moral rights, obligations, privileges, and powers.93 The metaphysical question posed in the Hart-Dworkin debate was a distraction; we have no good reason to think that our legal practices generate a distinctively legal domain of normativity, or quasi-normativity, whose metaphysics we must unravel. But the moral question is vital; it is contested every day, in court and out, with serious consequences for peoples’ lives. Of course, philosophers have thought about the moral question. There are answers all over the jurisprudence literature: in Dworkin’s comments on coercion;94 in Raz’s reflections on authority;95 in Shapiro’s paean to planning;96 and in Waldron’s discussions of disagreement.97 But rarely have these answers—and many more that I might have cited—been engaged as answers to the moral question, because they are usually attached to a metaphysical thesis, which draws all the attention.98 For far too long, the end of jurisprudence has been answering the question posed by the Hart-Dworkin debate. That debate is at its end, but jurisprudence is not. Jurisprudence does, however, need a new end. And we would do well to worry more about the moral consequences of our legal practices.