The One-System View and Dworkin’s Anti-Archimedean Eliminativism

Abstract

Many of Dworkin’s interlocutors saw his ‘one-system view’, according to which law is a branch of morality, as a radical shift. I argue that it is better seen as a different way of expressing his longstanding view that legal theory is an inherently normative endeavor. Dworkin emphasizes that fact and value are separate domains, and one cannot ground claims of one sort in the other domain. On this view, legal philosophy can only answer questions from within either domain. We cannot ask metaphysical questions about which domain law ‘properly’ belongs in; these would be archimedean, and Dworkin has long argued against archimedeanism. The one-system view, then, is best understood as an invitation to join Dworkin in asking moral questions from within the domain of value. Finally, I argue that Dworkin’s view can be understood as a version of ‘eliminativism’, a growing trend in legal philosophy.

Introduction

In Justice for Hedgehogs, Dworkin put forward an account of law that he referred to as the ‘one-system view’.¹ On that view, law is a branch of morality, and so all legal questions are really moral questions. Many of his interlocutors saw this as a radical shift.² I will argue here that it is better seen as a different way of expressing his longstanding view that legal theory is an inherently normative endeavor. We should read the argument in Justice for Hedgehogs in light of, and as continuous with, his arguments in earlier work. Law’s Empire can be read as saying that legal questions are all moral questions. What Dworkin means by law is: those rights and duties that actually justify coercion.³ I will trace this way of thinking about law through Dworkin’s career, to show that the one-system view is not new, but rather a distillation of what he always believed.

As well as seeing it as a shift, many of his fellow legal theorists have argued that it is an odd or undefended position: why assume law belongs in the moral domain? His argument for this is puzzling and brief. Thus, my second aim will be to unpack and reconstruct the argument for the one-system view, in order to provide a defensible reading of it, drawing on earlier arguments in Justice for Hedgehogs. There, Dworkin is concerned to emphasize a Humean point: the separation of fact and value. Fact and value are separate domains, and one cannot ground claims of one sort in the other domain. This distinction is crucial to Dworkin’s one-system view.

I will argue that, in light of the fact-value distinction, legal questions can only be moral or factual questions. Dworkin tries to show how they cannot be factual questions throughout his career, through his semantic sting argument, and other arguments against descriptive approaches.⁴ If legal questions aren’t descriptive, factual ones, what else could they be? They could be moral, and this is what Dworkin ultimately concludes. There is a third kind of thing they could be, however. One might think legal philosophers are asking questions about whether law is really a moral phenomenon or not: questions about its ultimate nature, about whether it would be right to place it in the domain of fact or value. Those questions ask us to step outside of a domain and make claims about a phenomenon from some supposed archimedean point. Dworkin has argued strongly against archimedeanism.⁵ I believe his claims in Justice for Hedgehogs can be made sense of if we think of them as anti-archimedean claims about law: we cannot step outside of any domain to talk about what law ‘really is’, so there is no question of whether it truly belongs in the domain of fact or value.

With that background understanding of Dworkin’s opposition to metaphysics, we can see why Dworkin places law in the moral domain. I will argue that the claim that law belongs in the domain of morality is not a claim about law’s true nature, but rather an invitation to join Dworkin in asking fruitful moral questions that can be answered from within the domain of morality.

Finally, I argue that this one-system approach to legal philosophy can be understood as a version of eliminativism, a growing trend in legal philosophy that says that there is no distinct concept of law, or no answerable question about the nature of law.⁶

Thus, this paper aims to do three things: to interpret Dworkin’s thought in a way that shows the one-system view to be a coherent part of his philosophy rather than a radical shift; to give at least a partial defense of Dworkin’s surprising claim that law is part of morality; and lastly, to show that that approach to legal philosophy can be understood as part of an important eliminativist trend in the field.

The One-System View: New or Old?

The One-System View

In Justice for Hedgehogs, Dworkin claims that we cannot accept the ‘two-systems view’, which treats law and morality as separate domains. Rather, we must adopt the ‘one-system view’, according to which law is a branch of morality. This means legal questions must be understood as moral questions. It is not that answering legal questions requires taking a moral stance, but that to understand legal questions at all we have to frame them as questions in morality.

On the two-systems view, ‘there is no neutral standpoint from which the connections between these supposedly separate systems can be adjudicated’.⁷ So, if we are to answer the question of how they are connected, Dworkin suggests that it must be answered as a moral or a legal question.⁸ But he says that the choice to treat it as either a legal question or a moral one will necessarily be question-begging. In either case, we will have to make an assumption about whether morality is part of the material we are permitted to examine. Yet, whether morality is in or out is precisely what is at stake. According to Dworkin, then, ‘The two-systems picture therefore faces an apparently insoluble problem: it poses a question that cannot be answered other than by assuming an answer from the start’.⁹

Hume and the Domains of Fact and Value

Dworkin’s ‘one-system’ account of jurisprudence in Justice for Hedgehogs was not, I will argue, a dramatic shift, but a different way of expressing his longstanding emphasis on legality and the thought that legal theory is an inherently normative endeavor. To understand this, I want to start not at the beginning of Dworkin’s career, but at the end. At the end, he framed his project in Humean terms that, in my view, greatly illuminate what he was doing.

Justice For Hedgehogs is a defense on a grand scale of a particular view of value, according to which it is truth-apt,¹⁰ it is an independent domain,¹¹ and it is unified.¹² These points are interrelated, but I will be particularly focused here on the idea that value belongs to a domain of its own, according to which moral claims can only be defended by further moral arguments. This idea comes from Hume’s famous claim that one cannot derive an ought from an is,¹³ which Dworkin treats as a foundational part of his argument. For some, this means that there can be no truth about morality, because empirical observation is all we have. But Dworkin argues that ‘Hume’s principle, properly understood, supports not skepticism about moral truth but rather the independence of morality as a separate department of knowledge with its own standards of inquiry and justification’.¹⁴ Here is not the place to defend Dworkin’s moral realism. I am interested not in whether he is right that the moral domain is one in which we can seek truth, but rather in the structure and foundation of his theory. For Dworkin, the Humean separation between fact and value is fundamental. If there is to be truth in either realm, it must be answerable to norms within that realm.

I read Dworkin as saying that there are two basic domains: fact and value.¹⁵ Each is freestanding and has its own norms of truth. Dworkin sometimes uses the term ‘domain’ more freely, talking about cosmology or morality as ‘domains’ of their own, but I believe these must each be properly understood as subdomains of the primary domains of fact and value.¹⁶ The Humean foundations of his argument support the idea that the two core domains are fact and value, and that any subdomains have to be situated within one of these domains. Interpretation is the method appropriate for the domain of value – though, since Dworkin argues that all value questions are answered through the interpretive method, he sometimes calls the domain of value the ‘interpretive domain’.¹⁷

Dworkin makes the further claim that we cannot step outside the domain of value to ask questions about whether morality is truth-apt or what kind of thing it is: ‘There is no neutral scientific or metaphysical plane on which we can stand finally to adjudicate which of different views about equal concern or liberty or democracy or any other opinion about right or wrong or good or bad is the best or true one’.¹⁸ This anti-archimedean point is very important, as I will discuss in more depth below.¹⁹

Law and Justified Coercion

Let’s now examine some of Dworkin’s earlier ideas. Dworkin defends an account of law he calls law as integrity. Law as integrity provides an account of why we look backwards for justification for the use of force.²⁰ This question of why we should be concerned with our political history is central in Law’s Empire.

[Integrity] insists that the law – the rights and duties that flow from past collective decisions and for that reason license or require coercion – contains not only the narrow explicit content of these decisions but also, more broadly, the scheme of principles necessary to justify them.²¹

This quotation is important, not only for providing an account of integrity, but for Dworkin’s clarification of what he means by law: rights and duties, ones that actually license coercion. These are real moral rights and duties, not some special category of the legally valid. When he gives an account of what the law ‘is’, he is doing a thoroughly normative project of figuring out what actual moral rights our past decisions give us.²²

In other words, he starts from the assumption that when we talk about law we are talking about justified power, and the only question is what makes it justified, not whether it is.

A conception of law must explain how what it takes to be law provides a justification for the exercise of coercive power by the state, a justification that holds except in special cases when some competing argument is specially powerful.²³

This starting point is important. Many positivists have objected to it, arguing that they are not interested in justifying coercion, but in describing what law is. But the fact that this is Dworkin’s starting point can tell us something about how he understood his project: it was not a description of what the law is, but a normative inquiry all the way down. In his Justice For Hedgehogs language, we might say that questions about law belong in the moral domain.²⁴

The domain terminology helps us to see that the shift in Justice For Hedgehogs really represents no shift at all. Law’s Empire can be read as saying that legal questions – at least the legal questions Dworkin sought to answer – are all moral questions. They are all questions about the justification of coercive force. The debate about law and morals only makes sense as one between rival political theories, where each aims to show our political practices as justifying coercion.²⁵ The whole project of figuring out what the law is begins and ends in normative territory: ‘[L]egal argument takes place on a plateau of rough consensus that if law exists it provides a justification for the use of collective power against individual citizens or groups’.²⁶

My interpretation fits Dworkin’s own account of his position in Justice for Hedgehogs. To begin with, he notes that Justice for Hedgehogs aims to supplement Law’s Empire and Justice in Robes, not replace them.²⁷ This indicates that he sees the project as continuous with his earlier work. He says that in his early work, he tried to fit within the ‘orthodox two-systems picture’, according to which ‘law and morals are different systems of norms and … the crucial question is how they interact’.²⁸ For that reason, he says, he defended interpretivism within this framework in his ‘The Model of Rules I’ (1967). But, in Justice for Hedgehogs, reflecting back on this earlier work, he says he ‘soon came to think’ that approach was problematic, and began to see things a different way. By ‘soon’ he really does mean ‘soon’: the footnote to that claim indicates that he already saw the problem by the time of ‘The Model of Rules II’ (1972).²⁹ This shift, then, occurred before even Law’s Empire, and so I think we can reasonably read that book in light of this ‘new’ one-system idea.

However, it is true that the one-system view isn’t made explicit until much later. He says: ‘I did not fully appreciate the nature of that picture…or how different it is from the orthodox model, until later when I began to consider the larger issues of this book’.³⁰ So, we should think of Justice for Hedgehogs as a shift in presentation, a way to better express what he had been grasping at all along – or at least since ‘The Model of Rules II’.

When he says the differences became fully apparent when considering the larger issues of the book, I believe this refers to his overarching argument about the independence of the domain of value from that of fact – his idea that morality is truth-apt and self-sufficient, and that arguments about morality can be grounded only in morality.

Where Does Law Belong?

So, it might be that Dworkin has long held a view that places law squarely within the domain of value, and that the one-system terminology is nothing more than a new way to express this. But the next, rather glaring, question is this: why should we care that that is what Dworkin has long maintained? Is it a plausible view at all? Why does Dworkin think we should place law within the domain of value and not fact? I will reconstruct his arguments for this below, in light of what I see as his most fundamental commitments.

The Domain of Fact

Here we can turn to one of Dworkin’s most notorious arguments: the semantic sting. ‘Semantic theories suppose that lawyers and judges use mainly the same criteria… in deciding when propositions of law are true or false; they suppose that lawyers actually agree about the grounds of law’.³¹ Holding a semantic theory leaves one vulnerable to the semantic sting: that is, semantic theorists are committed to the idea that argument requires shared ‘criteria for deciding when our claims are sound’.³² This is, Dworkin argues, too simplistic an idea of what it is to disagree. We sometimes disagree when we do not share criteria of application for the concept, but dispute about what would make it best.

Disagreement in law, Dworkin says, is of this latter sort: we do not agree on shared criteria of application. Yet we are arguing about the same thing, not talking past each other. Our theories of law, then, cannot be semantic ones. They must be interpretive; they must ‘try to show legal practice as a whole in its best light, to achieve equilibrium between legal practice as they find it and the best justification of that practice’.³³ They do so by starting with the preinterpretive data about what counts as a legal institution or practice, which we may not agree about entirely, but must largely share.³⁴ Of course, we need to have at least a provisional way of identifying what counts as law.³⁵ But, as I will argue, we can settle this question by looking to our practices and being clear which of those practices we mean to invoke by the use of a given concept.

The semantic sting argument leads Dworkin to reinterpret positivism as a moral theory, in the form of ‘conventionalism’. Positivism cannot be a semantic theory, so we must shift to understanding it, and any other theory that aims to be a genuine competitor, as a normative one. There is no other option: ‘since theories of law cannot sensibly be understood as linguistic analyses, or neutral accounts of social practice, I can think of no way to defend positivism or any other theory of law except by appeal to political morality’.³⁶

Hart finds Dworkin’s reasons for this claim unclear. Why can’t we have an external, purely descriptive, account of a person who takes the internal point of view?³⁷ Dworkin’s first version of the argument, the semantic sting, equates descriptive theorizing with semantic theorizing. He then criticizes semantic theories, and concludes that normative theories are all we have left. But this leaves open the response that positivists do not hold a view that depends on criterial semantics.³⁸

Dworkin, in a later response to Hart’s Postscript, broadens the argument to capture other plausible accounts of descriptive theorizing. He again suggests the semantic interpretation, but reminds us that there are no shared criteria for concepts like justice, though of course our semantic practices rule out certain uses of the concept.³⁹ The second possibility is an approach similar to how we think about natural kinds. That is, the theory can be descriptive because ‘the correct attribution of the concept is fixed by a certain kind of fact about the objects in question, facts that can be the object of very widespread error’.⁴⁰ But political concepts are not like this. They do not have some inherent essence or structure that is discoverable by a non-normative process.⁴¹ The third suggestion is that legal philosophers are engaged in a historical generalization about applications of the concept. However, this is not distinctly philosophical, but rather anthropological. And the claim that many people have thought something does not support the truth of that thing. Philosophers are claiming not that many people have thought law was a certain way, but that it is that way.⁴²

Dworkin considers each of these three approaches as an account of what Hart was doing. He says that positivists have failed to give any clear alternative account of their methods and how they can be descriptive.⁴³ Law has no DNA, and so the natural kinds approach is absurd. Hart and others never appeal to data that would make the historical generalization claim plausible. So Dworkin says, despite Hart’s protests: ‘I still think that my understanding of the enterprise in The Concept of Law is the best available’.⁴⁴

Dworkin’s ongoing series of objections to the descriptive approach constitutes his extended argument against placing law in the domain of fact. Dworkin’s approach was to reject in turn various possibilities for descriptive jurisprudence. This approach left space for his interlocutors to retreat – they could abandon the semantic territory and move to other, safer-seeming, descriptive ground. But trying to show that there is no acceptable version of descriptive analysis by proposing and rejecting examples is a boundless task. What Dworkin needed to do was demonstrate that no account of what law ‘is’ in a metaphysical sense will succeed. I will argue below that his anti-archimedean arguments play this role. Any metaphysically descriptive theory of law will fail because such a theory will require an impossible archimedean perspective.

Nonetheless, I do not think that Dworkin is entirely successful in heading off claims that law, or rather, questions about law, can ever belong in the domain of fact. On the contrary, there are many factual claims that we can make about law. Some of these are of limited interest: ‘What words are contained in this statute?’ is a descriptive question relevant to law. But we can use the tools of sociology, economics, and other descriptive methodologies to answer a vast number of important and interesting questions. A few examples of questions that would belong in the descriptive domain are: ‘What effect will repealing this bill have on the number of abortions that take place?’; ‘Do judges’ political leanings affect their judicial decisions?’; ‘What are the rates of compliance with the prohibition on drugs?’. We ought not to dismiss such questions. I believe Dworkin’s rejection of descriptive theorizing about law stems primarily from the fact that he is a philosopher. As a philosopher, he does not have the tools to answer questions of the sort just mentioned. The questions on the table, philosophically speaking, do not belong in the descriptive realm. Thus, having rejected placing law in the factual domain, Dworkin turns to morality.

The Domain of Value

In light of his objections to placing law in the factual domain, Dworkin’s ultimate view is that law belongs in the domain of value. But what does it mean to place law in that domain, and what justifies Dworkin’s claim that that is where it belongs? Dworkin’s argument for the one-system view is odd, and frustratingly brief.⁴⁵ In Justice for Hedgehogs, Dworkin argues that there was a ‘turn in Anglo-American jurisprudence, by positivists beginning in the nineteenth century, to the surprising idea that the puzzle about law and morals is neither a legal nor a moral problem but instead a conceptual one: that it can be settled through an analysis of the very concept of law’.⁴⁶ This led to a state of affairs in legal philosophy where it has been assumed on both sides that conceptual analysis is the way to answer this question: that ‘[w]e can excavate the nature or essence of that concept without making any prior legal or moral assumptions’.⁴⁷

But, Dworkin says, this approach is misguided. ‘We cannot solve the circularity problem of the two-systems picture through an analysis of the concept of law unless that concept can sensibly be treated as a criterial (or perhaps as a natural-kind) concept. But it cannot be’.⁴⁸ This, again, is his broadened version of the semantic sting, meant to sting any kind of descriptive approach.

Then comes what seems to be a pivotal move. Having argued that there is no way to do the conceptual project as a purely descriptive one, he says that we must shift to interpretivism instead.⁴⁹ We shift to seeing the opposing jurisprudential views as normative political theories.⁵⁰ He says we replace the old two-system view ‘with a one-system picture: we now treat law as a part of political morality’.⁵¹

As noted in Section II C above, this is not really a new approach. Dworkin often simply began by placing law in that domain, rather than presenting any argument for doing so. Indeed, Raz criticized him for this.⁵² And Lawrence Sager has objected that the move to the one-system view that places law within the domain of value is just as question-begging as the alternative, and thus no solution at all.⁵³

In my view, a more charitable interpretation is that he gave no argument for it because, on his account, none is needed. To further make out this surprising claim, however, we must take a detour into Dworkin’s views on metaphysics.

Against Metaphysics

I mentioned above Dworkin’s anti-archimedean stance, but now it is time to examine it more carefully. Arthur Ripstein calls Dworkin’s anti-archimedeanism ‘[t]he most significant and most central theme of Dworkin’s work’.⁵⁴ In my view it is part and parcel of a general anti-metaphysical outlook that Dworkin held.⁵⁵ Archimedeanism, for Dworkin, denotes a view that ‘purport[s] to stand outside a whole body of belief, and to judge it as a whole from premises or attitudes that owe nothing to it’.⁵⁶ In ‘Objectivity and Truth: You’d Better Believe it’, Dworkin argues against the view that we can step outside of morality to talk about its nature and what moral claims really are. Instead, the only arguments that can be raised are internal ones. This view is echoed in a number of later works.⁵⁷ Dworkin says that it is a mistaken view that, while regular people debate about the genuine importance of a value, philosophers aim to ‘provide accounts of what legality or liberty or equality or democracy or justice or community really is’,⁵⁸ while remaining neutral about the normative controversies.

Archimedeans, Dworkin argues, add a series of ‘external’ claims to ordinary moral claims. These are claims like: this is a report ‘of how things really are out there in an independent, subsisting, realm of moral facts.’⁵⁹ These external claims ‘take up positions on such metaphysical questions as whether there are moral properties in the universe, and, if so, what kind of properties these are’.⁶⁰ Dworkin says that all such statements can be translated into first-order moral claims, leaving nothing sensible in the category of ‘metaethics’. The idea that morality is part of the fabric of the universe, and other such claims, are just ‘inflated, metaphorical ways of repeating’⁶¹ the ordinary moral claim. The archimedeans’ fallacy ‘is to suppose that some sense can be assigned to the supposedly metaphysical claims that is not itself a normative sense’.⁶²

Is the same thing wrong with asking ‘what kind of thing is law?’ I believe Dworkin would say yes. Though his usual target is metaethics, the point generalizes to metaphysical claims. In the case of metaphysics, he says that there is no difference between the statement that there are mountains and they would exist even without humans, and the statement that ‘mountains are part of Reality As It Really Is’.⁶³ Dworkin cannot see, he says ‘what sense you can make of the second proposition, no matter how many capital letters you pack into it, that makes it mean something significantly different from the first proposition’.⁶⁴ This appears to be a rejection of metaphysical questions of the sort ‘what kind of thing are mountains [or law] in their true nature?’

It might seem that Dworkin’s anti-archimedeanism only applies to the domain of value. But I believe he is an anti-archimedean about the realm of fact too – meaning he rejects metaphysical questions about that realm the way he rejects metaethical questions about the realm of value. True, there are different norms appropriate to different domains. In science, we have access to bare truths which give our conclusions greater solidity.⁶⁵ But that does not mean that scientists must step outside their scientific domains to answer questions about what bacteria or mountains ‘really’ are. Indeed, scientific conclusions are also circular to some extent, according to Dworkin, given that we rely on one part of our science to check other parts of it.⁶⁶ So even if we are not interpretivists about science the way we are about value, it is still right to say that both domains ask their practitioners to remain within the relevant domain.

A metaphysical interpretation is, however, encouraged by some things he says. He insists that even though Hart takes his project to be descriptive and Dworkin takes his to be interpretive, they agree that they are studying ‘the very concept of law’.⁶⁷ And in reference to his own project, he asks: ‘In what way can deciding what law should be like help us to see what, in its very nature, it actually is?’⁶⁸

Phrases like ‘actually is’ and ‘nature’ seem to suggest he’s doing the same project as the analytic jurisprudes, whose aim is to access metaphysical truths.⁶⁹ But if Dworkin is understood this way, he would be taking an archimedean stance. Claims about the true nature of law could not be located in either the descriptive or the normative realm. They are claims about whether law belongs in the descriptive or normative domain. We can’t answer a question about law’s true nature from within the moral or the factual domain. Such accounts, then, have to be archimedean ones, striving to step outside of both fact and value to talk about what something really is: whether it ‘really’ belongs in the domain of fact or value.

We can imagine Dworkin saying something similar about law to what he says about mountains. We can ask the geological question about whether there are mountains, and we answer that question according to the norms of science. But we can’t answer the further question of whether mountains really exist, or what kind of entity they ‘really’ are, beyond the geological description. Turning to law, we might say that we can answer various moral and factual questions about law, but we cannot answer any further question about what law ‘really’ is.⁷⁰

Given the importance of Dworkin’s anti-archimedeanism and his rejection of metaphysics, it seems implausible to interpret him as making any claim of this sort, whether with respect to morality or to law. Dworkin is steadfast in his commitment to the view that we cannot access an archimedean point outside of ourselves that could tell us about what kind of thing morality really is. In light of this, Dworkin should not be read as saying that we can step outside the domains of fact and value to see what kind of thing law really is. His project must be one that can be located within one or the other domain: fact or value. It can’t be the archimedean one, asking questions about those domains or external to them.

It might be said that my rejection of any metaphysical ambitions in Dworkin’s work is too quick. He does, we might argue, have some metaphysical commitments, even if these are thin. For example, he seems committed to the view that, whatever else law is, it is the kind of thing that requires us to justify the coercion it exercises. But Dworkin’s points about the justification of coercive force are all first-order claims about what we must do and think about these institutional processes. Judges must act in such a way as to justify their decisions to those affected. This sort of claim says nothing archimedean about whether law, properly understood, is the kind of thing that requires justification of its coercive acts.

A different way of putting the point is this: Dworkin may say that law requires us to justify coercion. We could call this a thin metaphysical commitment. A positivist might then reply along the following lines: ‘That’s not law that you’re talking about – law, properly understood, can be the kind of thing that involves unjustified force. So whatever it is you are singling out is not the true nature of law’. How would Dworkin respond? If he were committed to archimedean metaphysics, he would engage with that claim and say that, to the contrary, law really is the kind of thing that involves justified use of force. But I think his response would be that such a claim is unanswerable: the positivist is trying to describe the true nature of law in a way that requires stepping outside a domain to give an archimedean account of what it really is. Dworkin would, or should, given his anti-archimedean commitments, simply refuse to engage with the positivist on that metaphysical question and continue to make claims from within the domain of fact or value.

We can answer questions within the moral domain or within the factual domain, but we cannot answer archimedean questions about those domains – questions which pretend to step outside one of the realms and see them from some outside perspective. If we cannot answer archimedean questions, the domains of fact and value are important. They designate the areas within which all the answerable questions reside.

Multiple Concepts

Section III argued that Dworkin resists placing law in the domain of fact, and that he thinks it should be placed in the domain of value, though, as noted, he provides little in the way of argument for the latter claim. Section IV outlined Dworkin’s anti-archimedean and anti-metaphysical commitments. Where does all this leave us?

Dworkin’s anti-archimedean commitments mean we cannot step out of any domain and talk about what the law ‘really’ is. There are descriptive questions that we would think of as related to law, which are located in the realm of fact, but Dworkin is not interested in these questions. Thus, he is left with the moral domain. He claims that law must be placed in the domain of value. But there is no argument for this claim, because any argument that the law properly belongs in the domain of value would have to be archimedean.

In my view, the key is in seeing that there are many questions about law, and many concepts of law, that coexist. Dworkin distinguished between the doctrinal, sociological, taxonomic, and aspirational concepts of law.⁷¹ These concepts serve different purposes. When we attempt to answer any question relating to one of these concepts, we must locate ourselves within one of the domains. Is it a question that can be answered wholly within the factual domain? Or is it a question that requires normative analysis? Once we properly frame our questions, and locate ourselves within one of those domains, the rest of the analysis is possible.

But what is not possible, on my reading of Dworkin, is any discussion of the domain in which law ‘properly’ belongs, or of which of these various concepts represents law’s true nature. To answer such questions would require an archimedean perspective, which, as we have seen, Dworkin opposes. We can answer questions within a domain, but we cannot step outside of a domain to say which concept is the ‘true’ concept of law.

I think the best way to make sense of Dworkin’s claim that law is a branch of morality is that the questions about law that most interested him belonged in the normative domain. He wanted to know what we ought to do, given this system we find ourselves in. This is a thoroughly moral question. He certainly continued calling this question one of ‘what the law is’. But even the Dworkin of Law’s Empire was doing a wholly moral project, as evidenced by the starting point mentioned above.

This interpretation might seem odd or revisionist. But I believe it is a way to take seriously commitments that run much deeper in Dworkin’s thought. If we accept his starting point, according to which law justifies coercion, then we can talk about what ‘the law’ – understood in that way – actually demands of us. Adopting that point of view, we can ask what sort of moral obligations a judge will have. But that conversation is different from the one metaphysically-inclined legal theorists want to have. Dworkin’s questions begin and end in the domain of morality.

Further, there is textual support for my reading. Recall that Dworkin says that under the one-system view, ‘we now treat law as a part of political morality’.⁷² In my view, the word treat is important here. Treat does not mean the same thing as ‘conclude’. It means something more indicative of choice – a choice to talk about one concept of law and not another. I think, when he talks about what the law ‘is’, we should understand this as taking as given a particular account of what we mean by law, and moving from there to further analysis of what ought to be done.

When Dworkin talks about the doctrinal concept of law, he means to invoke a concept that justifies the use of force. The question we are asking is what act would be justified in light of past institutional practice. Others might not agree that the doctrinal concept of law really does engage this normative question. But if the objection takes a form such as ‘that concept does not reflect the true nature of law’, it is an archimedean objection that Dworkin will not entertain. A different objection might be that this concept doesn’t open up relevant and interesting questions. This objection is at least sensible, but does not undermine Dworkin’s project. It simply implies that the two theorists have different aims, and may need to explore different concepts or answer questions located in different domains.

If we accept the starting point as designated by the concept Dworkin intends to examine, then we find ourselves wholly located in the domain of value, and can ask further questions about that moral practice. But there can be no argument about which starting point is the correct one – which starting point properly reflects the contours of the metaphysical phenomenon, ‘law’.

Further textual support comes when Dworkin says: ‘We must hold constant certain parts of our attitudes and convictions about law, as not under present study, in order to evaluate and refine the rest’.⁷³ Sometimes we provide an account of how things are, given a particular set of assumptions. Recall that Dworkin understands the project of talking about law as a project of talking about real rights, about instances of actually justified coercion.⁷⁴ I think of Dworkin as ‘holding steady’ the idea of the doctrinal concept as an interpretive concept that spells out our genuine moral obligations. Once we have established the concept we are working with, we can ask further questions, such as: what theory of adjudication would best make sense of law as justifying coercion?

A final source of support for my reading of Dworkin comes in his discussion of the ‘evil law’ dispute. He says that the important question is the underlying moral one, and we can set aside the issue of what to call a bad statute. Once we settle the moral questions, we have a choice about how to describe it.⁷⁵ He then goes on to say that this problem ‘is sadly close to a verbal dispute’.⁷⁶ A verbal dispute is one in which it makes no difference what term we use, and we can simply disambiguate concepts, clarify that we are working with Law₁ or Law₂, and carry on. This seems to be what Dworkin is doing. In his discussion of rights that aren’t enforceable by courts, which some people wish to describe as legal rights, he says: ‘This is indeed an available way to describe the situation: no one would misunderstand. The different vocabulary I suggest seems at least equally natural, however’.⁷⁷ Again, this sounds like a verbal dispute that we can happily circumvent by establishing what concept we have in mind. In saying that legal rights are those enforceable on demand in courts, this is what Dworkin aimed to do: set up a way of using the term that positions us to do the normative work he considered important. He did not care to argue about whether that was the proper use of the term – whether it ‘gets right’ what law is.⁷⁸

Although there is an element of choice in selecting what concept one will examine, this need not be arbitrary or disconnected from our practices. A choice of a particular concept of law is guided by the question being asked. But one might object: how do we avoid arbitrariness in selecting our concept? Waluchow raises this issue, suggesting that Dworkin fails to provide sufficient guidance about how we should identify the relevant institutional history and practices.⁷⁹

Ripstein also notes the problem, but says that Dworkin’s anti-archimedeanism helps him avoid it:

From Dworkin’s perspective, however, the problem must be seen as nothing more than a holdover from the same Archimedean ideas that he resists. Practices need definitive preinterpretive contours only if someone wants to police a practice while remaining outside of it. Real debate, whether about art, law, or justice, takes place when people are already participating in the practice, and they do not need any guidance from outside of that to show them the way in.⁸⁰

I think this response is successful. We begin where we begin, with the contours of the practice that seem reasonable. If our interlocutors disagree, they can suggest different boundaries, which may enrich our debate, or they can accept our starting point, and engage with us about ‘that concept’, even if they do not fully agree that this is the most important concept to discuss, or that it maps onto ‘what law is’. In this way we can be attentive to the practices that the participants consider important, without fighting about the precise boundaries of those practices.

Dworkin says that ‘[t]oo much jurisprudence has traveled from some declaration about the essence or very concept of law to theories about rights and duties of people and officials. Our journey must be in the opposite direction: vocabulary should follow political argument, not the other way around’.⁸¹ I take the italicized point to mean that it is not so important what term we use: we can call this stuff law if we like. But political argument, normative argument, is what matters. The question that interests Dworkin is what genuine rights we have.

This element of terminological choice does not preclude the idea that there are right answers to moral questions. Indeed, the arguments in the first half of Justice for Hedgehogs make plain that Dworkin thinks there are right answers in the domain of value. It is not a matter of choosing or stipulating the answer to a moral question. It is about the step before that inquiry: the choice to step into the domain of value and ask questions that can be answered therein. Once we are in that domain we are subject to its norms, and according to Dworkin, its norms insist that we are seeking truth. Thus, this interpretation is compatible with Dworkin’s commitment to moral truth.

One might still be concerned that the idea of ‘choice’ between concepts doesn’t accurately account for the idea that the participants are constrained by the practice which they interpret. But again, those constraints operate on interpreters once they have set the terms of their discussion and entered the domain of value. The right answers, and the constraints on how we talk about a practice, enter once we are already agreed that we are discussing a normative ideal, or a question of what rights our political system actually gives to those who live under it.

We can now return to the mysterious argument in Justice for Hedgehogs about placing law in the moral domain. I have been arguing that Dworkin’s anti-archimedeanism and his Humean distinction between fact and value can help us understand the argument. On my reading, the argument runs as follows: We cannot access a neutral, archimedean position on what law is or how it relates to morality. Putting aside ‘legal’ questions, since that is precisely the category that puzzles us, we can either answer moral questions or factual ones.

The thought is that there is no ‘legal’ domain, and talk about legal questions must be masking a different question.⁸² Sometimes it masks a predictive question. Sometimes it masks a moral one. But – and I think this is Dworkin’s point – there is no question from outside both these domains that tells us about the connection between law and morality. There is no way to answer the archimedean question of whether law ‘really’ belongs in the moral domain or not. This is why Dworkin can simply ‘place’ law in the moral domain without argument. He begins there in order to pursue important normative questions.⁸³

Eliminativism

I have been arguing that Dworkin’s view is moral all the way down: he is interested in the moral rights that people actually have, and how those are affected by our institutions and practices. While he uses the term law, it seems better to understand this as referring to one of many concepts, rather than as a statement about law’s singular nature. He is working with a normative concept of law, according to which law is something that ought to be obeyed, and that genuinely justifies coercive force. There is no metaphysical argument for that starting point – and there shouldn’t be. Dworkin is right to simply start by clarifying the terms of debate, and then engage in moral inquiry.

Does this mean Dworkin was an eliminativist? Eliminativism is a term for a cluster of positions recently gaining ground in jurisprudence, according to which some concept is nonexistent or some question is ill-formed, and we can discard it, and replace it with a series of other questions or concepts.⁸⁴ There is some ambiguity about what this term refers to, but here is not the place to engage in a full examination of the variety of views that might be called ‘eliminativist’. I will discuss two versions of eliminativism that might fit Dworkin’s views as I have interpreted them here.

First, the idea that there are multiple concepts might lead to eliminativism. The thought is that multiple concepts coexist and play different roles in our theorizing. But we need not determine that one or the other is the ‘true’ concept of law. On this version of eliminativism, what is ‘eliminated’ is the further question of which concept tracks the nature of law. There can be no argument about which concept ‘gets the nature of law right’, because they don’t compete on the same territory; they are simply doing different things.

Another version of eliminativism involves eliminating a particular concept. This version is defended by Lewis Kornhauser. Kornhauser identifies a number of different claims of varying strength about what it means to do without the doctrinal concept of law, but the unifying idea is that the doctrinal concept of law is unnecessary: judges, other officials, and citizens can all get by without the doctrinal concept of law.⁸⁵ The traditional view has supposed that there is a two-step process, wherein we figure out what the law is, and then decide whether we ought to follow it. On Kornhauser’s view, by contrast, we simply implement a one-step process of decision-making, where all reasons are weighed in the initial inquiry.⁸⁶

On this view, Dworkin might sound like an eliminativist. That is, he endorses a one-step process where the judge does not first figure out what the law is and then decide what to do, but simply seeks an answer about what she should do, morally speaking.⁸⁷ Kornhauser says that his position diverges from Dworkin’s. For one thing, Kornhauser’s view is explicit about eliminating the doctrinal concept of law in a way that Dworkin’s is not.⁸⁸ But in light of the fact that Dworkin’s view boils down to the position that we should simply focus on the moral questions – the questions of what morality requires of us in this specific historical setting – I think it is reasonable to interpret Dworkin’s view as a form of eliminativism.

But, of course, Dworkin uses the term ‘law’ constantly, and many find the idea that he would give up on the doctrinal concept of law implausible. Indeed, the thought of eliminating the doctrinal concept of law may seem implausible independent of our interpretation of Dworkin. Liam Murphy, for example, argues against eliminativism on the basis that we need to know what the law is, practically speaking, and so we cannot do as the eliminativist suggests, and simply replace law talk with other questions or ways of speaking.⁸⁹ In other words, ‘we need a view about the grounds of law if we are to be able to figure out the content of the law in force’.⁹⁰ Murphy argues that it is not a satisfactory response here to say, for example, ‘that on one understanding of what law is, same-sex couples have the legal right to marry while on another they do not’.⁹¹

But is that really so problematic? We might put it better as follows: there is one answer to the question of whether an official is likely to uphold a marriage between two people of the same sex. That is a factual question to be answered within the domain of fact. And there is a further question of what a judge, faced with a case that raises this question, ought to do, all things considered, given the institutional history we have and the moral stakes at hand. It seems reasonable to suppose that these questions coexist. Furthermore, we might actually do better when we disambiguate these questions, so that we don’t end up talking past each other when one person says that the law demands recognition of same-sex marriage, meaning the moralized concept of law, and another says that it does not, meaning to invoke the practical question of what will likely happen in a certain jurisdiction.

It is true that Dworkin spoke of the moral question as a question of ‘what the law is’. And he does not ‘deny the distinctness of questions about what the law is and what it ought to be’.⁹² Dworkin uses the example of a family’s morality to further spell this out. Family morality depends on how parental authority has been used in the past, and the importance of cohering with that for reasons of fairness. This is a separate question from what one ought to do in the absence of any family history. Similarly, his point is that the questions of ‘what law is’ and ‘what it ought to be’ are separate. But, crucially, they are both moral questions.⁹³ The family history did not produce a code of behavior that is separate from morality. The authority the family morality has just is the authority of morality itself. Thinking of the family morality as something distinct from morality is misguided, because if history and prior exercises of authority matter, this is itself for moral reasons. It is ‘principles of fairness … that make your family’s distinct history morally pertinent’.⁹⁴

Once we see that the whole inquiry is a normative one, including what Dworkin calls the ‘what is law’ question, we can see why this turns into eliminativism. There can be no debate about whether moral principles are part of the grounds of law: that debate has been rendered irrelevant by our very starting point, according to which we are using law to mean something like ‘the actual reasons we have for coercing people based on past actions’. It sounds as though Dworkin has begged the question by stepping inside the one-system view, but that is only question-begging if we think there is an answer to the question of what really counts as law. But there isn’t; there are multiple different concepts raising different questions, and there is no further question of which of these is the right one. This clearly amounts to at least the first form of eliminativism I mentioned above, where what is eliminated is the debate about which concept tracks the nature of law.

Dworkin recognizes that the one-system picture seems to suggest that there is no real distinction between theories of law and theories of adjudication.⁹⁵ The moral arguments that a judge makes while crafting her decision ought to be our focus. We should understand judges as searching for the right answer, as Dworkin has always said they do. But that right answer is a right answer in morality, not in law. In my interpretation, they are trying to figure out what is the right thing to do, given past historical events. We should not see them as trying to access some independent legal truth. Once we understand that the ‘right answer’ judges seek is the moral answer they are compelled to find given the rights that are at stake, it is hard to understand the point of stating that, in doing so, they really ‘apply the law’.⁹⁶

The Justice for Hedgehogs position seems to be that the only question that really matters is the question of what we should do, morally speaking. While our obligations arise in a particular context, as Dworkin makes clear with his example of the family, the ultimate question is still what ought to be done. But what ought to be done will be affected by context and factual history. The only genuine obligations are moral obligations, and both judges and citizens should be concerned to answer moral questions. Questions about the grounds of law resolve into moral questions about what ought to be done, and we could just as well say that the doctrinal concept of law is eliminated in favor of these moral questions.⁹⁷ This appears to amount to the second, stronger form of eliminativism I mentioned above. Thus, without settling the question of which version of eliminativism is the right one, we can see that Dworkin’s view seems compatible with both.

Of course, what judges do isn’t the only important question. Eliminativism seems harder to make sense of for citizens. What should people who just want to follow the law ask themselves? The eliminativist’s point is that the ‘what is law’ question doesn’t help the citizen. It has to be reframed. In Justice for Hedgehogs, Dworkin put the central question as follows: ‘Under what conditions do people acquire genuine rights and duties that are enforceable on demand in the way described?’⁹⁸ While this may be primarily a question for judges, it is also a question citizens can ask in trying to figure out how to order their affairs. They can ask what rights their government’s actions actually give them. This wholly normative question remains answerable on his framework. But we can ask it and be eliminativists: it is not about ‘what the law is’. It assumes what the law is, and goes on to answer moral questions stemming from that assumption.

Citizens can ask other questions, too, including ones that belong in the realm of fact. They might want to ask about the likely consequences of their actions. Eliminativism enables us to express this, because we aren’t looking for a singular answer about what ‘the law’ is, but rather seeking to answer a range of different questions, some of which belong in the normative domain and others in the factual one. The important point is to be clear from the start about which question we are asking and what domain it belongs in.

Conclusion

I have been arguing for an interpretation of Dworkin’s views according to which the Justice for Hedgehogs position is not new. In 1972 he said ‘My point was not that ‘the law’ contains a fixed number of standards, some of which are rules and other principles. Indeed, I want to oppose the idea that ‘the law’ is a fixed set of standards of any sort.'⁹⁹ Dworkin said this in the course of engaging with Raz’s view, which Dworkin says is overly concerned with individuating what counts as a law.¹⁰⁰ He goes on to say that his view does not commit him ‘to a legal ontology that assumes any particular theory of individuation’.¹⁰¹ In other words, the boundary-drawing project of identifying what counts as a law is not Dworkin’s project.

In his Reply to Critics in that same collection, he says that Hart and other theorists hold a misguided view according to which there is something called ‘existing law’.¹⁰²

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.¹⁰³

If questions about legal rights are questions about political rights, this already envisions law as a branch of morality, making the two-systems view seem much less like a new direction. So Dworkin has long insisted that arguments about law have to be normative. And that argument has long rested on the idea that a descriptive account of law simply makes no sense. He makes this argument in his Reply in a volume of essays on his work from 1983.¹⁰⁴ In that same Reply, Dworkin says Raz accuses him of endorsing the view ‘that there is such a thing as “the law,” meaning a discrete set of principles and other standards that alone is relevant in determining the decisions people are entitled to have from courts.'¹⁰⁵ But he says he does not endorse that idea. He says that the picture he paints is one of the judge applying moral principles to interpret our institutional history.¹⁰⁶ This picture, I think, is the same as the Justice for Hedgehogs view: the whole process is moral. There is no ‘figuring out what the law is’, only morally engaged deliberation on the rights of a litigant. Further, this sounds like a version of eliminativism.

I traced the thread of this eliminativist, one-system view through his earlier work, arguing that his presupposition that law is a matter of justified coercion indicates he always assumed law was part of the moral realm, even if he didn’t use the ‘one-system’ language. My interpretation of Dworkin here might seem radical, but it is grounded in his own thought. I have taken seriously the deep and powerful commitments he articulated across his vast body of work – his anti-archimedeanism, and his belief in moral truth.

Dworkin’s argument for the independence of value is the deep underpinning of his views about law. We can’t derive claims about what something ought to be like from claims about what it is. And similarly, many people have thought Dworkin’s idea that we should figure out what the law is by talking about what it ought to be, or what would justify it, was baffling.¹⁰⁷ That is why we have to understand Dworkin’s whole project as taking place in the normative domain. We can talk about what the law ‘is’ by talking about what it ought to be if and only if we realize the ‘is’ is actually an ‘ought’. The ‘What is law?’ question is really asking about what is actually morally demanded of us in a given context. That can only be answered from within the normative domain.