Notre Dame Law Review Volume 67 | Issue 3 Article 2 6-1-1999 Reflections on Dworkin and the Two Faces of Law Richard H, Fallon Jr. Follow this and additional works at: http://scholarship.law.nd.edu/ndlr Part of the Law Commons Recommended Citation Richard H. Fallon Jr., Reflections on Dworkin and the Two Faces of Law, 67 Notre Dame L. Rev. 553 (1993). Available at: http://scholarship.law.nd.edu/ndlr/vol67/iss3/2 This Article is brought to you for free and open access by NDLScholarship. It has been accepted for inclusion in Notre Dame Law Review by an authorized administrator of NDLScholarship. For more information, please contact lawdr@nd.edu. Reflections on Dworkin and the Two Faces of Law Richard H. Fallon, Jr.* I. INTRODUCTION It has been twenty-five years now since Ronald Dworkin began his efforts to redraw the map of jurisprudential debate' by offering a "third theory of law."' When Dworkin published his first important articles, a traditional and polar rivalry dominated the field. Positivists equated law with the rules recognized as authoritative within a particular legal system.' This association of law with rules led positivists, besides treating the existence of law as a matter of fact, to embrace a gap thesis. When the existing rules failed to resolve a controverted case,4 positivists asserted that judges had no choice but to fill gaps by making law much as a legislature would.5 Rejecting the positivists' identification of law with factually existing rules, 6 natural law theorists countered that the concept of law necessarily includes a moral element: a rule, however aggressively it might be enforced, could not truly be law unless it sat- * Professor of Law, Harvard Law School. I am grateful to a number of friends and colleagues who offered helpful comments on and criticisms of prior drafts of this Essay. Particular thanks go to Charles Fried, Frank Michelnan, Arval Morris, Lew Sargentich, Fred Schauer, Cass Sunstein, Lloyd Weinreb, Don Welch, and participants at workshops at Harvard, Vanderbilt, and the University of Washington Law Schools. Work was supported by a research grant from the Harvard Law School. 1 Dworkin's most important early piece was The Model of Rules, 35 U. CHI. L. REV. 14 (1967). His distinctive theory most clearly emerged in Hard Cases, 88 HARV. L. REV. 1057 (1975), and has evolved in works including three books: TAING RIGHTS SERIOUSLY (1977); A MATrER OF PRINCIPLE (1985); and LAW'S EMPIRE (1986). 2 See John Mackie, The Third Theory of Law, 7 PHIL. & PUB. AFF. 3 (1977). 3 See, e g., H.L.A. HART, THE CONCEPT OF LAw 77-96 (1961). 4 See id. at 12, 121-32. 5 Se id. at 12, 132, 141. 6 This formulation does not do full justice to sophisticated positivists, such as H.L.A. Hart, who recognized that law must be capable of supporting an "internal point of view" that differentiates the law's claim to obedience from that of a gunman, see id. at 55-56, 79-88, and that as a matter of "natural necessity . . . minimum forms of protection for persons, property, and promises . . . are . . . indispensable features of municipal law." Id. at 195. For Hart, the core of the positivism that he wishes to defend against natural law doctrines is that "it in no sense a necessary truth that laws [in order to is count as laws must] reproduce or satisfy certain demands of morality . . ." Id. at 181. NOTRE DAME LAW REVIEW [Vol. 67:553 isfied a moral test of acceptability. 7 This brand of natural law theory stood in an uncertain relationship with another version, possibly reflected in Blackstone, which held that law was not what legislatures said or judges thought, but law as it ought to be.8 On this view, the judicial task had no gap-filling or law-making aspect,9 but was solely one of discovering the law as it really was. Through its various iterations, Dworkin's third theory has attempted to bridge the gap between the two traditional theories. With the positivists, Dworkin has accepted that the concept of law makes sense only in reference to going legal systems; to know what the law is, it is necessary to begin with the materials that are recognized as law in a particular culture." Dworkin leaves room to accommodate the natural law view," however, by insisting that the materials that are recognized as authoritative within any legal system-the rules and standards that positivists have traditionally regarded as exhaustive of law-must always be interpreted. For interpretation, according to Dworkin, has an irreducibly moral element; the relevant materials must be interpreted in their best moral light." Dworkin thus sides with natural law theorists in recognizing a conceptual link between law and morals. Building on this foundation, he has further asserted that legal interpretation necessarily aspires to provide a moral justification for the law's claim to obedience. 4 He implies that a regime that was incapable of generating at least a presumptive, general duty to obey the law would not count as a properly "legal" system at all, but only as a scheme of organized coercion. 5 The measure of Dworkin's influence can be found on all sides of contemporary jurisprudential debate. The notion that law is an interpretive practice, in which legal materials must be given meaning by purposive agents, has caught on with a vengeance. 6 The 7 See, e.g., MARTIN P. GOLDING, PHILOSOPHY OF LAw 30-33, 46-51 (1975) (describing natural law theories). 8 1 WiLLIAM BLACKSTONE, COMMENA.RIES. 9 See id. at *69 (stating that the judicial duty is not to "pronounce a new law, but to maintain and expound the old one"). 10 See LAW'S EMPIRE, supra note 1, at 65-68, 87-88. 11 12 See Ronald A. Dworkin, "Natural Law" Revisited, 34 FLA. L. REV. 165 (1982). See LAW's EMPIRE, supra note 1, at 52-53, 62-68, 254-58. 13 See id at 52-53, 90. 14 See id. at 139, 191. 15 See id. at 101-13, 190-91. He is, nevertheless, sensitive to the relevance of context and the flexibility of language in making sense of others' seemingly contrary assertions about when law and legal systems exist. See id. at 87-113. 16 For a critical reaction, see Michael S. Moore, The Turn to Interpretation:A Turn for 19921 DWORKIN AND THE TWO FACES OF LAW idea that legal decisions must be morally justifiable in order to generate an obligation to obey has also emerged as a prominent 17 theme in constitutional as well as jurisprudential scholarship. 8 Although strongly influenced by Dworkin myself," I argue in this Essay that his effort to develop a third theory of law is, ultimately, less an unequivocal success than a deeply instructive failure. Law's Empire, Dworkin's most recent major work and the text with which- I shall be most concerned, is exemplary. Although enormously illuminating in the depth and sweep of its vision, Law's Empire fails in its attempt to synthesize the insights of positivism and natural law theories. Too much is lost in the amalgamation. In this Essay, I argue that the traditional opposition of positivism and natural law better exhibits the irreducibly Janus-faced character of the concept of law than does Dworkin's hybrid. I also attempt to show that a type of positivist theory, which Dworkin has labeled "soft conventionalism," is superior to his own as a theory of law and that such a theory can incorporate most of Dworkin's insights into the nature of adjudication within the distinctive context of the American legal system. : II. WHAT IS LAW? Although Dworkin clearly means to offer a theory of law, his slant on the omnibus question "what is law?" differs from that of other legal philosophers. Since Dworkin has developed his theory largely in opposition to the positivist theory of H.L.A. Hart, 19 it may be useful to begin with a sketch of Hart's inquiry. A. Hart's Inquiry H.L.A. Hart's The Concept of Law attempts to answer the question "what is law?" largely by propounding a theory of the nature of a "legal system" 2 -- the paradigmatic context in which law, as thm Worse?, 41 STAN. L. Rnv. 871 (1989). 17 See Paul W. Kahn, Community in Contemporary Constitutional Theory, 99 YALE L.J. 1 (1989). 18 19 See, e.g., Richard H. Fallon, Jr., A Constructivist Coherence Theory of Constitutional Interpretation, 100 HARV. L. REV. 1189, 1233-37 (1987). Dworkin, in The Model of Rules, supra note 1, explicitly attacked Hart's theory, as elaborated in THE CONCEPT OF LAW, supra note 3, and he has kept up' the attack in his subsequent writings. See, e.g., LAW'S EMPIRE, supra note 1, at 33-44. 20 See HART, supra note 3, at 16-17, 111-14. NOTRE DAME LAW REVIEW [Vol. 67:553 an instrument of social control, is exhibited." According to Hart's classic formulation, law (or more properly, I think, a legal 2 system)22 is a union of primary and secondary "rules." 3 Within this typology, primary rules regulate the conduct of the population as a whole; secondary rules, including a so-called "rule of recogni24 tion" that identifies what counts as law within the jurisdiction, define the powers of law-making and law-enforcing officials.25 Hart constructs his analysis around the concept of a social rule, which must be understood from both an internal and an external point of view. 6 From an external point of view-roughly, that of an observing anthropologist-a social rule is a standard of conduct that is generally complied with and deviation from which typically engenders criticism. From an internal point of view-that of an unalienated participant in the social life of the community-a social rule is a standard that is accepted as a guide to conduct and a basis for criticism, including self-criticism. Reliance on the concept of a rule allows Hart to pursue several tasks simultaneously. First, the concept explains the operation of law in ordinary social life: basic legal standards tend to be wellknown, and departures from them are typically regarded as occasions for criticism or self-criticism, if not official coercion. 29 Second, Hart hopes to account for the phenomenon of legal change: "power-conferring" rules authorize the creation of new rules by 3 designated officials. Third, he aims to explain the nature of adCf JOSEPH RAZ, THE CONCEPT OF A LEGAL SYSTEM 169-70 (2d ed. 1980) ("The of the concept of a law depends on the analysis of the concept of a legal sysSee HART, supra note 3, at 111-14. See id at 91-96. See id. at 92. 21 analysis tem."). 22 23 24 25 See id. at 91-96. Somewhat confusingly, Hart also uses the term "secondary rule" to encompass all power-conferring rules, a category within which lie also locates rules specifying the acceptable forms of private creation of legal obligations through, for exampie, the drafting of wills and the making of contracts. See id. at 78-79. 26 See id.at 54-60. 27 See id. at 54-60, 86-88. 28 29 See id. See id.at 39. The principal functions of the law as a means of social control are not to be seen in private litigation or prosecutions, which represent vital but still ancillary provisions for the failures of the system. It is to be seen in the diverse ways in which law is used to control, to guide, and to plan life out of court. Id. 30 See id.at 35-43, 93-94. 19921 DWORKIN AND THE TWO FACES OF LAW judication: judges typically apply the law, but are authorized by power-conferring rules to fill gaps and possibly to effect other changes."' Fourth, and most generally, by beginning with the concept of a social rule, Hart emphasizes that law is a phenomenon grounded in shared standards of appropriate conduct and 32 socially enforced pressure to conform. In contrast with Dworkin, however, Hart's rule-based analysis reflects much less concern with the procedure for identifying "the law" in any particular disputed case than with explicating the con3 cept of "law" generally. 3 Insofar as the "Hart-Dworkin debate" is worth judging in its own terms,' a central question is whether Hart's angle of vision-which links the concept of law closely to that of a legal system and accounts for the nature of a legal system in terms of rule-based foundations in social fact-is an illuminating one. B. Dworkin's Focus In contrast with Hart's focus on the nature of a legal system and the ways in which a legal system characteristically guides conduct and furnishes mechanisms for lawmaking and legal change, Dworkin develops his theory of law by asking how judges actually do, or should, decide "hard cases." When faced with a case having no conventionally accepted resolution, 6 judges, according to 31 S&e id. at 121-50. 32 Hart allows at least the possibility that, in an advanced legal system, the mass of the general population need not accept primary rules as standards for criticism and selfcriticism. It is enough if they generally obey the law, whatever their attitude toward it. See id. 59-60, 113. The system's officials, by contrast, must generally "accept" the rule of at recognition-the fundamental rule specifying how other legal rules are to be identified-in the sense of taking it as a "common standard[] of official behavior and apprais[ing] critically their own and each other's deviations as lapses." Id. at 113. As a result, law is always rooted in the facts of social practice, even if only that of the officials who (for there to be a legal system at all) must accept the rule of recognition, both the existence and the content of which are irreducible matters of fact. 33 As I have suggested, Hart attempts to achieve explication less by addressing forms of the question "what is the law?" than by elucidating the context in which the question is likely to arise: that of a legal system. Indeed, Hart's account of law as a union of primary and secondary rules (see HART, supra note 3, at 91, 96) works much better as a sketch of the nature of a legal system than it does as a definition of "law"-a concept that permits a host of usages. 34 For critical comparisons and contrasts, see E. Philip Soper, Legal Theoiy and the Obigation of a Judge: Te Hart-Dwordn Dispute, 75 MICH. L. REv. 473 (1977); Lloyd L. Weinreb, Law as Order, 91 HARV. L. REV. 909 (1978). 35 &e, eg, LAw's EMPIRE, supra note 1, at 10, 128-29, 255-56, 265-66, 411; Hard Cas- es, supra note 1. 36 The absence of a conventionally accepted or acceptable resolution may be attrib- NOTRE DAME LAW REVIEW [Vol. 67:553 Dworkin, do not simply proceed on the model of a deputy legislature Instead, Dworkin argues, judges seek to "interpret" what they take to be the relevant legal authorities and to discern what result the relevant authorities, when properly interpreted, dic8 tate. 3 As portrayed by Dworkin, the required interpretation turns out to be an endless, "creative," 9 and thoroughly "protestant' activity. Before engaging in legal interpretation, anyone must first develop a complex theory of "the point" of law and legal practice.4 1 A would-be legal practitioner must determine why law exists, why people engage in legal interpretation at all, and what interpretive standards must be observed if the point of law and legal practice is to be realized. Dworkin suggests that a practitioner must begin with a "preinterpretive sense" of what is generally done and accepted by those who engage in legal argument and decisionmaking.42 Then, treating generally accepted norms as provisional fixed points that any competent theory must encompass, the practitioner must take the "interpretive" step of "try[ing] to impose meaning on the [practice]-to see it in its best light-and then to restructure it in light of that meaning. ' Finally, having done this, the practitioner must interpret particular authorities in the way that the aims of the practice require-one that makes both the general legal practice of the community and its resolutions of particular controversies the best that they can plausibly be portrayed as being. 44 Different people will reach different interpretive judgments, partly because they will differ about how best to strike the balance between "fit" and normative attractiveness45 and 'partly because they will disagree about what is mor4 6 ally and politically right. utable either to the vagueness or ambiguity of relevant authorities or to the morally or socially problematic consequences of applying an otherwise applicable rule or standard. See FREDERICK SCHAUER, PLAYING BY THE RULES: A PHILOSOPHICAL EXAMINATION OF RULEBASED DECISION-MAKING IN LAW AND IN LIFE 209-10 (1991). 37 See, eg., LAw's EMPIRE, supra note 1, at 254-56. 38 See Hard Cases, supra note 1. 39 See LAw's EMPIRE, supra note 1, at 50-53. 40 Id. at 190, 413. 41 See id.at 52-53, 58-59. 42 See id- at 65-67. 43 I& at 47. 44 See id- at 52-53, 98-99, 255-56. 45 See i& at 254-58. 46 See id.at 254-60. Dispute notwithstanding, Dworkin maintains that there is "one right answer" to legal questions determined largely by standards of ultimate moral right. 1992] DWORKIN AND THE TWO FACES OF LAW 559 The conceptual connection between law and morals also .dictates what Dworkin calls "interpretive protestantism. ' , 7 The question (of what is morally right cannot be answered by appealing to social fact or settled legal authority. Each judge and legal interpreter must decide for herself, paying attention to the views of others only insofar as they can be regarded as reliable guides to moral and political truth." C. The Costs of Dworkin's Approach As a theory of adjudication in the English and American contexts, Dworkin's theory is, at least, challenging and provocative. But his approach imposes costs. By adopting a judge-centered perspective focused on the problem of how to decide hard cases, Dworkin in Law's Empire pays relatively little heed to Hart's questions of what the constitutive features of a legal system typically are, how those features interact with and depend on each other, how law functions in social life outside the courts, and what pressures judges feel to conform to established standards. Dworkin does not wholly ignore Hart's questions. He says repeatedly that law is an "interpretive practice" 9 in which an interpretation must achieve sufficient fit with the practice of the legal community to count as an interpretation rather than a revision.50 He also elaborates what he takes to be a necessary core of shared assumptions for an interpretive community to exist. 51 But while Dworkin does Id. at 266. According to Dworkin, moral and legal skepticism are either empty, see id. at 76-86, or misguided, see id. at 267-75, ideas, and the conceptual connection of law and morality makes law objective and determinate in the same way and to the same extent that morality is. Morality, in other words, fills any legal gaps and resolves any ambiguities that might otherwise exist. 47 Id. at 413; see also id. at 190 (characterizing "political obligation" as a related "protestant idea" calling for "fidelity to a scheme of principle each citizen has a responsibility to identify, ultimately for himself, as his community's scheme"). 48 See id at 64 ("[E]ach of the participants in a social practice must distinguish between trying to decide what other members of his community think the practice requires and trying to decide, for himself, what it really requires."). 49 See, eg., id. at 50, 87. For an interesting analysis of the concept of the practice, and a set of useful distinctions among different kinds of practices, see Thomas Morawetz, The Concept of a Practice, 24 PHIL STUD. 209 (1973). 50 See, eg., LAW'S EMPIRE, supra note 1, at 254-58. 51 According to Professor Postema's sensitive reconstruction, "four conditions define what, in Dworkin's view, is necessary and minimally sufficient for intelligible interpretive activity within a practice": (1) Generic or background -consensus: participants must share a language and understand the world about them in much the same way, have roughly similar NOTRE DAME LAW REVIEW [Vol. 67:553 not attempt to cut his theory free from such anchors, Law's Empire places little emphasis on system-maintaining social forces, including social pressure to conform. On the contrary, Dworkin leans distinctly in the opposite direction 2 with his recurrent suggestion that interpretation is an irreducibly protestant enterprise, in which each judge must develop her own theory of how to characterize the relevant legal materials in the best moral light. 3 On my reading, Dworkin's conception of interpretive protestantism implies that judges characteristically are not and should not be concerned with the rival interpretations that others will predictably develop. 54 This position seems untenable. At its heart, law has a function of guiding and coordinating human activity," and successful coordination requires attention to the likely views and responses of others.5 6 Within the American legal system, for example, an effective lawyer must make predictive judgments that attend, not just to the formal materials of the law, but to psychological and political interests and concerns, and, in general, participate in the same sufficiently concrete "form of life" . . . . (2) Boundary consensus: they must also agree on the "extension" or domain of the practice, i.e., on what behaviors, actions, decisions, claims (or rather claimings) count as falling within the practice, and which fall outside it. (3) Paradigm consensus: they must also recognize certain claims or propositions regarding what the practice requires to be true within the practice, if any are. Finally, (4) fit consensus: while participants may disagree widely regarding substantive "background" values (e.g., ideals of political morality) there must not be "too great a disparity" among participants regarding the amount of fit necessary to provide a viable distinction between interpretation and invention. Gerald J. Postema, "Protestant"Interpretation and Social Practices, 6 LAw & PHIL. 282, 298-99 (1987). 52 See LAW's EMPIRE, supra note 1, at 88 (discussing the mistake of emphasizing factors promoting consensus over those leading to disagreement). 53 See id. at 190, 252, 413. 54 See id. at 64-65 ("[E]ach of the participants in a social practice must distinguish between trying to decide what other members of his community think the practice requires and trying to decide, for himself, what it really requires."). Dworkin compares a judge's task to that of a chain novelist who must continue, in the best way that she can, a story that includes prior chapters written by other authors. Id. at 229. But he does not suggest that the chain novelist must seek to anticipate what future authors in the chain are likely to view as successful contributions, even when contemplating a "brilliant" but pathbreaking interpretation that characterizes past events in a light that past authors had never characterized them in. Id. at 247-48; see also id. at 58 (asserting that each participant in a practice reaches his own interpretation by "trying to discover his own intention in maintaining and participating in that practice" and characterizing social interpretation as "a conversation with oneself"). 55 See, e.g., Gerald J. Postema, Coordination and Convention at the Foundations of Law, 11 J. LEGAL STUD. 165 (1982). 56 See id. 1992] DWORKIN AND THE TWO FACES OF LAW factors that include changes in the social and economic culture and in the composition of the bench. Similarly, lower court judges must not simply ask what theory would portray past decisions in the best light; they must also pay heed to what result appellate judges are likely to reach. I do not wish to push this point too far. It is not my view that conscientious judges are always bound to decide a case as they think a higher court would. Within our legal system, for example, the relationship between higher and lower courts has a dialogic element; the creative handiwork of lower courts can help affect how a higher court will view a particular issue, if and when the higher court explicitly undertakes to decide it. Nevertheless, lower courts clearly are not created equal to higher courts, and the lower courts' obligation of conceptual faithfulness to precedent cannot be understood independently of higher courts' own understanding, in the psychological sense, of what their decisions mean. Even at the level of a court such as the American Supreme Court, a Justice has an obligation to look forward and outward to other judges and lawyers, as well as backward at the authorities requiring interpretation, and to reason in ways that the profession will regard as sensible and lawful.58 A judge or Justice who rendered a decision that she knew would be regarded as legally insupportable by the overwhelming preponderance of the legal community--even if she believed that her interpretation showed settled authorities in the best moral light-would most often, if not always, be outside the bounds of law.59 Conceivably, Dworkin would not want so much made of his claims for interpretive protestantism. ° He undoubtedly aspires to 57 The failure to make these predictions would foster inconsistency and uncertainty, multiply the number of appeals, and unreasonably burden litigants by increasing litigation costs. 58 Dworkin suggests that a judge ought to take public morality into account, see LAW'S EMPIRE, supra note 1, at 249, but makes no parallel claim about normative commitments of members of the legal profession. Cf id. at 64 ("[E]ach of the participants in a social practice must distinguish between trying to decide what other members of his community think the practice requires and trying to decide, for himself, what it really requires."). 59 See Postema, supra note 55, at 192-93. The qualification may be necessary to deal with implicit conventions about the permissible role of a judge in attempting to promote what is effectively a change of law. The implicit rules of judging do not invariably bar the planting of a seed that a judge may reasonably hope will change the legal landscape in the fullness of time. 60 Read narrowly, Dworkin's protestantism claim might mean only that a judge or Justice must ultimately decide for herself how much weight she ought to give to the NOTRE DAME LAW REVIEW [Vol. 67:553 maintain a delicate balance: to account for the centrifugal forces that produce as much consensus and cohesion as characteristically exist in law while at the same time explicating the phenomena of uncertainty, argument, and creative insight.61 In Law's Empire, however, his line of argument is so carefully crafted to explain and justify the latter phenomena that the former get somewhat short shrift. The emphasis is all on intellectual thrust and parry, not on the felt experience of an external and peremptory source of authority or of a need to achieve coordination with the decisional patterns of others. Whatever interpretive protestantism might cash out to mean, Dworkin's argument has tipped in a direction decisively different from Hart's. D. Rules, Facticity, and the Social Construction of Reality Dworkin's reasons for declining to follow Hart are many and varied, but his central concern involves the difficulties of founding an analysis of law, as Hart attempted to do, on the concept of a social rule. As abundant and helpful criticism has established, the concept of a "rule," as that term is perhaps most familiarly understood, is too crude to do the work Hart assigns it. The term "rule," which later authors have contrasted with standards or principles, 62 may connote a norm that admits concise, canonical formulation and that necessarily applies in an all-or-nothing way.0 Not all norms that are relevant to legal decisionmaking are of this anticipated views of other judges or Justices, there being no canonical formulation to resolve this question for her. Cf LAw's EMPIRE, supra note 1, at 313-14 (Hercules, Dworkin's prototype of the ideal judge, "must rely on his own judgment in answering" questions of political morality "not because he thinks his opinions are automatically right, but because no one can properly answer any question except by relying at the deepest level on what he himself believes."). But this reading of protestantism is too weak to be interesting. Cf Postema, supra note 51, at 288 ("Of course, ultimately one can only come to one's own view of the practice. In this trivial sense, even to defer to the majority is to come to 'one's own' view of the practice, .viz., that its meaning is determined by the majority."). 61 See LAW's EMPIRE, supra note 1, at 88-93. 62 See, e.g., TAKING RIGHTS SERIOUSLY, supra note 1, at 22-28; Duncan Kennedy, Form and Substance in Private Law Adjudication, 89 HARV. L. REV. 1685 (1976). For criticism of Dworkin's distinction between rules and principles and a suggestion that rules can be "overridden" as much as principles can, see SCHAUER, supra note 36 at 12-15. 63 It is unclear whether Hart wished the term "rule" to be limited in such a way as to distinguish it from vaguer or more flexible principles, conventions, and standards. See Margaret J. Radin, Reconsidering the Rule of Law,, 69 B.U. L. REV. 781, 794 & n.39 (1989). He was at pains, for example, to assert that the rule of recognition could be vague or indeterminate. Other passages, however, seem to trade implicitiy on the paradigmatic use of the word "rule" to identify standards capable of quasi-syllogistic applicability. 1992] DWORKIN AND THE TWO FACES OF LAW kind. Common law "rules" lend themselves to alternative formulations, so that competent lawyers can disagree about what the rule is, as well as about whether it applies to a particular set of facts. Norms of constitutional and statutory interpretation are also likely to be vague or conflicted, or to prescribe kinds of balancing processes that predictably will result in disagreement among lawyers and judges. On the other hand, abundant commentary has also demonstrated that the basic point of Hart's rule-based analysis is not to show the uniform nature of all legal "rules," but to illuminate how law is grounded in social fact and to clarify why there is no strong, conceptual connection between the relevant social facts and standards of ultimate moral right." Hart's basic emphasis can easily be maintained by a theory that accounts for law's facticity through reference to a variety of types of precepts6 5 -- rules, principles, conventions,' and so forth-departure 64 See ag., Macde, supra note 2, at 6. 65 See, e.g., NEIL MAcCORMCtK, H.L.A. HART 40-43, 123 (1981); Joseph Raz, Legal Prindples and the Limits of Law, 81 YALE LJ. 823, 845 (1972); Jules L. Coleman, Book Review, 66 CAL. L. REV. 885, 897 (1978). 66 See Harry T. Edwards, The Rol of a Judge in Modern Society: Some Reflections on Current Practice in Federal Appellate Adjudication, 32 CLEV. ST. L. REV. 385, 398-99 (1983-84). The first set of conventions that I have in mind pertains to permissible forms of reasoning, sensed and observed by almost all members of the American legal profession. Precisely because we take these habits of mind so for granted, it is difficult to identify the ways and occasions in which they constrain us. The context in which their role and importance is perhaps most evident is analogical reasoning; lawyers gradually acquire a feel (from law school, from everyday discussion with their colleagues, from the responses to briefs and oral arguments) for what kinds of efforts to link concepts, rules, and cases from distinct spheres of the law are credible and what kinds are not. The net result is that lawyers have a sense, different from and more consistent than laypersons' comparable sense, of what constitutes a plausible or forceful reference or comparison and what constitutes a weak or strained one. [A second set of conventions existing among judges] relate[s] to common standards of judgment in case handling: for example, . . . a willingness among judges to adhere to existing precedent . . . . This last category of conventions . . . consists of a collection of intertwined sensitivities: a feel for what sorts of decisions can fhirly be described as "clarifications," "reforms," or "adjustments" of the body of law and what must be described as "departures" from it; a suspicion of the latter; and a distrust of all lines of argument that would lead toward the latter. It is tis last-mentioned group of attitudes . . . that I think Cardozo was referring to when he spoke of "the duty of adherence to the spirit of the law." Put more bluntly, it is an awareness of where the trodden path lies, and a sense, not only of the danger, but of the impropriety of straying too far from it. Id. (citations omitted). 564 NOTRE DAME LAW REVIEW [Vol. 67:553 from which is likely to be viewed in the relevant community as appropriately subject to criticism or other sanction. Indeed, Hart's mode of analysis is entirely consistent with a view, often associated with Wittgenstein, that inverts the usual understanding of the relationship between some (if not all) legal rules and agreement in legal judgment.' According to this theory, rules and conventions do not cause agreement; rather, it is the fact of agreement that allows us to say there are rules or conventions.' In order for rules to be constitutive of a legal system, lawyers and judges do not need to be able to recite (all of) the rules that they follow, nor must rules enter the conscious mind, as such, in order to guide anyone to a legal conclusion. With respect to the interpretation of precedent, for example, it is sufficient that there is substantial uniformity of judgment, that there is expressed disapproval of deviant judgments, and that a threat of disapproval of deviant reasoning is experienced by lawyers and judges as an aspect of the social reality that is the law. For other accounts of the nature of convention and its role in legal reasoning, see Owen M. Fiss, Conventionalism 58 S. CAL. L. REV. 177 (1985); Dennis M. Patterson, Laws Pragmatism: Law as Pracdie & Narrative, 76 VA. L. REV. 937 (1990); Postema, supra note 55. 67 For a discussion of the appropriate community, see supra note 32. 68 For a sensitive exposition of this view and a discussion of its debated relation to Wittgenstein's own thought, see Radin, supra note 63. 69 Stanley Fish has argued that the term "rule" is misapplied if used in the former sense and redundant or misleading if used in the latter. See, e.g., Stanley Fish, Fish v. Fiss, 36 STAN. L. REV.- 1325 (1984). Fish's argument, roughly summarized, is that it is impossible to formulate- the "rules" of legal practice independently of the goals and values characteristic of that practice and the situations within the practice where legal interpretations are offered. Once the situations for the rules' application are specified in this way, however, the "rules" become redundant, since anyone who grasps the situation will understand how to proceed without reference to the rules. My own tentative view is that this critique helps to bring out the mistake of thinking that all rules, just because they are "rules," must bear the same relationship to behavioral regularities or mental phenomena. In law, I believe a crucially dialectical relationship exists between rules and conventions-which do, to be sure, importantly reflect or summarize characteristic behaviors of those engaged in legal practice-and the practice that the rules purport to regulate. For a participant in the practice, in times of uncertainty and puzzlement about how she ought to proceed, may attempt to formulate the rules for herself in order to be able to refer to them to guide her behavior. See Fiss, supra note 66. 70 To use the term "rule" this capaciously-as Hart does, for example, in characterizing the "rule of recognition" as a social "rule"--is arguably misleading. This usage goes substantially beyond Frederick Schauer's helpful definition of rules as entrenched generalizations. See SCHAUER, supra note 36. The "rule of recognition" could plausibly be characterized as a complex "practice" of law-interpreting officials that defies reduction to a single rule. See Postema, supra note 55, at 168. If the rule of recognition is really a practice, however, the practice is very much a social one, aimed at achieving coherence and coordination, see id. at 192-93, and the guiding interest in achieving coherence and coor- 1992] DWORKIN AND THE TWO FACES OF LAW To put the point another way, Hart's approach illuminatingly 72 71 parallels the work of other philosophers and social scientists wh6 have theorized about what is sometimes called "the social construction of reality."' Phenomena such as law, legal systems, rules, conventions, and principles are not part of the physical universe; they are socially created, and there is little if any natural necessity about their form or content. But though phenomena such as legal 'systems, legal rules, and conventions of legal reasoning are human creations-imaginably different and susceptible to change-they confront particular individuals as external and frequently coercive.74 Indeed, their externality is a social fact, re7 5 flected in pressure to conform. The socially constructed aspect of legal practice deserves both elucidation and emphasis. Judges undoubtedly have to engage in interpretation, at least in some sense of that term, and Dworkin has made a major contribution both in emphasizing this and in showing how complex an enterprise legal interpretation may be. But it is helpful to be as clear as possible about what judges have to interpret, about the constraints to which they are subject, and why.76 To my mind, Law's Empire's emphasis on "protestant" inter- dination gives rise to and reinforces "regularities in . . . identifying, interpreting, and applying rules of law [that] amount to conventions around which mutually interdependent expectations are focused." Id. at 194-95. To my mind, the conventional aspect of law and legal interpretation and the social pressures surrounding them provide ample reason for assimilating what Hart calls the "rule" of recognition to social rules and conventions. 71 See, e.g., PETER WINCH, THE IDEA OF A SOCIAL SCIENCE (1958). 72 See, mg., CLIFFORD GEERTZ, LOCAL KNOWLEDGE (1983); INTERPRETIVE SOCIAL SCIENCE: A READER (Paul Rabinow & William Sullivan, eds. 1979). 73 e P=E L. BERGER & THOMAS LUCKMANN, THE SOCIAL CONSTRUCTION OF RE-' ALrrY (1966). 74 See id. at 60-61 (explaining how the "institutional world ... is experienced as an objective reality," even though it is in one sense a product of human thought). 75 &e ANDREW ALTMAN, CRITICAL LEGAL STUDIES: A LIBERAL CRITIQUE 179-86 (1990). 76 In TAKING RIGHTS SERIOUSLY, supra note 1, Dworkin argued that Hart's concept of a rule was too narrow and unbending to capture all of the constituent components of a legal system; he viewed "principles" as necessary elements as well. In LAW'S EMPIRE, supra note 1, however, although he allows that rules and conventions have a place in pre-interpretive understandings of law, id. at 65-66, Dworkin appears to believe that their role in interpretive accounts will be contingent at best. He offers two main reasons. First, Dworkin believes rule- or convention-based accounts to be incompatible with the phenomenon of theoretical disagreement in law that he wishes to explain: the phenomenon that lawyers and judges frequently disagree in their judgments concerning what the pertinent rules or conventions are or whether they apply to particular cases: See id. at 122-30. Clearly, however, it must be compatible with the concept of a rule or convention for there to be disagreement about what a rule or convention means, or even whether a particular rule or convention exists at all. See Morawetz, supra note 49, at 215-20. As I NOTRE DAME LAW REVIEW [Vol. 67:553 pretation tends to slight the social forces at work in giving rise to and maintaining the phenomenon we know as "law." III. MORALIZED INTERPRETATION Like his closely related claim concerning interpretive protestantism, Dworkin's argument that legal interpretation necessarily attempts to portray legal authorities in the best moral light almost inevitably clouds the centrality of convention and facticity to the concept of law. suggested above, see supra text accompanying notes 68-70, the best measure of the existence of a rule is frequently agreement in judgment; a positivist or conventionalist theory need not enter the debate whether rules or conventions actually cause agreement, or whether it is the fact of agreement that causes us to say there are rules or conventions. It suffices to sustain a rule-based account if there is substantial uniformity in judgment, there is expressed disapproval of deviant judgments, and the awareness of potential disapproval if legal reasoning is not carried on in ways conventionally viewed as correct is experienced by lawyers and judges as an aspect of the social reality that is the law. Moreover, as Hart acknowledges may be the case with what he calls the rule of recognition, there is no conceptual difficulty about recognizing that rules or conventions can be vague, conflicted, or indeterminate. However indeterminate the conventions that comprise a legal system such as ours, the basic, positivist point still holds: the principles, standards, and criteria that jointly comprise what Hart calls the rule of recognition owe their status as such to understandings of the legal profession that exist as a matter of fact, not as a matter of moral or logical necessity. What counts as "law" and "legal argument! is deeply and thoroughly contingent, but all of the contingencies are best explained by reference to the facts of social life and legal practice. Here Dworkin lodges a second objection. He believes that rule- or convention-based accounts of the criteria of validity within a legal system are inadequate to explain the phenomena of legal creativity and legal change. See LAW's EMPIRE, supra note 1, at 13637. Once again, however, Dworkin's conception of a rule or convention is unnecessarily static and unbending. The set of conventions specifying how the law should be identifled-what Hart calls the rule of recognition-may change, just as other social conventions and practices change. See JOSEPH RAz, THE AuTnoary OF LAW. ESSAYS ON LAW AND MORALITY 98 (1979) ("The rule of recognition is a customary rule; hence it is constantly in the process of change."); Gerald J. Postema, Tim Normativity of Law, in IsSUES IN CONTEMPORARY PHILOSOPHY, THE INFLUENCE OF H.LA. HART, at 81, 100-01 (Ruth Gavison ed., 1987); Postema, supra note 55, at 178-79. What is met with criticism today may not meet with criticism tomorrow. A court may refuse to accept some aspect of the rule of recognition, and it may even happen, though it need not, that "a very surprising piece of judicial law-making concerning the very sources of law may be calmly 'swallowed.'" HART, supra note 3, at 150. Within our constitutional law, for example, it is easy to imagine that this is what happened during the New Deal: beginning in 1937, the Supreme Court simply ceased to recognize as law rules and principles that had previously enjoyed that status-not only rejecting Lodiner but effectively rewriting the commerce and contracts clauses as well. If a transformation of this kind occurs, the rule of recognition will itself have been* altered. But the system, as Hart recognized, can remain rule- or conventiongoverned in its fundamental operation. See id. 1992] DWORKIN AND THE TWO FACES OF LAW A. Law and the "InterpretiveAttitude" Although Dworkin founds his theory of law on an account of how judges do and should decide hard cases, he furnishes no strong reason for thinking that judges must approach hard cases the same way in every system that we would characterize as a legal system.7 7 Nor is it clear that he means to claim this. 7 At times, he suggests that jurisprudence is not properly concerned with law as a general, cross-cultural phenomenon, but is always and irreducibly interpretive of the practices of a particular society.7' At other times, however, his arguments suggest that it is precisely because law is an inherently interpretive concept, regardless of its cultural context, that the most interesting substantive questions are likely 8 to involve particular interpretations of particular practices. 0 The underlying issue is a hard one. A deep understanding of law must grasp its significance in a humanly constructed, culturally s contingent, and variable universe of meaning.8 At the same time, illuminating cross-cultural comparisons are possible, and these require the deployment of a conceptual apparatus adequate to mark the characteristic features of phenomena such as law and legal systems that permit them to be identified, referred to, and contrasted. Hart, for example, aspires to identify the characteristic, cross-cultural features of law that make comparative jurisprudence possible, but tries to preserve a space for local, hermeneutic insight by insisting that he is elucidating the concept of law, not providing a hard definition. 2 This is a fine line to walk, but it has the virtue of making comparative social science intelligible. If Dworkin's theory is to do similar work, he too must make some claim about the nature of law that is similarly cross-culturally appli- 77 Cf. Philip Soper, Dworkin's Domain, 100 HARV. L. REV. 1166, 1166-67, 1181-82 (1987) (book review). 78 See LAW'S EMPIRE, supra note 1, at 411 ("General theories of law, for us, are general interpretations of our own judicial practice."). 79 See, e.g., id. at 79, 102, 423 n.15. 80 If Dworkin is not making a claim of this kind, it would seem to follow that he is not so much disagreeing with theories like Hart's as attempting to refocus the question that theories of law attempt to answer. For an illuminating discussion, see Kenneth Kress, The Interpretive Turn, 97 ETHICS 834, 842-43 (1987). 81 On the difficulties of cross-cultural applications of concepts whose meanings may be culturally dependent, see, ag., GEERTZ, supra note 72; 2 CHARLES TAYLOR, INTERPRETATION AND THE SCIENCES OF MAN IN PHILOSOPHY AND THE HUMAN SCIENCES: PHILOSOPHICAL PAPERS 13 (1985); WINCH, supra note 71. 82 See HART, supra note 3, at 16-17. NOTRE DAME LAW REVIEW [Vol. 67:553 cable. Although his signals are mixed, I take him to be claiming that a central, recurring characteristic of "law" as we know it is that it must be interpreted, and that its interpretation requires the adoption of an "interpretive attitude."' Judges, in other words, must assume that law and legal practice "serve[] some interest or purpose . . . that can be stated independently of just describing the rules that make up the practice," and they must further assume that "the strict rules must be understood ... or modified or qualified or limited" by the practice's "point."" Based on this understanding they must "try to impose meaning on the [practice]-to see it in its best light-and then to restructure it in light of that meaning."8 Whether taken as a cross-cultural statement about law or a more localized assertion about American practice, the claim that judges must attempt to see law and legal practice in their "best light" and to "restructure" them accordingly seems to me a strong one. As with his claims about interpretation being a "protestant" enterprise, Dworkin is attempting to account simultaneously for constraint and individuality, social control and individual moral responsibility;8 6 but, again as with his characterization of interpretation as protestant, he cannot entirely have it both ways. In Law's Empire, his emphasis seems unmistakably individualistic and moralized. Assessed in relative isolation, the claim that the nature of law inherently requires judges to assume a moralizing "interpretive attitude" 87 invites persuasive objections. Descriptively, we should not mistake the relatively open-ended character of American practice for a necessary truth about law. In other systems, rules may bind more tightly. Moreover, even within our system, it is clear that judges can, and frequently do, embrace the apparent clarity of familiar rules or seek the resolutions that they believe least likely to incur resistance from professional colleagues.' Nor, when they do so, do they necessarily violate the conventions that define the minimal obligations of judicial office.89 Indeed, if we 83 See LAw's EMPIRE, supra note 1, at 47. 84 85 86 87 88 89 (1983) Id 1 See i& at 88. IdSee Meir Dan-Cohen, Law, Community, and Communication, 1989 DUKE LJ. 1654. Cf Robert M. Cover, Foreword: Nomos and Narrative, 97 HARV. L. REV. 4, 53-60 (criticizing the "apologetic and statist orientation" of "current understandings" of 19921 DWORKIN AND THE TWO FACES OF LAW want to understand law as a social phenomenon, it is crucial to appreciate how legal rules-including relatively clear rules of jurisdiction and hierarchy-can sometimes insulate judges from a psychological sense of responsibility for their decisions. This psychological sense of insulation may help in turn to explain why unjust rules of substantive law can be more or less routinely enforced 9 even by judges who apprehend the injustice. 0 The debate, however, is by no means entirely descriptive. Dworkin frames his position at least partly in rebellion against the prospect of judicial self-distancing from substantive injustice enforced in the name of law; he sees it as morally desirable to portray the judicial function as one of attempting, within what seem in Law's Empire to be broad bounds, to bring positive law into line with moral justice. 91 If adopted as a practical strategy, however, the partial equation of legal with moral right seems to me' to underestimate the significance of contingent social circumstances. If legal and social conventions are strong and tight, judicial resistance in the name of justice is unlikely to occur and even less likely to succeed. If, on the other hand, legal and social conventions are loose and weak, to deprecate the significance of substantive and jurisdictional rules for judicial decisionmaking is not only to invite unpredictability and disorder but, in reasonably democratic regimes, to subvert democracy. Courts may be better at identifying right and justice than other branches of government, but they may also be worse. At stake are issues of power, which should be debated in those terms, and not resolved as an aspect of a claim 92 about the nature of law. the judicial role that allows judges to escape responsibility for the ultimate moral rightness or wrongness of the decisions they render in the name of the law). 90 See id. In the American legal system, rules of jurisdiction and hierarchy are relatively rigid and uncontested, and their widespread recognition and enforcement as rules play a crucial system-maintaining role. See id. at 53-60. A theory characterizing legal interpretation as a deeply moralized and protestant enterprise not only fails to illuminate, but actually obscures, the roles of jurisdictional and hierarchical rules. See id. at 35 n.98. 91 See LAW'S EMPIRE, supra note 1, at 90-113 (arguing that a conception of law, including its theory of adjudication, must be tested for moral attractiveness as well as fit with existing practice). 92 Cf. Frederick Schauer, Formairn, 97 YALE LJ. 509, 542-43 (1988) (arguing that questions about whether judicial decisionmaking should be rule-bound are concerned with power and its allocation). NOTRE DAME LAW REVIEW [Vol. 67:553 B. Legal and MoralJustification These, however, are not the arguments that Dworkin principally engages in Law's Empire. His aspiration, clearly and admirably, is to recast a tired debate by establishing a new argumentative connection. Dworkin argues for moralizing the concept of law to the extent that he does because he thinks this approach is necessary to justify an affirmative answer to the question whether there is a moral obligation to obey the law.93 Dworkin begins with the proposition that "[a] state is legitimate [only] if its constitutional structure and practices are such that its citizens have a general [moral] obligation to obey political decisions that purport to impose duties on them."94 Reasoning against this background, he claims that every legal argument must presuppose an argument about the moral legitimacy of the legal regime. More precisely, every legal argument reflects a conception of legal practice, and every conception of legal practice must "deploy some argument why law on that conception provides an adequate justification for coercion." 5 With these structural premises in place, the way to the conclusion seems plain: the necessary aim of legal arguments, and especially of those offered in support of a judicial decision, is to justif ) the exercise of the state's coercive force in moral as well as legal terms. On the surface, this is by no means an implausible position. Law aspires to guide human conduct by furnishing reasons for action. Although reasons are capable of diverse classifications, a familiar scheme identifies two relevant types: self-interested reasons and moral reasons.9 6 It is implausible that a legal system could regularly furnish self-interested reasons to do what the law requires. If there is anything like a generally applicable reason to obey the law, it is therefore tempting to think that the foundation must be moral. When argument proceeds along this line, the relevant question is obvious: what counts as a claim concerning a judicial decision's moral legitimacy? At a minimum, Dworkin insists that a judicial justification must assert a decision to be dictated by the See LAw's EMPIRE, supra note 1, at 190-91. Id. at 191. Id.at 139. 96 See Joseph Raz, Hart on Moral Rights and Legal Duties, 4 OXFORD J. LEGAL STUD. 93 94 95 123, 131 (1984). 1992] DWORKIN AND THE TWO FACES OF LAW moral theory that paints past political decisions, including statutes and legal precedents, in the best moral light. This, undoubtedly, is justification of a sort. If we accept that legal decisions are morally legitimate insofar as they reflect the best interpretations of the past political decisions of a regime that is morally adequate, and if we further accept Dworkin's controversial theory of adjudication, then all judicial decisions aspire to furnish moral justification for the exercise of the state's authority. Under this view, however, it is conceivable that a decision enforcing chattel slavery could reflect the best interpretation of a political regime that was not so pervasively unjust that its past political decisions could justify nothing. 7 If this can be so-if a judicial decision can 'Justify" slavery as a matter of law-do we want to say that the judge in asserting the arguments that establish the result necessarily claims that the result possesses moral, as opposed to legal, legitimacy?9" Dworkin's ambivalence emerges in response to this question. In characteristic style, his argument depends on a distinction. Legal arguments, he explains, justify morally by displaying the virtue of "integrity" -- the virtue of a community that keeps faith with the commitments of principle that are implicit in its past political decisions. But the kind of moral justification achieved by arguments appealing to integrity is not necessarily the ultimate justification typically signalled by the moral "ought." Where only the former can be achieved, a legal justification will justify morally, but only in an inferior sense that leaves open whether anyone would be morally justified (in the stronger sense) in doing, much less morally required (in the stronger sense) to do, what the law commands."tu The relationship of these two types of moral justification to our ordinary moral vocabularies is never adequately explained, and I doubt that it ever could be. C. Retrospect and Prospect For me, this is the central difficulty with Dworkin's third theory of law. Rather than muddling the concepts of moral justifica- 97 This was State v. Post, 20 98 Many of plainly did not 99 100 the likely situation in the United States prior to the Civil War. &e, e g, N.J.L 368 (1845). the northern judges who upheld and enforced the Fugitive Slave Act think so. For a penetrating study, see ROBERT M. COVER, JUSTICE Ac- cUSED: ANTISLAVERY AND THE JUDICIAL PROCESS (1975). See LAW'S EMPIRE, supra note 1, at 96. &e id. 110-12, 190, 218-19. at NOTRE DAME LAW REVIEW (Vol. 67:553 tion and moral legitimacy in this way, I would prefer to keep law and morals conceptually separate in roughly the way that positivists such as Hart attempt to do. But I, probably in common with a good many others, have learned too much from Dworkin simply to embrace a pre-Dworkinian definition and defense of legal positivism. The testing question for a contemporary positivist or conventionalist theory is whether it can deal adequately with Dworkin's challenges and possibly incorporate his central insights. IV. THE TWO FACES OF LAW The central historical debate that Dworkin's theory attempts to recast is about how best to understand and study a concept that is irreducibly Janus-faced.' On one side, the side that positivism portrays, law is a social phenomenon-an organized, coercive institution for the regulation of human conduct. Although laws and legal systems are frequently unjust, they still function in the characteristic, organized, coercive ways, and 'the fact that they deserve criticism, even condemnation, does not detract from their effectiveness in regulating conduct or from their status as law. On the other side, the side that natural law theories have attempted to capture, the concept of law expresses a human ideal and a standard of criticism. 0 2 When someone says that "an unjust 'law is no law at all," or that "what some have called 'the Nazi legal system' was really just a regime of organized terror," she has surely spoken intelligibly and possibly eloquently. The deep question involves the relationship between law's two faces. A. The Confusions of Amalgamation So far, no one has succeeded in bringing the two faces of law within a unitary analysis, beyond recognizing that both have a claim on our thought and that both inform our vocabulary. To my mind, Dworkin's failure makes it doubtful that a synthesis could ever succeed. Law looks into two worlds, that of the real and that of the ideal, and there is no strong conceptual connection between the two. 03 As Lloyd Weinreb has argued, much of our fa101 Cf Ruth Gavison, Natural Law, Positivism, and the Limits of Jurspradence A Modem Round, 91 YALE L.J. 1250 (1982) (book review) (emphasizing that the natural law and positivist positions, in sophisticated modern formulations, are not so much contradictory as pursuing the study of law in different ways and that each has insights to offer the other). 102 103 See, e.g., JOHN FINNIS, NATURAL LAW AND NATURAL RIGHTS (1980). 1 use the phrase "strong conceptual connection" to leave open the possibility that 1992] DWORKIN AND THE TWO FACES OF LAW miliar moral vocabulary, possibly reflecting a deeper, subconscious understanding, assumes the existence of a fundamentally just natural order. 1°4 Our legal vocabulary, which overlaps substantially with our moral vocabulary, invites similar assumptions about a natural moral order that runs deeper than our conventions.1' 5 In the vision of the better world that the vocabulary of rights and justice invites us to take as foundational of legal discourse, there would be no collision between legal right and moral right, no injustice under law. It is obvious that our world is not that which our discourse, insofar as it sounds in idealized terms, assumes or imagines. But law, as a concept, looks into that ideal world, even as it looks in a quite different direction into the world of social fact. Our understanding of law will be deeply impoverished if we fail to recognize this duality. Where does this analysis leave us in cases in which the issue is joined-in cases, for example, in which the conscientious judge says that the law is that the slave must be returned, and the person characterized as a slave protests that a law so unjust is really no law at all? The answer, no less facially paradoxical for its familiarity, is that they are not necessarily disagreeing about the rules or conventions of the legal system, but, if at all, about what the judge and those subject to judicial orders ultimately ought to do. In short, the disputants are using the word "law" in different senses. The judge (I am assuming) invokes the result dictated by application of settled rules or conventions. The person denominated a slave makes a claim of justice: the 'relevant rules or conventions, law, in order to function as such, must satisfy certain conditions (such as being intelligible and, in general, prospective) that could be argued to reflect an "inner morality of law." &e, eg., Lon L. Fuller, Positivism and Fidelity to Law-A Reply to Professor Har, 71 HARv. L. REv. 630 (1958). As Professor Hart has written, "if this is what the necessary connexion of law and morality means, we may accept it. It is unfortunately compatible with very great iniquity." HART, supra note 3, at 202. By a "strong conceptual connection" between the two faces of law, or between law and morals, I am referring to the kind of connection that would imply that an unjust law, or at least a seriously unjust law, is not law at all. By insisting that no strong conceptual connection exists between law and morals, I also do not mean to suggest that there could be a body of law whose substantive content was morally abhorrent to all relevant segments of the population. For a further discussion of what I call "sociological" connections between law and morals, see infra notes 110-11 and accompanying text. 104 See LLOYD L. WEINREB, NATURAL LAW AND JUSTICE 30 (1987). Similar conclusions are common among social scientists. See, e.g., BERGER & LUCKMANN, supra note 73, at 6162; MELVIN J. LERNER, THE BELIEF IN A Jusr WORLD: A FUNDAMENTAL DELUSION vii-viii (1980). 105 See Weinreb, supra note 34. NOTRE DAME LAW REVIEW [Vol. 67:553 if they dictate a result that is so seriously unjust, should be changed or disobeyed. Any reasons that might ordinarily call for obeying the law, including the judge's promise to do so, must yield to supervening moral imperatives. This analysis, which seeks on the one hand to acknowledge law's two faces, may seem on the other to take away what it purports to give. It may appear to privilege the face of law that is rooted in social fact by assimilating the face that the natural law tradition exalts to the independently identifiable realm of morals. But this, I think, is the best analysis available; if a privileging is detected, I see no way to avoid it. Two reasons support this conclusion. First, whatever else might be said about it, the positivist's usage cannot plausibly be classified as a linguistic error. On the contrary, in our ordinary legal vocabularies the face of law that identifies law and legal systems with social fact seems to be dominant and the natural law tradition recessive. Virtually no one argues that the dictates of moral right are necessarily law within an existing legal system. The natural law influence characteristically asserts itself, if at all, only in legal debates about appropriate interpretive principles or, even more rarely, in assertions that an unjust law is not law at all. Second, and more controversially, the positivist usage can be defended on the ground that it deals more effectively with what I take to be the greater practical danger: that people may be lulled into an association of law as it is declared by established institutions with law as it ought to be and into a confusion of legal duty 0 as identified by state authorities with moral duty." 6 As against Dworkin's amalgamating approach, the reasons for preferring a simple acknowledgement of law's Janus-faced character are even more powerful. Given the existence of law's two faces, the surface allure of Dworkin's theory lies in its insistence that positive law must be understood as aspiring to instantiate a vision of law as it ought to be, and that judges not only can, but should and must, strive (within the limits defined by the requirement of "fit") to bring the "is" and the "ought" into a relationship of identity. As I have suggested already, however, Dworkin's effort to link 106 See Gavison, supra note 101, at 1283. But see J.L. Mackie, Obligations to Obey the Law, 67 VA. L. REV. 143 (1981) (arguing that, although a general obligation to obey the law cannot be derived from any plausible moral theory, we so badly need such an obligation for reasons of social co6rdination that we should "invent" one). 1992] DWORKIN AND THE TWO FACES OF LAW positivism and natural law via a theory of legal interpretation presents us, not with the two clear faces of law, but with a confusing amalgam. The confusion is most evident in those instances when, by Dworkin's own account, the interpretation that is morally best is nonetheless one that yields unjust results."" When this occurs, we have unjust law that will, on Dworkin's theory, be judicially validated as justified, with the claim of justification confusingly straddling the domains of the real 8 and the ideal across which the 1 0 two faces of law respectively look. 107 Se supra text accompanying notes 97-100. Dworkin appears to insist on the conceptual necessity of a link between the two faces of law, with interpretation functioning as the mediating term. His argument, if I understand it correctly, unfolds in two stages. First, law and legal practice are interpretive concepts, and judges must therefore develop their own interpretive theories of what obligations the practice imposes on them. See JAW's EMPIRE, supra note 1, at 50-52, 254-58. Second, given the nature of interpretation, the best interpretation of every legal practice requires judges, in doubtful cases, to base their results at least partly on normative grounds. The best legal theory, which correctly identifies the law in every case, will be the one that achieves the best mixture of "fit" with existing legal materials and normative attractiveness. Id. Regarding this argument, the testing question is whether we can imagine a possible legal system in which the test applied by judges to determine the validity of legal claims does not incorporate a criterion of moral truth or attractiveness. I think that we can-a system, for example, in which the rule is that all doubtful claims should be resolved in favor of the defendant. What is easier to imagine is a system in which the applicable adjudicative principles call for decisions in doubtful cases to conform to conventional morality-that is, to accord with the values that predominate in a particular society-rather than to what is defensible in light of objective moral principle. If so, the applicable test would rely solely on social fact, not on moral right, and Dworkin's claim that the ultimate test of legal validity necessarily includes moral truth as one of its criteria cannot survive. Dworkin may think that he needs to claim less., He might intend only to claim that every imaginable system of legal rules requires "constructive" interpretation, which blends concerns of fit with "normative attractiveness." As I have argued above, however, it is implausible to think that the conceptual logic of law somehow requires judges to adopt what Dworkin calls an interpretive attitude, even when this attitude is permissible. Alternatively, Dworkin might intend that, even under a rule such as that suggested above (doubtful cases are to be decided in accordance with conventional morality), questions will remain about (i) which cases are in fact doubtful and (ii) what conventional morality is or requires and that these questions cannot be resolved unless a judge adopts an interpretive attitude that balances descriptive "fit" with considerations of normative attractiveness. Again, however, this claim seems too strong. It is imaginable, at least, that judges would pursue purely positive or descriptive answers to the unresolved questions as far as human nature permits, with any irreducibly residual influence left to relatively unreflective sympathy, empathy, prejudice, and taste. 108 See supra text following note 105. To talk of law's two faces is of course to speak. metaphorically, but I do not know how to make my point other than by metaphor. Another metaphor, that of Wittgenstein's duck-rabbit-a figure that can be seen either as a duck or as a rabbit-might serve as well. See LUDWIG WrrTGENSTEwN, PHILosoPHIcAL INVESTIGATIONS 194 (G.M.E. Anscombe trans., 3d ed. 1976). Although we can see the figure as either a duck or as a rabbit, it would be very misleading to say that the figure NOTRE DAME LAW REVIEW [Vol. 67:553 B. Is There a Duty to Obey? Dworkin's contrary suggestion notwithstanding, recognizing that the concept of law has two distinct aspects helps to clarify the relationship between legal duties and moral duties. More particularly, this recognition helps to explain why it is natural but erroneous to believe that there is, at least in reasonably just regimes, a "general" moral obligation to obey the law."° First, recognizing law's two faces helps to explain why there is likely to be a very substantial overlap between the positive law and the prevailing moral beliefs in any particular society. Both serve similar interests in minimizing conflict and facilitating communal life." 0 In addition, deep psychological and sociological forces press in favor of "legitimating" explanations of the external world, including institutions, such as the law, that are socially constructed.' If too great a dissonance exists between what the law commands and what is believed to be reasonably just, the situation is likely to be unstable; it may even be pathological. Nevertheless, recognizing the likelihood of overlap between law and morals is far from asserting the existence of a particular, discretely identifiable, conceptual connection between the two. Although it is predictable that there will be substantial connections between what the law in a given society happens to be and what the people of is a combination of a duck and a rabbit. So it is with law. We can see law as externally given and potentially coercive-a matter of social fact. But much of the time, as unalienated citizens of a reasonably just society, we may see the law as typically furnishing reasons for action that sound in moral terms. Cf infra notes 109-14 and accompanying text (discussing whether there is a general moral obligation to obey the law of reasonably just regimes). Indeed, this may be such a familiar aspect of our experience that we would find something seriously amiss, even conceptually problematic, about a body of coercive rules that purported to be law but that provoked in us a sense of moral outrage or alienation. Nonetheless, although law may appear to us in these very different ways, it would be fundamentally misleading to say that law was some mixture or amalgamation of potentially coercive social facts and freely embraced moral norms. Rather, law is a phenomenon that sometimes appears to us in each of the two ways, and our understanding is best expressed by recognizing the duality, not by insisting on a conceptual unity that demands some form of blurring, blending, or amalgamation. 109 The insistence on generality is crucial, since there are likely to be good moral reasons to obey particular laws under virtually any regime. The testing question must therefore be whether the fact of legal obligation furnishes a moral reason to do what the law requires even in cases in which no other moral reason would exist. For a useful discussion of how to sort out this and related issues, see KENT GREENAWALT, CONFLICTS OF LAW AND MoRALTrrY 47-203 (1987). 110 See, e.g., HART, supra note 3. 111 See BERGER & LUCKMANN, supra note 73, at 60-61; LERNER, supra note 104. 19921 DWORKIN AND THE TWO FACES OF LAW that society believe is morally justifiable, the precise nature of the connections is sociologically variable. In turn, the familiar (but not invariable) overlap between conventional morality and substantive law helps to explain why most citizens of most regimes will tend to view the law as reasonably just. Moreover, if citizens tend to regard the law as reasonably just, then it may seem natural, under a variety of -theories, to think that there must be a generally applicable reason to obey." 2 In my view, however, there is no "general" moral obligation to obey the law."' This, I should hasten to say, is a conclusion of limited practical relevance." 4 There often are strong moral reasons to do what is commanded by particular laws, many of which only prescribe *that which would be morally right or forbid that which would be morally wrong anyway." 5 In addition, I am inclined to accept the idea of a moral obligation to support reason6 ably just systems of social cooperation among relevant groups." Laws frequently are instrumental in establishing such systems of cooperation-schemes in which everyone drives on one side of the road and supports the provision of public goods, for example. Nonet heless, the role of law in such cases is instrumental, not intrinsic. The same obligation of support would arise if the coop- 112 For critical surveys of such theories, see, ag., GREENAWALT, supra note 109; RAZ, supra note 76; Mackie, supra note 106. 113 See supra note 109 (discussing the significance of the "generality" requirement). In reaching this conclusion, I am generally persuaded by the arguments offered in RAZ, supra note 76, at 233-49. Raz proceeds largely by dismissing familiar arguments purporting to establish an obligation to obey. He argues, for example, that law-breaking is not necessarily bad on the ground that it sets a bad example; sometimes there will be no example, or the example may, be good. Id.at 237-38. Arguments based on contract fail because there is nothing in the lives of most citizens that could constitute even a tacit promise to obey. Id. at 238-39. Nor are most citizens estopped across-the-board from breaking the law as a result of holding themselves out as generally law-abiding; an estoppei argument would work only in cases involving detrimental effect as a result of actual reliance. Id. Arguments based on asserted duties to uphold just institutions and to respect principles of fair play or fair reciprocation are discussed below, see infra text accompanying notes 118-19. 114 Among other things, this conclusion does not imply that the governments of reasonably just regimes lack legitimate authority to govern; the questions of legitimate authority to govern and citizens' obligations to obey are distinct. See Rolf Sartorius, Political Atlwrity and Political Obligati'n, 67 VA. L. REV. 3 (1981). 115 See eg., RAZ, supra note 76, at 233-49. 116 The leading effort to define and justify such a duty, albeit not one that I agree with in all respects, is JOHN RAWLs, A THEORY OF JUsTIcE 333-91 (1971). NOTRE DAME LAW REVIEW [Vol. 67:553 erative7 scheme arose voluntarily or through extra-legal convention." More importantly, the obligation to support just institutions, as I would construe it, does not extend to cases in which individual action and inaction, if generally engaged in, would have no adverse consequences for the institutions involved.11 The obligation to support just institutions is simply not implicated, for example, if I roll through rather than coming to a full stop at a stop sign at a deserted intersection, or if, after stopping, I drive through a red light on an empty street at 3 a.m. These examples may, of course, seem trivial. What is really at stake, it might be argued, is not so much an instrumental obligation of support but a noninstrumental, notion of reciprocation or "fair play,"119 which requires those who benefit from others' obedience to a reasonably just system of laws to accord the same benefit to others. To my mind, however, any noninstrumental obligation of obedience could attach only to laws that were good in themselves, not to every law just in virtue of its being a law of a reasonably just regime. Consider the laws of otherwise reasonably just regimes that regulate private religious or consensual sexual practices that (in my view) ought, as a moral matter, to be left to individual conscience or choice. In confronting laws that offer seemingly disagreeable prescriptions, the conscientious citizen should perhaps ponder carefully whether the moral judgment that the community has embodied in law might not be better considered than her own.120 If she judges that it is not, however, I see no noninstrumental moral obligation of compliance1-not 117 (1979): See A.D. WOOZLEY, LAW AND OBEDIENCE: THE ARGUMENTS OF PLATO'S CRrro 73 Whatever there is to be said for such a law is whatever there is to be said for the practice which the law requires; and the reason for acting in accordance with the law, or for conforming one's conduct to it, lies in the value of the practice. The practice derives none of its value from the law, and [just because] it is the law provides no reason for conformity .... Id. 118 See GREENAWALT, supra note 109. 119 See, e.g., H.L.A. Hart, Are There Any Natural Rights?, 64 PHIL REv. 175 (1955); John Rawls, Legal Obligations and the Duty of Fair Play, in AW AND PHILOSOPHY 3 (Sidney Hook ed., 1964). 120 See, e.g., SCHAUER, supra note 36, at 125; cf. Heidi Hurd, Challenging Authority, 100 YALE L.J. 1611 (1991) (arguing that the authority of law is "theoretical" rather than practical); Donald H. Regan, Reasons, Authority, and the Meaning of "Obey", 3 CAN. J.L. & JUR. 3, 3-11 (1990) (arguing that law provides "indicative" rather than "intrinsic" reasons for compliance). 121 Throughout this discussion, I mean to put aside the possibility of instrumental 1992] DWORKIN AND THE TWO FACES OF LAW even, as proponents of the position would argue, an obligation susceptible of being overcome by competing moral considerations.1 22 It may be hard to find a moral duty to engage in proscribed forms of sexual conduct;12 but where there are no adverse consequences implicating a moral obligation of support for the law, the existence of a general moral right to engage in legally forbidden behavior will often be dispositive of whether there is an all-things-considered moral duty to obey.1 24 No "general" obligation to obey the law need figure in the calculus.' I If we are inclined to think that there is a general moral obligation to obey the law, applicable even to cases involving traffic regulations on visibly deserted streets and the regulation of private sexual conduct, at least part of the explanation may be that law's moral face encourages us to associate law with that which is moral12 ly right and therefore to equate legal duty with moiral duty. harms to a reasonably just regime of law-harns that might be present in some, but surely not in all, cases. 122 See LAW'S EMPIRE, supra note 1, at 110-13, 191, 202-06. 123 This is not to suggest that morality is exclusively or even primarily bound up with the concept of duty. "Moral" reasons may well exist for engaging in sexual activity-that is, reasons rooted in a conception of the kind of life that is morally healthy or good. 124 Among the possible complicating factors, perhaps the most interesting is Joseph Raz's suggestion that conscientious citizens whose sense of personal identity is linked to their membership in a reasonably just society may have moral reasons, possibly rising to the level of a moral duty, to express their respect for the society through obedience to its laws. Se RAZ, supra note 76, at 250-61. Although not persuaded by Raz's analysis in all respects (among other things, framing the conclusion in the language of "duty" seems to me to be misleading), I agree that self-identification with, and respect for, one's society can generate reasons, including moral reasons, to obey the society's laws, even when the underlying psychological attitude is not itself morally required. Socrates' acceptance of a punishment of death, see PLATO, Crito, in THE COLLECTED DIALOGUES OF PLATO 27 (Edith Hamilton & Huritington Calras eds., 1961), arguably exemplifies both the attitude that I have in mind and the kinds of reasons with which that attitude may be associated. But see Frances Olsen, Socr'ate on Legal Obligation.. Legitimation Theory and Civil Disobedience 18 GA. L. REV. 929 (1984) (arguing that Socrates accepted death by refusing to escape, not to show his respect for the laws of Athens, but to provide a vivid example of, the laws' injustice). 125 Judges constitute a special case in a double sense. First,,judges commonly promise to obey the law as a condition of assuming judicial office. Second, in order for law to fulfill its basic purposes, "it must be possible to regard the activity of law applying as governed by some reasonably coherent pattern," Postemua, supra note 55, at 176, and the requisite coherence might well be impossible if judges did not follow established interpretive conventions. 126 Cf.Sartorius, supra note 114, at 16 ("Within a just society, the general pattern of compliance with what the law requires would be voluntay in the sense that the normal motive for obedience would not be the fear of legal sanctions but rather the belief that what the law required was right independent of the fact that the law required iL"). Professor Lyons apparently agrees, but reasons to the very different conclusion that the only NOTRE DAME LAW REVIEW [Vol. 67:553 But this, as I have suggested, is a mistake insofar as the term "legal" refers to law's positivist face. The magistrate may find a breach of legal duty in the cases I have hypothesized. Depending on the prevailing interpretive conventions, her legal analysis may be correct. Nevertheless, there will be no breach of a moral duty, or even a moral requirement of competing moral reasons to justify disobedience, because there is no general moral obligation to 127 obey positive law. V. CONVENTIONALISM In Law's Empire, Dworkin distinguishes between his formal or meta-theory, which claims that "law is an interpretive concept, "128 and his substantive interpretation, which he styles "law as integrity."' It is an entailment of law as integrity that the law in any particular case is that result which emerges from the theory that portrays the past political decisions of a regime in the best, most consistent light. Dworkin defends this substantive theory partly by 13 which is substanarguing its superiority to "conventionalism," TM tially an improved, modernized version of Hart's positivism.1 morally and theoretically defensible conception of legal justification would be one that equated legal justification with moral justification. See David Lyons, Jusitfcation and Judicial Resonsibility, 72 CA_ L. REV. 178, 193-99 (1984). Lyons's reasoning leads to an implication, which seems to me unsustainable, that an "unjust law" is not really law at all. 127 To say that there is no general moral obligation to obey the law is by no means to imply that law fails in its aspiration to guide conduct by providing reasons for action in many, if not most, cases in which the law of a reasonably just regime applies. For the law of reasonably just regimes will typically provide what Professor Regan has classified as "indicative" reasons for obeyance. In other words, law's being what it is will tend to indicate the existence of some further fact of intrinsic moral relevance. See Regan, supra note 120, at 3-11. The indicated fact might be that particular conduct is morally wrong or offensive to other citizens (where this is the judgment that relevant officials have reached) or that solutions to coordination problems (such as the problem of getting everyone to drive on either the left or the right) can realistically be achieved only on the terms indicated by law. Indeed, the reason-fumishing significance of law is so great with respect to coordination problems, see Postema, supra note 55, that the difference between saying there is a moral obligation to support the reasonably just scheme of social cooperation established by law and saying that there is an obligation to obey the law as law may sometimes be so thin as to border on being artificial. See Mackie, supra note 106, at 150. 128 LAW'S EMPIRE, supra note 1, at 50. 129 Id. at 225. 130 &e id. at 114-50. 131 See id. at 115-16. Dworkin dismisses "positivism" in the early part of LAW'S EMPIRE as a "semantic" theory of law that, he says, fails in principle because debate over what "law" is is not plausibly a debate about the semantic rules governing the use of the word "law." Instead of being semantic, as he claims positivism is, Dworkin thinks a good theory 1992] DWORKIN AND THE TWO FACES OF LAW Among conventionalism's improvements over positivism is its avoidance of misleading connotations that law is necessarily posited or commanded by some identifiable sovereign entity. By contrast, the label "conventionalism" suggests that the foundations of law lie in the kind of social situation in which conventions tend to develop: "[C]onventions are not mere regularities of behavior but, rather, regularities arising out of and reinforcing a system of mutual expectations and a commonly recognized need for coordinated activity."132 Within the terms of this definition, many legal rules or interpretive practices are conventions in the technical sense: they reflect a response to situations in which, unless coordination is achieved, widely shared moral or practical interests will be frustrated, and in which it may be more important that there should be some coordinating convention than that the convention should have a particular content.1 3 To take a familiar example, it is more important that everyone should drive on the right or on the left than that it be either the right or the left on which everyone of law must be "interpretive" in the sense described above. See supra Part III. "Conventionalism," in Dworkin's usage, is an interpretive theory that incorporates most of the substantive features of "positivism." There is one possibly significant difference. Consistent with his own generally "adjudicavist" approach to the concept of law, see Stephen -R. Perry, Second-Order Reasons, Uncertainty and Legal Tlory, 62 S. CAL. L. REV. 913, 958-59 (1989), Dworkin conceives the point or function of law under a conventionalist theory to be one of providing fair warning of legal consequences and protecting reasonable expectations. See LAW's EMPIRE, supra note 1, at 95, 117. Under traditional positivist approaches, the function of law is more centrally one of guiding behavior by providing reasons for action. See, eg., RAZ, supra note 76, at 50-51; Perry, supra, at 950-62; Postema, supra note 55, at 187-88. I shall use the term "conventionalism" more or less interchangeably with "positivism" without endorsing Dworkin's ascription of purpose. Dworkin also discusses "pragmatism" as a second rival to his own theory. "Pragmatism," as defined by Dworkin, holds that judges are free to decide cases however they choose but that judges characteristically should and do decide cases in ways that promote the society's values and policy interests. See LAW's EMPIRE, supra note 1, at 72, 147-49, 151-53. I have serious doubts about this account of a pragmatist theory of law. Like" Dworkin's own theory, legal pragmatism as described by him seems more a theory of adjudication than a theory of law. Pragmatism, as he portrays it, cannot account for how law operates in social life outside the courts or for why judicial decisions should be taken as authoritative. For these purposes, some account rooted in social fact remains needed. It is also unclear how Dworkin's "pragmatism" maps onto that of the apparently burgeoning school of self-styled legal pragmatists, whose views are vastly more diverse and complex than Dworkin's ideal type would suggest. See e.g., RICHARD POSNER, THE PROBLEMS OF JURISPRUDENCE (1990); Daniel A. Farber, Legal Pragmatism and the Constitution, 72 MINN. L. REV. 1381 (1981); Symposium on the Renaissance of Pragmatism in American Legal Thought, 63 S. CAL. L. REV. 1569 (1990). 132 Postema, supra note 55, at 176. The undoubted inspiration for modem conventional theories of law is DAVID LEWIS, CONVENTION: A PHILOSOPHICAL STUDY (1969). 133 See Postema, supra note 55, at 199. NOTRE DAME LAW REVIEW [Vol. 67:553 drives. Having suggested my sympathy with a conventionalist approach to the study of law, I want briefly to address Dworkin's opposing arguments. A. , Dworkin's Attack Conventionalism, as defined by Dworkin, follows Hart in rooting law in the facts of existing practices, social rules, or conventions.M According to conventionalism, law consists of conventions or other precepts that reflect widely shared agreement about the existence or nonexistence of legal duties, including conventions governing the practice of legal interpretation. Dworkin's chief claim against conventionalism is that it cannot account for judicial decisionmaking in hard cases or, more generally, for disagreement in law: if the governing conventions are valid only in virtue of being accepted as such, how can we explain the familiar kinds of disagreement among judges and lawyers about What the relevant conventions are and about how they apply in particular cases?' s 5 Having knocked down conventionalism as a view that treats all legal questions as questions of fact about the existence and relevance of legal rules, Dworkin briefly introduces a second version 136 of conventionalism, which he labels "soft conventionalism." Soft conventionalism adopts the positivist tenets that the existence and validity of law are matters of social fact and that there is no strong, conceptual connection between law and morals. 37 But soft conventionalism departs from the "harder" variety in recognizing that, as a matter of contingent fact, the conventions of any particular legal system may specify some form of appeal to moral principles as the appropriate means of dispute resolution in cases in which the more narrowly legal materials, understood in the conventional ways (which may be more or less determinate), fail to provide a uniquely correct answer. Because many of the relevant rules or conventions may be vague or conflicted, and because the interpretive exercise necessary to resolve doubtful cases may be 134 See LAw's EMPIRE, supra note 1, at 114-17; see also Postema, supra note 55, at 16667 (characterizing the rule of recognition that provides the foundation for Hart's theory of law as a "convention"). 135 See LAw's EMPIRE, supra note 1, at 126-39. 136 Id at 124-30. For examples of other theories that would fit the description, see, e.g., Jules L. Coleman, Negative and Positive Positivism, 11 J. LEGAL STUD. 139 (1982); David Lyons, Principles, Positivism and Legal Theoly, 87 YALE LJ. 415 (1977). 137 See supra note 103. 19921 DWORKIN AND THE TWO FACES OF LAW contestable, soft conventionalism can explain why judges will predictably diverge in their judgments about how cases should be decided and about what factors judges should rely on in deciding cases. Yet, as Professor Coleman has put it, "[t]he controversy among judges does not arise over the content of the [convention prescribing reference to principles or policy goals]. It arises over which norms satisfy the standards set forth in [that convention].,,"s8 B. The Virtues of "Soft Conventionalism" Soft conventionalism is unlikely to satisfy anyone's notions of theoretical elegance." 9 It is something of a makeshift, founded on the view that law must be rooted in social fact, but adapted to accommodate three plain truths emphasized by Dworkin. First, judges and lawyers frequently write and argue as if there were more law than can be accounted for by any theory that makes the existence of uniquely correct outcomes, rules, or conventions depend on virtual consensus in factual judgment. Second, in the American context at least, the phenomena of legal evolution and change can be explained only by recognizing that judicial reasoning frequently has a "political" aspect that is related by family resemblance to, yet is distinct from, many other forms of decisionmaking familiarly characterized as "political." 4 ° Third, there is no recognized distinction in our legal practice between an aspect of judicial reasoning aimed at discovering the law and an aspect that is aimed, after the law is determined, at filling identified gaps or implementing reforms. I had these insights in mind when, at the outset, I termed Dworkin's theoretical efforts "deeply instructive." Soft conventionalism, however, can embrace these insights, while at the same time explaining in a way that Dworkin's own theory does not how law is rooted in social fact. In addition, within the American system at least, soft conventionalism provides a plausible account of how ideology and moral argument can per- 138 Coleman, supra note 136, at 156. 139 See, eg., Kent Greenawalt, The Rule of Recognition and the Constitution, 85 MICH. L. REV. 621, 667-68 (1987). 140 On different senses of "political" and their relevance to analysis of judicial decisionmaking, see Richard Hodder-Williams, Six Notions of "Political" and the United States Supreme Court, 22 BRIT. J. POL. Scl. 1 (1992); Richard H. Fallon, Jr., Common Law Court or Council of Redsion?, 101 YALE LJ. 949, 960-61 (1992) (book review). NOTRE DAME LAW REVIEW [Vol. 67:553 meate legal debate while at the same time playing a subordinate role limited by the facts of social practice. To see that the role of morals is subordinate, and restricted by convention, it suffices to keep two points in mind. First, conventions do far more work than sophisticated debate frequently imagines, often by defining the realm of the legally unthinkable. Within our practice, for example, it is simply unimaginable-there are no plausible legal arguments holding-that the first amendment forbids me to yawn or that the Koran is the law of the United States.141 Second, the moral arguments that count as legal arguments are restricted by the rules or conventions (and in some cases by virtually unanimously acknowledged rules or conventions) of our legal practice. "It is God's will" is not a legal argument within our practice, because the governing conventions rule it out of bounds, even though the governing conventions do not rule other moral arguments about fairness and efficiency similarly out of bounds. In other words, a partial separation exists between law and morals in our practice; and the question of which moral arguments also count as legal arguments is one that can only be answered by reference to social fact. A judge or lawyer may dismiss an argument based on God's will as legally irrelevant, not because she believes that there is no God or that God's will cannot be known, but simply because the rules or conventions of legal practice provide an authoritative legal reason for refusing to take theological morals into account. VI. CONCLUSION: DwoRKIN AND "SoF CONvENTIoNALIsM" In Law's Empire, Dworkin does not attempt to refute "soft conventionalism" directly. Instead, he tries to subsume it as an 4 "underdeveloped form" of his own theory." In my view, however, Dworkin has matters almost precisely backwards. As I have suggested above, his theory lacks a sufficiently rich account of the nature of a legal system, and of how law functions in everyday life, to stand as a comprehensive theory of law in the same sense as Hart's theory, for example. Furthermore, Dworkin's claim that all judges must always attempt to portray existing legal materials in 141 For forceful expressions of similar points, see Ken Kress, Legal Indeterminacy, 77 CAi. L. REV. 283 (1989); Lawrence B. Solumn, On tie Indeterminacy Crisis: Critiquing Critlcal Dogma, 54 U. CHI. L. REV. 462, 470-73 (1987). 142 See LAw's EMPIRE, supra note 1, at 127-28. 19921 DWORKIN AND THE TWO FACES OF LAW 585 the best moral light is implausible. To put the point sharply, Dworkin is least illuminiting when talking about what "law" is as a general or conceptual matter; he is most illuminating (even if not always entirely persuasive) when talking about how anyone-and especially judges-should identify what "the law" is in the special context of the English and American legal systems. Nor, I hasten to add, should this characterization be regarded as in any way trivializing Dworkin's enterprise. Though broader in compass, the question "what is law?" does not necessarily have greater philosophical depth or practical importance than more local questions about what the law is or whether particular laws ought to be obeyed. By contrast with Dworkin's theory, soft conventionalism is a theory of law in the traditional sense. A soft conventionalist can follow Hart in recognizing that law is grounded in the facts of actual social practice. But soft conventionalism can also, as Dworkin acknowledges, accept the possibility that the reigning conventions (or what Hart has called the rule of recognition) in certain legal systems-such as those of England and the United States-might put judges under a conventional or social duty to resolve controversial questions, including questions about the rule of recognition itself, by application of interpretive techniques that may sometimes include appeal to moral principle. If this is so, someone holding a conventionalist theory might accept or incorporate Dworkin's theory as the best account of how judges do and should decide cases in the English and American systems-though she would not, have to, since rival accounts that are equally compatible with conventionalism are also available. 4 ' 143 For example, my own thoroughly conventionalist theory of constitutional interpretation, though much influenced by Dworkin, departs from his theory in important respects. See Fallon, supra note 18, at 1233-37 & nn.206, 213.