1. I use “legitimate” and its cognates in their normative, moral senses and not in either their legal senses or their empirical, descriptive, or sociological senses. Thus I am not considering here questions about the de facto support that a population has for the extant political regime or constitutional system, nor about the relation between (normative) legitimacy and (empirical) stability. For more on this distinction, see Christopher F. Zurn, Deliberative Democracy and the Institutions of Judicial Review (2007), at 76–77. That book also contains my considerations in support of the normative proposition—simply presupposed here—that constitutionalism and democracy are both required for political legitimacy.

2. Because the trouble is generic, I do not intend the positive arguments of this paper to rely upon any particular substantive conception of the legitimacy of constitutional democracy. Of course, the trouble would not arise if constitutional democracy were not required for political legitimacy.

3. Randy E. Barnett, Restoring the Lost Constitution: The Presumption of Liberty (2004).

4. Frank I. Michelman, Brennan and Democracy (1999); Michelman, Constitutional Authorship, in Constitutionalism: Philosophical Foundations (64–98) (Larry Alexander ed., 1998); Michelman, How Can the People Ever Make the Laws? A Critique of Deliberative Democracy, in Deliberative Democracy: Essays on Reason and Politics (145–171) (James Bohman & William Rehg eds., 1997); Michelman, Jürgen Habermas's Between Facts and Norms, 93 J. Phil. (307–315) (1996) (book review). I think Michelman's clearest formulation of the paradox is to be found in Constitutional Authorship, though only in Brennan and Democracy does he suggest determinate ways to come to terms with the paradox in practice.

5. See Barnett, supra note 3, at 11–31, referring to Lea Brilmayer, Consent, Contract, and Territory, 74 Minn. L. Rev. (1–35) (1989). Specifically, his argument is directed against contractualist political theories in a broad sense but is not directed to contractarian or contractualist moral theories. I ignore here Barnett's (unacknowledged) replay of Hume's arguments against consent, to the effect that various empirical circumstances that individual subjects find themselves within under already constituted states render their putatively free consent questionable and therefore unreliable as a basis for claiming legitimacy. These familiar arguments—e.g., that acquiescence to the law might reflect the extremely high cost of exit rather than tacit consent to the law—are not conceptual barriers to a contractualist account of legitimacy, no matter how weighty they may be under actual empirical conditions.

6. For a powerful institutional analysis, see Mark A. Graber, Dred Scott and the Problem of Constitutional Evil ch. 5 (2006).

7. While slaves were legally cognized as chattel and were legally barred from enjoying any of the privileges and immunities of citizenship, they were notoriously counted (at a 40 percent “discount”) toward the population of their respective states, thereby giving a very important numerical boost to the Slave Power states with respect to their proportional representation both in the House of Representatives (U.S. Const. art. I, § 2, cl. 3) and in the selection of the President through the Electoral College (U.S. Const. art. II, § 1, cls. 2–3; and amends. XII and XXIII). For more on the expected and actual historical interaction between federal representation and the counting of slaves, see Graber, supra note 6, pt. 2. The USC itself also gave a twenty-year safe harbor to the international slave trade by barring in art. I, § 9, cl. 1 any federal laws that might outlaw the importation of slaves and it included a fugitive-slave provision in art. IV, § 2, cl. 3, requiring nonslave states to return any escaped slaves to their owners.

8. See the way in which the amendment procedures specified in U.S. Const. art. V ensure the effective unamendability of USC provisions concerned with the allocation of Senate seats (since every state must consent to any re-allocation); and ensure the actual unamendability of the twenty-year guarantees of both the slave trade (U.S. Const. art. I, § 9, cl. 1) and counting slaves as three-fifths of a person when calculating a capitation tax (U.S. Const. art. I, § 9, cl. 4).

9. Other groups of persons who were also constitutively left out of the U.S. constitutional settlement present more complex examples of delegitimating exclusion: adult females, free men without significant personal property, and Native Americans. Yet I think it safe to say that in different ways they were constitutively barred from the possibility of consenting to or withholding consent from the colonists and their new governments, not to mention constitutively excluded from collective political democracy.

10. For discussions of the ways in which art. XIII of the Articles of Confederation were ignored in designing the USC's ratification procedures, see David Kay, The Illegality of the Constitution, 4 Const. Comment. (57–80) (1987); and Bruce Ackerman & Neal Kumar Katyal, Our Unconventional Founding, 62 U. Chi. L. Rev. (475–573) (1995). The Federalist Nos. 40, 43 (James Madison) explicitly recognize that the ratification procedures depart from those specified in the Articles, but Madison dismisses the significance of this objection on the grounds that the Articles' unanimous-state-consent provision would give too much veto power to a single intransigent state and that the antifederalist opponents of the USC had not really pressed the objection. Note that neither of Madison's arguments is responsive to the problem of procedural illegality.

11. See the documentation of the claim that amends. XIII and XIV were strictly illegal constitutional amendments in Bruce Ackerman, We the People: Transformations (1998), at 99–252. The claim is contested in Akhil Reed Amar, America's Constitution: A Biography (2005), at 364–380.

12. The stories are importantly different. The German Basic Law was written largely by delegates appointed by the heads of the German states in 1948, approved by the state parliaments, and promulgated in 1949. Nevertheless, the Occupying Powers retained effective veto power over the proposal should it have not met with their approval. The Japanese constitution was mostly written by two U.S. military lawyers in 1946 after the Occupying Powers disapproved of an earlier indigenous proposal, and was ratified in national legislatures that same year.

13. The story is considerably more complex than this—involving the United States' direct rule through the Coalition Provisional Authority from April 2003 to June 2004, then the rule of the Iraqi Interim Government under the aegis of the Transitional Administrative Law from June 2004 until the election of the Iraqi Transitional Government in January 2005, and finally the October 2005 ratification and adoption of the Iraqi Constitution—but I think the formula in the text adequately foreshortens the story from normative a point of view. For an insightful and thought-provoking account of the Iraqi situation written in February 2004, see Andrew Arato, Sistani v. Bush: Constitutional Politics in Iraq, 11 Constellations (174–192) (2004).

14. There are surely other common usages of the language of legitimacy that are directed at more individual governmental actions; e.g., at the processes and outcomes of trials, the decisions of appellate courts, the procedures and outcomes of legislative or administrative decisions, and so on. Rather than a survey of linguistic usage in this area, then, I am simply trying to delineate and focus upon one type of usage of “legitimacy” and its cognates when addressed to the most general assessment of the worth of a governmental system.

15. Ronald Dworkin, The Model of Rules, 35 U. Chi. L. Rev. (14–46) (1967).

16. Although Kant is the obvious inspiration for my conception of the logic of regulative ideals, and I believe that this account is consistent with Kant's usage of the notion, I do not here make any claims to present accurately Kant's view of regulative ideals. The same is true for my example of a “broadly Kantian” notion of moral autonomy: I do not aim to reproduce his views exactly. One particular difference deserves note, however: for Kant, regulative principles or ideals are to be strictly distinguished from constitutive ones, in that the former are standards guiding reason, whereas the latter are standards imminent in objects; see, e.g., Immanuel Kant, Critique of Pure Reason (Norman Kemp Smith trans., St. Martin's Press 1965) (1781), at 449–451, 514–518. I make no use of this contrast in my account of the logic of regulative ideals, even though the contrast is fundamental to Kant's use of the term “regulative.” For an insightful discussion of how regulative ideals function, see Thomas McCarthy, The Philosophy of the Limit and Its Other, 2 Constellations (175–188) (1995).

17. A clear illustration of these features of the logic of regulative ideas in Kant's work is the discussion of the practical idea of a “holy will” and its relation to human morality in Immanuel Kant, “Remark to the Corollary of 7. Fundamental Law of Practical Reason,” in Critique of Practical Reason (Lewis White Beck trans., Macmillan 1956) (1788).

18. The distinction between proceduralism and substantialism is defined more precisely at Zurn, supra note 1, at 76–80. It plays a large role throughout the book as well.

19. Barnett, taking his cue from Locke as filtered through Robert Nozick, refers to natural rights as inalienable and repeatedly catalogs them “as the rights of several property, freedom of contract, self-defense, first possession, and restitution”; Barnett, supra note 3, at 73. Barnett's argument for natural rights appears to be functional: if we want to live happy, prosperous lives in functioning and perduring societies, then the catalog of rights must be respected—see id. at 78–86. Barnett's particular functionalist justification of natural-rights substantialism is not of interest here, though one cannot help wondering both how rights justified merely hypothetically are supposed to gain the categorical force usually accorded to natural, inalienable rights and how all kinds of societies apparently keep functioning perfectly well even with massive violations of these same rights.

20. Id. at 44.

21. Barnett thus conceives of the justice of a constitutional system to be an instance of imperfect procedural justice, to use Rawls's nomenclature, where there is a procedure-independent test for justice (here, the catalog of natural rights), but there is no feasible way of structuring decision-making rules (here, the constitutional system of lawmaking and applying) that will ensure, in all cases, that a just outcome is achieved; see John Rawls, A Theory of Justice (1999), at 74–75.

22. Barnett, supra note 3, at 51.

23. The choice of criminal trials is not accidental; it is Rawls's example of imperfect procedural justice, where we have a substantive test for just outcomes but no feasible procedures for guaranteeing such outcomes. Using Barnett's natural-rights theory in concert with the threshold conception of legitimacy yields, in Rawls's terms, a unrealizable demand for perfect procedural justice.

24. One might object that perhaps Barnett could suitably weaken his legitimacy test so that only very extreme rights deprivations would count as delegitimating. Aside from the fact that this threatens to render almost all political systems legitimate despite the intentions of his rather stringent brand of libertarianism, it seems incompatible with the catalog of rights that Barnett considers natural and inalienable and with the fulsome content he often interprets them as having.

25. Consider, in addition, Barnett's endorsement of the USC as a basically legitimate lawmaking system. He bases this judgment largely on the existence of the Ninth Amendment (concerning constitutionally unenumerated rights retained by the people), which he interprets as guaranteeing the protection of natural rights. But the Ninth Amendment was ratified some three years after the USC itself. That means that the original, unamended Constitution was, strictly speaking, illegitimate by Barnett's account and so could not have furnished the just authority for the amendment procedure itself—at least if we accept a threshold conception of legitimacy.

26. The stages indicated here are used merely as examples; different variations have been proposed in different social contract theories.

27. My thinking about reflexivity is deeply indebted to the important use of the concept made in conceptualizing the threat of persistent social inequality to legitimate deliberative democracy by Kevin Olson. See, e.g., Kevin Olson, Reflexive Democracy: Political Equality and the Welfare State (2006).

28. John Rawls, Political Liberalism (1996), at xxiii–xxx, 36–38, 54–58.

29. See Michelman, Constitutional Authorship, supra note 4.

30. My reconstruction of Michelman's positive response differs substantially from that given in Ciaran Cronin, On the Possibility of a Democratic Constitutional Founding: Habermas and Michelman in Dialogue, 19 Ratio Juris (343–369) (2006).

31. Michelman, Brennan and Democracy, supra note 4, at 50–51.

32. A later paper loosens the thickness of the required consensus, demanding not that all citizens agree on the same fundamental principles but only that there is an overlapping consensus of individual citizens' positive assessments of the legitimacy of the given system of government and law. See Frank I. Michelman, Ida's Way: Constructing the Respect-Worthy Governmental System, 72 Fordham L. Rev. (345–365) (2003).

33. This strategy is structurally the same as Rawls's argument that an “overlapping consensus” on political fundamentals among irreconcilable comprehensive doctrines is possible and could develop over time out of mere political compromise through the workings of public reason in the context of free institutions: Rawls, Political Liberalism, supra note 28, at 158–168.

34. In another context, Michelman explicitly endorses regulativism with respect to the possibility of giving a democratic legitimation for human rights: “No logic excludes the possibility of there being something that is morally necessary to do, which we cannot ever finally know or show that we have done. We call such thing a regulative ideal.” Frank I. Michelman, Human Rights and the Limits of Constitutional Theory, 13 Ratio Juris 76 (63–76) (2000).

35. Michelman, Brennan and Democracy, supra note 4, at 60. The argument is tailored to promoting a particular understanding of the function of constitutional review and its institutionalization in an electorally unaccountable judiciary. I critically evaluate Michelman's theory of judicial review but endorse his notion of openness as criterial for the requisite sensitivity of institutions of constitutional review in Zurn, supra note 1, at 163–220 and 271–272. Here I am not concerned with which actual institutional arrangements actually carry the function of applying constitutional law to concrete controversies.

36. Frank I. Michelman, The Problem of Constitutional Interpretive Disagreement: Can “Discourses of Application” Help?, in Habermas and Pragmatism 114 (Mitchell Aboulafia, Myra Bookman & Catherine Kemp eds., 2002).

37. Id.

38. Brown v. Board of Education, 347 U.S. 483 (1954). Brown now functions as what I call elsewhere a “firmament” case, acknowledged in the legal community as unimpeachably correct (Zurn, supra note 1, at 11) even though it was not always treated so; see, e.g., the attack on it in Herbert Wechsler, Toward Neutral Principles of Constitutional Law, 73 Harv. L. Rev. (1–35) (1959).

39. Parents Involved in Community Schools v. Seattle School District No. 1, 127 S. Ct. 2738 (2007).

40. Michelman, Brennan and Democracy, supra note 4, at 8.

41. See Jeremy Waldron, Law and Disagreement (1999), at 11–12, 55, 73–75, 101–103, 112–113.

42. Id. at 111 n.62. The argument is fully laid out in ch. 8 at 164–187.

43. Michelman likewise claims, when reflecting on reasonable interpretive pluralism, that he is not denying that there is moral truth about correct applications of legal principles, only insisting that such truths are “politically unavailable” to us under modern pluralism; Michelman, Human Rights, supra note 34, at 71. Habermas points out that this puts Waldron and Michelman in the same boat: they neither endorse moral skepticism nor reject moral cognitivism. The problem is essentially about our epistemic limits concerning moral content in collective action situations, not a problem of the basic metaethical status of that content. Jürgen Habermas, On Law and Disagreement: Some Comments on “Interpretive Pluralism,” 16 Ratio Juris (187–194) (2003).

44. This is, of course, another way of stating Rawls's point about the “burdens of judgment.”

45. This strategy is most clearly articulated in response to Michelman's challenge in Jürgen Habermas, Constitutional Democracy—a Paradoxical Union of Contradictory Principles?, in Time of Transitions (113–128) (Ciaran Cronin & Max Pensky eds., 2006) (originally published 2001 in Political Theory); and to the challenge of legal disagreement posed by Michelman and Waldron in Habermas, On Law and Disagreement, supra note 43. Cronin, On the Possibility, supra note 30, provides an insightful reconstruction of the debate between Michelman and Habermas.

46. I am not directly interested here in Habermas's specific conception of the legitimacy requirements for constitutional democracy nor in the general outlines of his account of the processes, institutions, and laws appropriate to its instantiation. For fuller treatments of these topics, see Zurn, supra note 1, at 227–243.

47. Habermas, Constitutional Democracy, supra note 45, at 122.

48. Jürgen Habermas, Between Facts and Norms: Contributions to a Discourse Theory of Law and Democracy (William Rehg trans., 1996), at 39, 110.

49. Habermas's conception of progressive learning arises from a combination of the Hegelian idea of dialectical progression and the specific stage-sequential logic of learning evident in developmentalist psychology as in Piaget and Kohlberg. A clear exposition and defense of this conception of learning as Habermas applies it to underwrite some of his most ambitious claims concerning social evolution can be found in David S. Owen, Between Reason and History: Habermas and the Idea of Progress (2002).

50. Thus worries expressed by “agonistic” theorists that Habermas's political theory prematurely forecloses the “democratic openness” that agonists take to be constitutive of healthy politics seem to me to be misplaced. For two different versions of the complaint that Habermas proposes a static, fully reconciled, and prematurely “closed” theory of constitutional democracy, see Bonnie Honig, Between Decision and Deliberation: Political Paradox in Democratic Theory, 101 Am. Pol. Sci. Rev. (1–17) (2007); and Chantal Mouffe, The Democratic Paradox (2000). Cronin, On the Possibility, supra note 30, at 367–368, makes a similar point in defending Habermas.

51. Although some formulations in Michelman indicate that historical trends are important to legitimacy judgments, in general he tends to emphasize their presentist character to underline the idea that contemporary citizens must judge their current governmental system as worthy of respect: see especially his summary of his views in Frank I. Michelman, Reply to Ming-Sung Kuo, 7 Int'l J. Const. L. (715–730) (2009), at 723–730. Thus, while Michelman's view is clearly a form of regulativism, I do not think it counts as developmentalist.

52. I believe this goes a long way to assuaging Olson's and Honig's worries about the epistemic viability of predictions of progress, since on my account such predictions are not purely future-oriented. Cf. Kevin Olson, Paradoxes of Constitutional Democracy, 51 Am. J. Pol. Sci. (330–343) (2007), at 333–334; and Bonnie Honig, Dead Rights, Live Futures: A Reply to Habermas's “Constitutional Democracy,” 29 Pol. Theory (792–805) (2001). Olson's introduction of the notion of path-dependent constitutional development as an important component of the logic of legitimacy plays the same role in his theory, if I am not mistaken, as does my insistence that legitimacy assessments attend carefully to the past and present as well as the future.

53. Habermas, Constitutional Democracy, supra note 45, at 123 (almost all emphases added).

54. My reflections here are indebted to Alessandro Ferrara, Of Boats and Principles: Reflections on Habermas's “Constitutional Democracy,” 29 Pol. Theory (782–791) (2001); and Honig, Dead Rights, supra note 52, even as I approach the problems differently. I am not frontally concerned here with a question that Habermas briefly considers in Constitutional Democracy, supra note 45, and that preoccupies both Ferrara and Honig: namely, how can it be legitimate for past generations' agreements to bind the present demos?

55. Throughout his political theory and political writings, Habermas insists on a distinction between, on the one hand, the universal project of constitutional democracy, with its abstract system of rights categories and functionally based separation of powers, and, on the other, the acceptably different realizations of constitutional democracy under given contingent cultural, social, and material conditions in distinct nation-states. For instance, with regard to rights, he claims “we can understand the catalogs of human and civil rights found in our historic constitutions as context-dependent readings of the same system of rights”; Habermas, Between Facts, supra note 48, at 128. The same goes for the ideals of the separation of powers, the rule of law, rational adjudication, a free and open public sphere, and so on; in each case, the same abstract ideals are realized in the specific traditions of a given polity, where the bone and marrow of such traditions consists in conflicts concerning the proper interpretation and implementation of those very ideals.

56. One might cite here the eventual inclusion of slaves through U.S. Const. amends. XIII, XIV, XV (1865–1870), women through amend. XIX (1920), and nonpropertied males through amend. XXIV (1964).

57. Or if we do ask this, it is only in the deflated sense that Michelman specifies: “constitutional framers can be our framers—their history can be our history . . . —only because and insofar as they, in our eyes now, were already on what we judge to be the track of true constitutional reason.” Michelman, Constitutional Authorship, supra note 4, at 81.

58. At various points, though inconsistently, Habermas seems to endorse this more general conception of what remains the same across the generations of citizens in a particular political system: “What binds them together is the performative meaning of the very practice of constitution making. The ‘purpose’ of this practice is supposed to be understood in the same way as the founding, development, and preservation of a voluntary association of free and equal citizens governing themselves by means of modern, i.e. positive and legitimate, law.” Jürgen Habermas, Postscript: Some Concluding Remarks, in Habermas and Pragmatism (223–233) (Mitchell Aboulafia, Myra Bookman & Catherine Kemp eds., 2002), at 225.

59. Cronin, On the Possibility, supra note 30, explicitly endorses the opposite of what I am recommending: a contextualized and historicized version of the identity between founders and successors that is inherently particularistic and finds expression in a determinate collective political identity that is constructed through imaginative, retrospective projection. While Cronin argues on the basis of salutary sociological effects in fostering shared identity among citizens, this is a different question from whether a political system is normatively legitimate.

60. Recall that according to Habermas's more concretistic conception, “All the later generations have the task of actualizing the still-untapped normative substance of the system of rights laid down in the original document of the constitution.” Habermas, Constitutional Democracy, supra note 45, at 122. But to put it bluntly, what if “we” happen to have an unlucky heritage; what if that original document simply does not contain all that constitutional democracy requires, even in nuce? If we cannot plausibly trace a constitutional innovation back to the normative content of the original document, is it then to be abandoned even if it is a demonstrable improvement from the general point of view of the project of realizing the ideals of constitutional democracy?

61. Dred Scott v. Sandford, 60 U.S. 393 (1857).

62. Could a judgment of legitimacy at one point in time be warranted by the evidence available at that time—perhaps after the 1996 Russian presidential election one might be warranted in thinking the developmental path towards increasing democratization propitious and likely irreversible—and later be shown to have been wrong by subsequent developments—perhaps after the 2007 legislative and 2008 presidential elections showed distinct signs of democratization regress? Yes, I think it makes sense to assert that a legitimacy judgment could have been warranted at time 1 given the evidence available at time 1, and at the same time to assert that, given the evidence available at time 2, it is clear in retrospect that the judgment of time 1 was not correct. Thus legitimacy judgments are fallibilistic in a similar way that scientific propositions are: Priestly may have been both warranted and incorrect in his assertions about the existence of phlogiston in 1774, as shown by Lavoisier's results publicized in 1775. This also connects with my point above about the irrelevance of the beliefs of the founders about legitimacy: their assessments at time 1 of legitimacy may have been warranted, but they are not dispositive. In the end, only the evidence—about the actual past, present, and likely future performance of the political system in terms of the stipulated ideals of constitutional democracy—is dispositive.

63. Here I am in full agreement with Michelman, much of whose “writing denies the cogency—once examined—of authorship as a ground of constitutional legitimacy,” Michelman, Reply to Ming-Sung Kuo, supra note 51, at 719.

64. I have in mind here both commitment-focused theories (e.g., Jon Elster's and Steven Holmes's accounts of constitutional precommitment or Jed Rubenfeld's commitmentarian account of constitutional binding across generations) and ratification-focused theories (such as Alexander Hamilton's theory of the constitution as the ratified will of the people or Bruce Ackerman's more complex theory of constitutional change through sustained popular deliberation and endorsement). It is, however, well beyond the scope of this paper to make good on such claims.

65. One obvious candidate wrong is the impropriety of paternalism, now writ large into international military relations; see, e.g., the lengths gone to in order to show that such paternalistic interventions admit of at least a possibility in principle of normative acceptability in Arthur Isak Applbaum, Forcing a People to Be Free, 35 Phil. & Pub. Aff. (359–400) (2007).