I. Liberalism, Epistemic Abstinence, and the Paradoxes of Evaluative Neutrality

A right of freedom of expression is naturally associated with the political phi­losophy of liberalism. Human rights documents list freedom of expression alongside freedom of religion, freedom of association, freedom of marriage, and other rights strongly associated with liberalism.^[289]

Liberalism is, however, deeply paradoxical at its core.

Freedom of religion and freedom of association likewise require evaluative neutrality. Freedom to practice only the “correct” religion is not religious free­dom. Nor is freedom to associate only along lines the government approves freedom of association. Freedom of religion must mean freedom to practice re­ligions that are “wrong” about religious truths. And freedom of association must mean freedom to associate with the “wrong” people for the “wrong” purposes.

Yet, here is the problem. Any philosophical account of political morality will, perforce, take a stand on what is true, right, and valuable and what is not. It will and must be “partisan” in favor of its own conclusions. Thus, it must regard as error and possibly malign those ideas that it rejects.

Liberalism in any of its renditions is no different. If liberalism is the correct political morality, all positions inconsistent with its tenets are incorrect.

A. Liberalism and Illiberal Religions

Liberalism as apolitical philosophy is traceable to the European wars of religion. Unsurprisingly, therefore, freedom of religious practice is central to liberalism. Yet freedom of religious practice demands a governmental stance of evaluative neutrality that, given both extant and conceivable religious tenets, generates paradoxes.

Suppose, for example, that a religious group believes in the necessity of theocratic rather than democratic rule. Liberalism cannot allow that group to actualize what its religion demands. It cannot be “neutral” regarding theocracy. It cannot claim to know that liberalism is the correct political philosophy but at the same time profess “epistemic abstinence” regarding whether this religious group’s tenets are correct. If a liberal form of government is most consistent with the dictates of reason, then theocracy and the precepts of theocratic religions are not.

Theocratic beliefs are not the only religious beliefs the realization of which in practice liberalism cannot allow. Religions take positions on all kinds of issues on which liberalism is partisan. A religion may dictate separation of the races or rigid gender roles. It may demand the repression of sexual speech and the outlawing of abortion. The list of possible “illiberal” positions that religions may hold is endless.

Some liberals believe, contrary to what has just been said, that a liberal government can reject illiberal religious practices without at the same time taking a position on the truth of the religious tenets that underlie them. On this view, a liberal government can say to the frustrated adherents of illiberal religions, “Your religious views may be correct or incorrect. The government - the liberal majority - is not in a position to say. What the government is in a position to say is that your views are ‘reasonably rejectable’ by those outside your religion on whom you wish to impose the practices your views support, and that for that reason they may not be practiced.

This type of response is unavailing. It requires that there be beliefs that are, on the one hand, correct, but are, on the other hand, “reasonably rejectable.” Now there is one sense in which correct beliefs can be reasonably rejectable, but it is not a sense that is helpful to liberal neutrality. There are many propositions about the world that, given the evidence before me, I should reasonably conclude to be false. Yet some of those propositions will turn out to be true. If I am unfamiliar with Einsteinian physics and the evidence that supports it, I might reasonably reject the proposition that nothing can exceed the speed of light. After all, in the experience of ordinary people, the idea of a finite limit to speed - or to space and time - is counterintuitive. Similarly, one who knew only that O. J. Simpson was a wealthy ex-football star, announcer, and actor, and who did not watch or read about his trial, could reasonably reject the proposition that O. J. killed Nicole.

Obviously, this commonplace sense of reasonable rejectability does not help the case for liberal neutrality because it cannot account for the distinctiveness of religious beliefs within liberalism. Any belief could be reasonably rejectable by someone in this sense. Liberalism itself could be reasonably rejectable.

The only other sense of reasonably rejectable that the liberal might invoke is one that assumes that the “truth” of religious propositions rests on an epistemol­ogy that is different in kind from the epistemology underlying claims of truth for ordinary factual and moral propositions. Yet for the typical religious adherent, her religious beliefs are continuous with her beliefs about other matters, and all her beliefs are mutually reinforcing to the extent they cohere. There is no separate sphere of religious beliefs that is hermetically sealed off from nonreli­gious beliefs and that rests on a distinctively religious epistemology.

As I said, the above argument should be uncontroversial. If liberalism or the set of propositions to which it can be reduced is true, then religious tenets that conflict with the propositions of liberalism are false. Liberalism is in this respect just one more sectarian position, and those who decry the “religion” of secularism that they find characteristic of modern liberal societies have surely affixed the correct label to their concern, even if what disturbs them is something they should instead welcome. If, however, liberalism is a sectarian position, then is it in fact better justified than its competitors?

The problem arises with respect to religious and other comprehensive the­ories of the Good that are nonindividualistic or other-regarding in nature. Ac­cording to such theories, an individual’s good is inextricably dependent upon the good of others, such as the salvation of their souls or their adherence to norms regulating behavior not directly harmful to others. Because an individ­ual’s good is dependent on others in this way, the individual can be just as harmed by others’ failures to attend to their souls or their virtue as by direct physical assaults, theft, deception, or sensory offense. Andjust as an individual may legitimately employ the coercive apparatus of the state to protect against physical harm, theft, and offense, so too on these views of the Good may the individual employ that coercive apparatus to save souls, enforce virtue, and so forth.

The liberal tries to refute these views. The liberal, however, does not face these views head on and show that the salvation of souls is meaningless or incapable of being accomplished through force, or that no one is harmed in any way other than through physical assault, theft, deception, or offense.

To assess the justifiability of liberalism as it relates to religious views, let me move from the abstract treatment of liberalism’s logical implications and look at a concrete case. Ann has been brought up in a religion within the Judeo- Christian family of religions. She believes in the existence of a God, whom she pictures vaguely as a person, and in the moral authority of the leaders of her church. She perhaps also believes in the divinity of the biblical Christ. Although Ann first came to these beliefs by accepting the teachings of her parents, she now holds these beliefs based on what she views as independent and superior reasons. For example, she believes most of the accounts of God’s and Christ’s miraculous deeds in the Old and New Testaments are credible. She holds these beliefs based on the number of witnesses, their independently tested reliability, and the number of intelligent people who accept these accounts as true. The miracles for her establish the authority of the moral teachings in the Bible, which are in turn independently verified by their consonance with her own feelings about moral issues and with the teachings of people she finds admirable. Moreover, these mutually reinforcing beliefs for Ann make sense of human existence in general and her own life in particular.

The purpose of this portrait of Ann is to show how continuous Ann’s religious epistemology is with her epistemology in general, and how all of her beliefs, evidentiary criteria, and methods of reasoning cohere.^[291] Ann’s religious beliefs are supported in exactly the same way as are her beliefs that George Washington was the first President of the United States, that Nairobi is the capital of Kenya, that Barry Bonds hit seventy-three home runs, and that nothing can exceed the speed of light.

Further, no reason exists for not portraying Ann as a person who entertains the possibility that many of her beliefs, including her religious beliefs, might be wrong (though, of course, she does not currently believe they are wrong, or else they would not be her beliefs). Ann may worry about the veracity of biblical sources, or about the coexistence of a benign, all-powerful God with evil. She may wonder why the truth of Christianity was revealed at a particular time and place in a manner that could not fail to make its acceptance local, or whether, had she been reared in Tehran, she would believe in Islam even if she were given all the above evidence of Christianity’s truth.

I have written thus far about Ann’s beliefs and the reasons she has for holding them. I have not mentioned faith, much less contrasted it with reason. For one thing, I am not sure how to make an epistemological distinction between faith and reason.^[293] Ann has “faith” in modern physics, in her personal physician, and in her husband’s fidelity. Of course, her faith in these is grounded in reasons she has or believes she has. Likewise, Ann’s religious faith is not a faith in just anything. She does not believe she could have faith in whatever she wants to believe, religiously or otherwise. Her “faith” is, for her, a product of her reason.

Ann holds various views on public policy that rest in part on her religious beliefs. She believes that fetuses are morally protectable and have a right to life that trumps women’s rights regarding reproduction. She believes that animal life is worthy of protection, and that meat-eating and vivisection should be illegal. She believes that pornography is blasphemous because it degrades sexuality that God intends be ennobled; she, therefore, wants pornography legally banned. She believes that public money should be used to support the private schools that teach her religious beliefs.

Now, there are various versions of liberalism, and not all of them would find all of Ann’s public policy recommendations illicit. Versions that I shall label “purist” - the versions endorsed by Bruce Ackerman,^{[294] [295] Ronald Dworkin,11 Charles Larmore,[296] Thomas Nagel,[297] and John Rawls[298] - would reject all four of Ann’s policy recommendations if they can be “justified” only by Ann’s reli­gious beliefs. “Impure” liberalisms, such as that of Kent Greenawalt,[299] would reject only the latter two, because “secular” reasoning can reach no conclusions regarding issues such as abortion and animal rights. Both pure and impure ver­sions of liberalism are subject to my critique.}

1. impure liberalism. The impure liberal would allow religious argu­ments to shape public policy whenever “secular” arguments cannot resolve the issues at stake. Greenawalt argues that two of Ann’s concerns, abortion and animal rights, raise issues that secular reasoning cannot resolve. Thus, liberal­ism must permit religious views to shape the resolution of these issues.¹⁶

Impure liberals such as Greenawalt seriously underestimate the degree to which religion’s nose is inside the liberal’s tent on their account of impure liberalism.¹⁷ Greenawalt at times appears to believe that only a modest number of issues lying at the periphery of liberalism’s agenda will be incapable of secular resolution, and that, therefore, religion’s influence on public policy will be, in principle at least, limited and well-defined.¹⁸ Greenawalt, though, does recognize that core issues, such as the just distribution of wealth, seem to involve not only “secular” reasons that are “neutral” regarding theories of the Good, but also reasons that are partisan regarding the Good. Neutral reasons are incapable of adjudicating whether resources, objective welfare, subjective welfare, or something else is the object of liberal distributive policies.¹⁹ Moreover, among partisan reasons, Greenawalt does not distinguish between the secular and the religious, much less favor the former over the latter.²⁰

¹⁶ See KentR. Greenawalt, “Religious Convictions and Political Choice: Some Further Thoughts,” 39 DePaulL. Rev. 1019, 1029-33, 1042(1990).

¹⁷ See Michael J. Perry, Love and Power 19-20 (1991). Perry notes that the line between secular and religious will be a hard one for Greenawalt to draw. Id. at 19. If religious and secular epistemologies are but a single epistemology, then it follows that the line between “secular” and “religious” will be arbitrary and at best a matter of convention.

¹⁸ Thus, to illustrate his thesis that religious arguments may sometimes be resorted to in a liberal democracy to determine public policy, Greenawalt points to issues of moral status that he labels as “borderline” issues - for example, the issues of abortion and animal rights. See Greenawalt, supra note 7, chs. 6, 7.

¹⁹ Id. at 173-87.

²⁰ See Greenawalt, supra note 7, at 191-92. For an attack on Greenawalt’s impure liberalism from a more purist direction, one arguing that “secular” beliefs are capable of resolving all of Greenawalt’s hard issues, see Peter S. Wenz, Abortion Rights as Religious Freedom 112-13, 119-31, 135-9 (1991). Wenz distinguishes secular from religious beliefs as follows:

Religious beliefs are those that cannot be established by appeal merely to secular premises and methodologies. Secular premises and methodologies include what passes for common-sense knowledge in our society (e.g., fire burns people, punishment deters crime), the scientific be­liefs that underlie our technology (e.g., electrons and bacteria exist), the methodology accepted in our society for the generation of scientific and technological knowledge (e.g., observation, microscopes, carbon dating, and mathematical calculations can be useful), and the values con­sidered essential to society (e.g., peaceful coexistence among its members, limits on assault and murder) or essential to our type of society (e.g., individual liberty, private property, plural­ism, and hard work). More generally, secular premises are drawn from secular beliefs. I term “secular beliefs” all those agreements of belief, thought, and practice that are the basis of the cooperation and mutual understanding needed among people to maintain and perpetuate our society.

Religious beliefs are those that cannot be supported cogently with arguments or demonstrations whose premises include only secular beliefs. What is more, for First Amendment purposes, all religious matters are religious because of their relationship to religious beliefs. For example, “creation science” is religious because it results from an interpretation of the Bible. The Bible is a religious text because of its relationship to belief in the existence of God, whose actions in history it is supposed to record. Belief in the existence of God cannot be supported cogently by the use of common sense, science, technology, or accepted scientific methodologies, so it is a religious belief.

Indeed, the impure liberal’s notion of what issues are underdetermined by secular arguments, and thus the legitimate subjects of religious ones, threatens to bring the entire liberal tent crashing down. The impure liberal seems to take controversy among self-proclaimed liberals as a sign that secular reasoning has gone as far as it can go and must be supplemented by alternative brands of reasoning, such as religious reasoning.²¹ Abortion, animal rights, and what is to be distributed are highly controversial among liberals, none of whom has committed an error of syllogistic inference or has relied on incorrect accounts of empirical facts. Therefore, according to the impure liberal, these are issues that can only be resolved by recourse to sectarian views. If, however, the impure liberal is correct about this, then the dam of principle holding back thoroughgo­ing religious influence on policy making is completely breached. For on almost every policy issue - the meaning and scope of freedom of speech, economic freedom, privacy, free exercise of religion, equal treatment, community self­determination, the justification of punishment, procedural rights, and so on - self-proclaimed liberals are deeply divided. Although one cannot be a “liberal” if one rejects all conceptions of free speech, privacy, and so forth, no liberal party line exists on any aspect of liberalism. Impure liberalism as a theory is too skeptical about the power of secular reasoning to resolve issues that are con­troverted among liberals to provide a convincing argument against wholesale admission of religious arguments in policy making.²²

Finally, and perhaps most importantly, the impure liberal’s breach of the dam barring religious arguments fails to take into account the interconnectedness of beliefs. If religion can “get it right” about, say, abortion, why can it not “get it right” about such things as, say, sin, or other matters that the impure liberal would make the exclusive province of the secular? The grounds that support accepting a religion’s view on one issue are quite likely to support accepting its

I call this the epistemological definition of religion because religion is defined in terms of how we gain knowledge and how we support our claims to knowledge.

Id.,at 112-13.

Wenz’s distinction between “secular” and “religious” is unconvincing, both as to matters of material fact and as to matters of value. Both common-sense and scientific factual knowledge, as well as common moral beliefs, rest ultimately on undemonstrable metaphysical views. As Ronald Dworkin puts it, “[Wenz’s] test is not acceptable, because government must make and impose decisions on a wide variety of moral issues about which people disagree profoundly, and which cannot be decided on empirical grounds or by appeal to any convictions shared by everyone or by methods that are in any other way “integral” to any collective way of life.” Ronald M. Dworkin, “Unenumerated Rights: Whether and How Roe Should Be Overruled,” 59 U. Chi. L Rev. 381, 421-2 n. 60 (1992).

²¹ See Raz, supra note 1, at 42.

²² See generally id. at 23, 42-3. See also Thomas Spragens, Jr., The Irony of Liberal Reason (1981). Spragens argues that liberalism has been undermined by its celebration of technological conceptions of reason and its value noncognitivism. Enlightenment-based animus to religious authority may have resulted in placing technocratic “reason” in an exalted status; “reason’s” failure to resolve value disputes may have left liberalism vulnerable to attack. views on other issues. Put differently, one cannot buy religious views at retail, specific issue by specific issue; they are available only at wholesale.

2. pure liberalism. The pure liberals - Ackerman, Dworkin, Larmore, Nagel, and Rawls - believe that secular reasons should resolve all issues in the public sphere, including how the line defining that sphere is to be drawn. The issues of abortion and animal rights can and must be resolved without recourse to religious views. The pure liberal would find all of Ann’s policy recommendations unjustifiable to the extent that they are based on the arguments Ann provides for them.

What grounds does the liberal have for finding Ann’s policy recommen­dations unjustifiable except for competing religious beliefs or beliefs that all religious beliefs are untrue, or at least unjustifiable? The liberal’s rejection of religion-based policies suggests some sort of epistemological divide or discon­tinuity between what we can claim justifiably to know secularly, so to speak, and what we can claim justifiably to know religiously, the latter being an infe­rior form of knowledge for purposes of public policy, though perhaps not for other purposes. No such epistemological divide exists, however, that will give liberalism its justification for rejecting the legitimacy of Ann’s proposals.

It is easy to dispose of one possible basis for rejecting Ann’s policies, namely, that the religious beliefs she relies on are not empirical. Without entering the epistemological thicket surrounding the status of empirical knowledge and its connection to theories that ultimately rest, in part, on nonempirical norms, it is obvious that liberalism cannot avail itself of any empirical/nonempirical divide. For the propositions of liberalism are not empirical, but metaphysical and normative.

More likely, the liberal will want to argue that religious beliefs are less subject to reasoned assessment than moral beliefs, such as those of liberalism. If this ar­gument is not just another way of rejecting the justifiability of religious beliefs altogether, which itself would require that other religious beliefs - skeptical ones such as agnosticism or atheism - are justifiable, then the argument is implausible. Moral reasoning rests either on nonempirical premises or on in­ferences that are not matters of logical entailment. This is surely the case with liberalism. Reason in the narrow sense that might be employed to discredit religious arguments equally discredits normative arguments.^[300] And reason in the broader, reflective equilibrium sense that might make one moral view more reasonable than another, seems just as applicable to religious views.^[301]

Liberals attempt to disenfranchise religious arguments on other grounds. Some, such as Ackerman and Nagel, believe that reliance on religious arguments, such as Ann’s, to establish public policy is inappropriately non­neutral or nonimpartial and therefore “unfair” to those who do not share Ann’s religious views.^[302] This tack is question-begging at two levels. First, to repeat an earlier point, the demonstration that the Ackerman/Nagel conception of fair­ness trumps religious grounds would require arguments for that conception and against the competing religious grounds that liberals such as Ackerman and Nagel fail to make. How could they make them, though, without doing theology as well as moral philosophy?

Second, Ann would dispute the claim that acting on her religious views in effecting public policy is in any way unfair to those who reject her religious views. Fairness, neutrality, and impartiality are concepts, like equality, that are empty vessels for substantive norms. If abortion is the killing of a being that should be legally protected, then in what sense is it unfair to those who reject this truth to outlaw abortion? If fairness requires placing legal duties on persons only if those duties are truly morally required, then from Ann’s perspective, passing laws against abortion is not unfair. Of course, from the perspective of those who disagree with Ann, such a law is unfair, but its unfairness lies in its error and not its source. This conflict of views about fairness is, therefore, inevitable. Unless Ackerman/Nagel liberals advocate, in the name of fairness, a unanimity requirement for public policy - and why is that “fair” (and what public policies are the default policies in case no unanimity exists, anyway?) - then public policies are fair if they are morally correct. Moral correctness may truly be the function of some forms of equal treatment or equal regard, but if Ann’s views are correct, then it follows from their correctness that they already incorporate the correct conception of equal regard.^[303]

The Ackerman/Nagel liberal is arguing that Ann’s views are incorrect, at least the component of them that dictates that they be coercively imposed. The liberal is using the unfairness of the coercive imposition as grounds for claiming Ann’s views are incorrect. I have shown that some senses of unfairness fail to support the liberal’s argument.

Nagel, for example, argues that coercive imposition of norms is unfair (in a morally overriding sense) if such norms are reasonably rejectable. Norms are reasonably rejectable in this disqualifying sense if they rest on grounds that are not publicly accessible. Religious grounds are not publicly accessible to those outside the religion. Therefore, it is morally true - contrary to the teachings of Ann’s religion - that it is morally wrong for Ann to seek to impose coercively her views on abortion, support of religious schools, and so on.^[304]

Let me skip the question of whether Nagel’s reasonable rejection conception of fairness fails its own test, that is, whether someone like Ann could reasonably reject it, or find the arguments for it inaccessible given her religious beliefs. Assume Nagel’s test is not self-undermining in this way. Does it render illegit­imate Ann’s policy recommendations if we assume the absence of unanimity, and if we prescind the question whether, as matters of substantive morality and religious truth, abortion, pornography, and so forth should be condemned?

Two senses in which grounds and arguments might be reasonably rejectable or publicly inaccessible exist. One sense is too strong to help Nagel, the other too weak. In the strong sense, Ann’s beliefs are reasonably rejectable and publicly inaccessible if we can understand why others do not share those beliefs. Ann has had a unique life, with unique experiences. She, but not others, witnessed certain events from a specific point of view. She, but not others, was taught certain things by particular people. Her particular beliefs - on all sorts of matters - are as reasonably rejectable by and inaccessible to others as, say, our beliefs about air travel, television, and the historical Christ are reasonably rejectable by and inaccessible to Stone Age tribes in New Guinea who have never seen an airplane, a television show, or a Christian.

This sense of reasonable rejectability/public inaccessibility is obviously too strong to serve Nagel’s purpose, for it rules out imposition of any norms whose rejection is understandable in this way. We can surely understand how a “rea­sonable” slave owner in the Antebellum South could reject imposition of a norm of racial equality. From his perspective - given the web of his other beliefs - blacks were inferior human beings. Nonetheless, Nagel would surely not think it illegitimate to eradicate slavery coercively in the face of opposition by such “reasonable” slave owners.^[305]

The other sense of reasonable rejectability/public inaccessibility is too weak to serve Nagel’s purpose. That sense would exclude the imposition of only those norms that can be reasonably rejected by someone after being privy to all the experiences of those who accept the norms. Clearly, Ann will not accept rejection of her norms as reasonable in this sense; her acceptance means that she thinks acceptance is reasonable and rejection unreasonable. From the point of view of the one whose norms they are, no accepted norms can be reasonably rejected.^[306]

One gets the same results if one substitutes some sort of dialogic test for the reasonable rejectability/public accessibility test, that is, if one allows co­ercive imposition of only those norms accepted through dialogue among all the affected parties. Dialogic tests for the legitimacy of coercive imposition of norms come in two types: those requiring actual dialogue among the affected parties, and those requiring only a hypothetical, idealized dialogue. An actual dialogue test is, in effect, a requirement of unanimity. Requiring unanimous acceptance of coercive norms, however, cannot be morally justified for two rea­sons. First, the reasons for failure of unanimity can include antisocial motives, ignorance of factual matters, illogic, deception, misunderstanding, immatu­rity, fatigue, linguistic incompetence, and a myriad of other factors that should discredit the objector.³⁰ (No actual dialogue ever occurs in an “ideal speech

Suppose someone in control of the coercive apparatus of government encounters aliens from another planet who warn him, based on research that is beyond human capacities, that the Earth faces annihilation if nations do not disarm. (See The Day the Earth Stood Still, 20^th Century Fox, Farmington Hills, Mich., 1951.) Would it be “illiberal” in Nagel's sense for him to rely on this knowledge, given that others are universally skeptical when he tells them about the encounter? His claims are reasonably rejectable in the strong sense I have identified, though not in the weak sense. Which sense should govern his actions, and does some third sense exist that is intermediate between the two I've identified? Is his encounter with space aliens any less accessible to others than his knowledge of any nonreplicable historical event, such as that a spider crawled upon his bathroom mirror at 2:15 p.m. Tuesday? If not, may he rely upon his knowledge of such historical events, and what if they are “miraculous”?

Note that impure liberals like Greenawalt, while they would allow Ann to consult her religious views on some issues, nonetheless require that she be able to distinguish those views from purely secular ones. For Greenawalt, the latter are to govern decision making on most issues and are to govern public argument on all issues. See Greenawalt, supra note 7, at 215-28. How is Ann, though, to distinguish her religious views from her secular ones? No sharp divide separating one set of her beliefs from another exists, and why would it matter whether she is influenced by Aristotle or by Christ? And where does Aquinas fit? It looks as if “secular” can only be distinguished from “religious” using criteria like Nagel's “publicly accessible reasons.” However, I have argued these criteria do not work because they are either too strong or too weak. ³⁰ Actually, although all of these factors may prevent achievement of unanimity, some of them will vitiate the results of unanimous agreement even if unanimity is achieved.

The problem, pointed out by Walzer, is how to constrain actual dialogue to obviate these problems without at the same time leaving the parties to the dialogue with nothing left to discuss:

Forthe rules of engagement are designed to ensure that the speakers are free and equal,to liberate them from domination, subordination, servility, fear, and deference. Otherwise, it is said, we could not respect their arguments and decisions. But once rules of this sort have been laid out, the speakers are left with few substantive issues to argue and decide about. Social structure, political arrangements, distributive standards are pretty much given; there is room only for local adjustments.

Michael Walzer, “Moral Minimalism,” in W. R. & A. Spadafora, eds., From the Twilight of Probability 3,11 (1992). Seealso Herlinde Pauer-Studer, ed., Constructions ofPractical Reason 129-30 (2003); Richard Rorty, “The Advantages of Moral Diversity,” 9 Soc. Phil. & Pol^,y 38,53 (1992) (“[T]he conditions that are necessary to assure just and fair debate in the public sphere appear to presuppose the happy outcome of just those debates.”) Put differently, we have to know how the dialogue should come out substantively in order to know how to constrain it procedurally.

situation.”^[307]) Second, and more fundamentally, an actual dialogue/unanimity requirement necessitates some sort of default position, coercively imposed (as well as an allocation of resources to maintain the actual dialogue, collected from someone), to govern the parties unless and until they unanimously agree on a new set of coercively imposed norms.^[308] Yet, no set of default norms has itself passed the test of unanimity in actual dialogue. Requiring a dialogue about the default position, therefore, leads to an infinite regress of default positions and dialogues.^[309]

For these reasons, most dialogic tests for justifying coercive imposition of norms rely on a hypothesized and idealized dialogue among parties who are rational, well-motivated, and dialogically constrained in certain ways. Because such dialogues are hypothetical, they can be replicated monologically without devoting resources to bringing parties together, ensuring actual communication, and sustaining a dialogue over time. Moreover, the results of the monological exercise can be known virtually instantaneously, thus obviating the need to have, and thus to justify, a coercively imposed default position. The key to imagining the idealized hypothetical dialogue lies in what types of knowledge, reasoning ability, emotional responses, motivations, and so on one envisions the parties to the dialogue possessing. If one allows for differences among the parties, then one faces the possibility that they would not agree to any particular set of norms. A dialogic test that allows for differences is functionally like the strong reasonable rejectability/public accessibility test. One - and Ann - can understand why those who have had different experiences from Ann’s would disagree with her norms were they to engage in dialogue. One can also understand, though, why slave owners would disagree with norms of racial equality.^[310]

If one assumes complete interpersonal transparency among those in the hypothesized dialogue, then Ann’s experiences are everyone’s. This type of hy­pothesized dialogue is no different from the monologue of the solitary thinker and the weak sense of reasonable rejectability/public accessibility. If Ann con­ducts this imaginary dialogue, she will conclude that her norms would be unan­imously accepted, for everyone would share her experiences and thus her (from her view reasonable) beliefs.³⁵

Nagel wants to deny the epistemological continuity of religious beliefs and those beliefs that should govern public policy. He argues that there are some beliefs, most notably religious beliefs, that are reasonable for someone to accept who at the same time recognizes that rejection of those beliefs by others is also reasonable.³⁶ As I pointed out above, however, in saying that it is reasonable for others to reject what one believes, one is only saying that one can understand how others, possessed of reason, have fallen into error. One does not conclude from this that it is reasonable for the world to be ruled by error.³⁷

liberals. See H. Jefferson Powell, “Reviving Republicanism,” 97 Yale L. J. 1703, 1709 (1988). Ackerman, for example, argues that neutral dialogue precludes asserting that one’s conception of the Good is superiorto others’. Ackerman, supra note 10, at 37. If, however, someone argues that her conception of the Good is superior and gives evidence ofits superiority, is she violating this constraint? See Perry, supra note 17, at 10. If not, then the dialogic test doesn’t exclude religious arguments. If so, then the dialogic test rejects political positions that no one would want rejected.

On the indeterminacy of dialogic tests generally, see George W. Dent, Jr., “The ‘Tensions’ of Liberalism,” 38 Phil. Q. 481, 484-5 (1988).

³⁵ See David Estlund, “Book Review,” 20 Pol. Theory 694, 696 (1992). If one is to imagine a dialogue that is without limits on time, in which the participants are all similarly motivated, and in which all information - including the personal experience kind - is equally available, then one will envision a convergence of beliefs. Religious beliefs will be no different in this respect from other beliefs, including (self-referentially) the belief that normative truth is what emerges from such an interpersonally transparent dialogue regarding what ought to be done. See Nagel, supra note 5; see also Richard Arneson, “Socialism as the Extension of Democracy,” 10 Soc. Phil. & Pol’y 145, 170-1 (1993) (arguing that dialogic tests are really just monologic counterfactual thought experiments regarding what all rational, informed, etc., people would agree to).

The tendency among followers of Habermas is, indeed, to build all of the substantive po­sitions they themselves accept as correct into the preconditions for the idealized dialogue. See Walzer, supra note 30, at 11. The “dialogue” then will produce “agreement” on precisely those substantive positions. How could it not do so? In the end, this merely shows that however much we acknowledge that points of view exist different from our own, points of view that are in some way reasonable for those whose points of view they are, we are stuck inside our own point of view, which for us is the measure of truth of all other points of view. Even the “postmodern” recognition that our point of view is just that - a point of view - and that it has been socially constructed, has no practical consequence for us. No available alternative to our point of view exists.

³⁶ See Nagel, supra note 5, at 161-4.

³⁷ It may be useful at this point to rehash what I am attempting to demonstrate by arguing for epistemological unity. My first point is that liberalism’s truth entails the falsity of some of Ann’s beliefs. Pure liberalism entails that all of Ann’s beliefs regarding public policy of those that we have mentioned are false. Impure liberalism entails that her beliefs regarding public support and endorsement of her religion are false, and perhaps also her beliefs that pornography should be banned. My second point is that liberalism’s claim to its truth and the falsity of Ann’s opposing

Nagel clings to the hope that someone like Ann, with her beliefs, can see that others are reasonable in rejecting her beliefs, and that, in some trumping sense of “reasonable,” not having her beliefs imposed on them is also reasonable, even though her beliefs about, say, abortion (obviously reasonable from her point of view) include the belief that her belief about abortion should be imposed. Nagel thinks erroneously that the “reasonable” alternative to coercive imposi­tion of Ann’s beliefs is governmental neutrality or relegation of the matter to private choice.³⁸ However, given what Ann’s views are regarding abortion - or beliefs cannot rest on the claim that liberalism’s supporting epistemology is different from - and fairer than, more respectful of autonomy than, more consistent with our free and equal nature than, etc. - Ann’s religious epistemology.

My point is that if liberalism is correct and illiberal religious views, therefore, are incorrect, then liberalism is correct not by virtue of its operating in a separate epistemological domain, one in which it does not have to meet the evidence supporting illiberal religious views head-on. No meta-epistemological vantage point exists from which it is possible both for illiberal religious views to be correct and also for liberalism to be correct that imposing correct religious views is unfair in some trumping way (where a component of such views is that imposing them is all right).

In addition to offering the reasonable rejectability/public accessibility test and the dialogic test as grounds for excluding religious positions from determining public policy, the liberal might offer the Rawlsian veil of ignorance as a conceptualization of fairness that supports such an exclusion. Would persons behind the Rawlsian veil eschew reliance on religious convictions? They would not want to have false views forced upon them, but likewise they would not want to be prevented from acting on true views. Moreover, risk aversion does not solve this dilemma, because without doing theology, one cannot meaningfully compare the two risks. One cannot determine the position regarding religious beliefs that would emerge from behind the Rawlsian veil without specifying what persons behind the veil believe religiously. Furthermore, if one so specifies, each specification will produce a different outcome. Persons behind the veil would exclude illiberal religious beliefs only if their religious beliefs were liberal, i.e., only if their religious beliefs held the beliefs’ coercive imposition wrong in principle (or held such imposition to be wrong in practice because unlikely to succeed). See Galston, supra note 23, at 146-9.

The instinct among liberals to place religious claims on a different epistemological level is so powerful that it affects even those who expressly repudiate such a move. Marshall forcefully asserts that religion is on the same epistemological level as the political theory of liberalism but then, in the same article, argues that admitting religion into the public forum should be resisted. His reason is that religion, as the answer to humanity’s anxiety over death, evokes such powerful and often dark emotions that it must not be given access to state power. Compare William Marshall, “The Other Side of Religion,” 44 Hastings L. J. 843, 845-7 (1993) with Id. at 858-9, 863. Interestingly, Marshall’s position assumes that religious answers to the mysteries of life and the terror of death are either false or unknowable. For his proto-utilitarian argument against religious participation in the public square is advanced from no particular point of view, and surely not the point of view of religions that deny that Marshall’s dangers justify their exclusion from the public square. Put differently, even if Marshall’s gloomy forecasts are well-founded, the consequentialism animating Marshall’s arguments is of no higher status as an argument than religious arguments that deny the value he attaches to outcomes. He would surely fail to convince one who was willing to shed blood to save souls.

For a liberal position similar to Marshall’s, see Galston, supra note 23, at 177-82. For an acceptance of Marshall’s premise regarding the need to avoid violence, but a rejection of his estimation of the threat posed by admitting religion to the public square, see Gedicks, supra note 24, at 696 n.122.

³⁸ Nagel’s view is that some beliefs are objective in that they appeal to more than the fact that they are believed and to a common ground of justification. See Nagel, supra note 5, at 159-62. pornography, animal rights, and public funding of religious education - obviously the “neutral” or private choice position is nothing more than rejection of Ann’s views.³⁹ Ann does not believe that abortion, viewing pornography, eating meat, or failing to fund religious education is wrong just for her, but that these acts and omissions are wrong for everyone.⁴⁰ No neutral position, intermediate between Ann and those who reject her views, exists.⁴¹

Perhaps the strongest argument the liberal might use against Ann without engaging Ann’s views frontally, other than one based on practical impossibility, would be an argument that denies that we can ever be sufficiently certain about any religious beliefs, or beliefs about the Good, to justify their coercive impo­sition. According to this argument, we must allow diversity of such beliefs to flourish as an experiment to test which beliefs ultimately gain universal assent, the best gauge of their truth.⁴² The view is analogous to the marketplace of ideas justification of freedom of expression and has analogous theoretical difficulties. Universal assent as a test of truth of religious views is unproblematic, not be­cause universal assent establishes truth, but because universal assent leaves no dissenters to object to its coercive imposition. Short of universal assent, we are faced with the fact that the number of possible experiments yet untested is, and always will be, infinite; yet coercive decisions are, in fact, always and inevitably being made, always potentially in error, and always foreclosing experiments that might have led to views commanding universal assent. Permitting abor­tion, if in error, costs millions of lives. Moreover, permitting abortion prevents

Raz points out that no one appeals to the fact of his own belief to justify the belief. Raz, supra note 1, at 37-9. Either all beliefs satisfy Nagel’s criterion of appealing to a common ground of justification, or his criterion is far too strong and rules out direct perceptions and experiences as legitimate bases for imposing beliefs. Much of science and all of morality would be excluded as legitimate grounds for coercion. See Raz supra note 1, at 45-6; Perry, supra note 17, at 12,120. Actual consent is too strong a condition to demand for coercion, and reasonable consent makes consent superfluous: No gap exists between what is a reasonable principle and a principle to which it would be reasonable to consent. See Raz, supra note 1, at 37-9.

³⁹ The same point can be made with respect to the education of children. Whether conducted by the state or by parents with the coercive backing of the state, education cannot be neutral regarding religion. Pure liberalism has real difficulty handling the education of children without assuming that education does not affect the later adult’s ability to choose among theories of the Good.

⁴⁰ “Privatization” of Ann’s views - translating them from “it is wrong for anyone to abort” to “it is wrong for me to do so” - is nothing other than rejecting them.

⁴¹ As Patrick Neal puts it,

Liberalism understands itself as a political theory which permits no substantive conception of what constitutes the good life for individuals to take public, political priority over any other; it is hence neutral with regard to the question of the good.... The politics of neutrality is conducted within a language which is, like its competitors, non-neutral; those who do not speak it as a matter of course in liberal societies are provided, sometimes against their wishes, with a translator. In the last analysis, that translator is the state.

Patrick Neal, “A Liberal Theory of the Good?,” 17 Can. J. of Phil. 567, 578-9 (1987).

⁴² See D. A. Lloyd Thomas, In Defense of Liberalism 36-8 (1988). us from carrying on the experiment of living in a society without abortion. More obviously, certain views of the Good requiring everyone to conform to certain norms cannot be tested in the liberal marketplace of diverse individual lifestyles. On some religious views, the costs of their not being coercively imposed are quite high (for example, millions of innocent lives lost, millions of souls un­saved), and no “neutral” ground is available.

I come to the conclusion, then, that liberalism and religion are on the same epistemological level, and that the knowledge each claims, if it be knowl­edge, has the same pedigree in experience and reason. There are not two ways of “knowing,” religious and secular/liberal; there are not both sectarian and secular/liberal “truths.” As a consequence of epistemological unity, liberalism must establish its tenets by rejecting conflicting religious ones, not by the il­lusion of “neutrally” banishing them to the “private” realm, where they can somehow remain “true” but impotent, but by meeting them head on and show­ing them to be false or unjustified.^[311] Liberalism is, as many critics claim it to be, the “religion” of secularism. That does not mean that liberalism is false or that antiliberal religious views are true. What it does mean is that both liberalism and antiliberal religious views inhabit the same realm and make conflicting claims within it. Liberalism is not at a different level, where it can remain neutral and impartial with respect to religious controversy that is truth-seeking within a restricted domain, but not within the domain of liberalism.^[312]

That liberalism and religion are epistemological rivals has two basic impli­cations. One, obvious and banal, is that the truth of those core, defining tenets of liberalism entails the falsity of all conflicting religious tenets. That much follows from the law of noncontradiction. More importantly, and the burden of the bulk of my argument, liberalism cannot establish its core tenets and its repudiation of illiberal religious ones (and banishment of religion from public policy making) by claiming to inhabit a different epistemological realm from that occupied by religion, a realm whose truths not only trump those of religion, but whose truths can be seen to be reasonable even by those whose religious truths they trump. No epistemological perspective exists from which one can simultaneously hold Ann’s views and, barring a belief that to do so would be self-defeating, hold that the state should not impose them.^{[313] [314] Furthermore, to the extent that liberalism defines itself by the proposition that religious views can be “true” and important enough to protect, but cannot be fairly imposed through public policy, a public policy that draws its “truths” from a different epistemological well, to that extent the unity of epistemology undermines liberalism.46}

Liberalism can rest on agnosticism regarding some truths, but not regarding its own truth. Toleration of illiberal religions may rest on the value of autonomy (so long as those religions do not threaten autonomy), or toleration of illiberal religions may rest on a prediction that intolerance would provoke a backlash that would threaten liberalism to a greater extent than toleration. The case for toleration of illiberal religions, however, cannot rest upon their possible truth without self-contradiction.^[315]

Ultimately, then, the only reason to exclude religious views from the realm of coercive public policy - for the liberal or anyone else - is because those views are wrong. And the liberal’s particular problem is that he believes it wrong to extirpate erroneous views coercively.

In the end, the source of the liberal’s regard for autonomy is that each of us has a particular point of view from which we cannot escape. The source of the liberal’s dilemma is that some points of view will be closer to the truth than others, including the truth about the value of autonomy itself.

Liberalism’s attempt to claim neutrality vis-a-vis religious views through some sort of epistemic abstinence is a failure. And from the failure issues a paradox: Liberalism can be neutral only toward those religions and religious views that are compatible with the tenets of liberalism. Which is to say that if liberalism is defined in part by neutrality toward religious beliefs, liberalism is impossible.

B. Liberalism, Freedom of Association, and Illiberal Groups

I turn now to that second pillar of political liberalism, freedom of association. It is my contention that the same paradox that bedevils liberalism’s stance

⁴⁶ What the unity of epistemology undermines is liberalism as a political theory resting on a distinction between the epistemology of political theory and the epistemology of religious and other comprehensive world views. The unity of epistemology does not undermine concerns for autonomy, social harmony, fairness, etc. Those concerns must, however, be defended within the same epistemological domain as their contradictories. Furthermore, they cannot be correct if their interpretation denies this unity of epistemology. See also Stanley Fish, “The Dance of Theory,” in L. C. Bollinger and G. R. Stone, eds., Eternally Vigilant 199, 227,231 (2001); Fish, supra note 44, at 997:

[Liberalism] is therefore committed at once to allowing competing world views equal access to its deliberative arena, and to disallowing the claims of any one of them to be supreme, unless of course it is demonstrated to be at all points compatible with [its] principles.... It follows then that liberalism can only “cherish” religion as something under its protection; to take it seriously would be to regard it as it demands to be regarded, as a claimant to the adjudicative authority already deeded in liberal thought to reason. toward illiberal religious bedevils its stance toward associations formed around illiberal principles. There are many historical and contemporary examples of such groups. To take two, a gated community might organize around precepts that include: sex-segregated schools; exclusion of television programs, movies, and print media that contain violent or sexual material; mandatory religious observance; rule by a nondemocratic community government; and so on. Or a Native American tribe might mandate various forms of sexual inequality within tribal life and practice racial inequality with respect to tribal membership.

To this point I have been treating liberalism as a rather univocal political/ moral conception. It will be useful to complicate that picture somewhat in showing the paradoxical relation between liberalism and freedom of illiberal associations.

1. two philosophical liberalisms. There are two basic philosophical conceptions of liberalism that are in tension with one another. On one con­ception, the conception that has been dominant in the discussion of religion, liberalism demands “neutrality” among choices of the Good - choices about how to live, with whom to associate, and what to believe.^[316] On the other concep­tion, liberalism itself represents a choice of the Good - in this case, the choice of a cosmopolitan way of life.^[317] Both conceptions of liberalism are problematic, each exposing the principal theoretical weakness of the other.

a. Liberalism as a “Neutral” Umbrella for Illiberal Persons, Associations, and Communities. As individuals, we cannot be neutral about what is good and what is true. To live is to make choices - to pick A over B because we prefer A, or value A, or believe A to be right. We practice a particular religion and not others. We choose to read some things and not others, and to say some things and not others. We choose to associate with some people and not others. At the individual level, no one lives the life of a neutral liberal, one who does not side with any particular points of view, values, or persons. There is no such life. Such a life is a theoretical as well as practical impossibility. At the individual level, all lives must be “illiberal.” This is the most fundamental illiberal bedrock upon which liberalism as neutrality must build.

If the individual may choose - because he must choose - a version of the Good, liberalism as neutrality demands that he not be legally prohibited from acting on that choice. That is, on liberalism as neutrality, it is the state and its laws that must be neutral in the required sense of neutrality.^[318]

The picture then is of a neutral state umpiring among the myriad beliefs, values, and ways of life that necessarily illiberal individuals might choose. But the picture is in fact more complicated than that. If an individual may choose a conception of the Good, may he not associate with like-minded others - in families? in cities? in nations? Indeed, if the individual’s conception of the Good itself dictates these associations, then his choice of conceptions of the Good is really a choice of in which illiberal society to live. His choice may, of course, be as limited as a choice of mate or friends, or a choice of a particular religious congregation with which to associate for a couple of hours weekly. But it may be as total a choice as one to live within a gated community that subscribes to a code regulating religion, reading matter, education, and dress, or to live within an even larger society with similar rules.^[319]

Liberalism as neutrality then is consistent with everyone living very illiberal lives. Everyone might opt into total communities, communities in which reli­gion, speech, occupation, and association are completely regimented. Indeed, it is possible that everyone would opt into the same total community. Could such a society truly be a “liberal” one?

The defender of liberalism as neutrality might reply that as long as everyone freely consents to such regimentation and can opt out of total communities at any time, then total communities and liberalism can co-exist. In other words, as long as the state preserves the freedom to exit any community, relationship, and so forth as the corollary of the freedom to enter those communities and relationships, we have a liberal state.

This picture of liberalism raises several issues. First, what if someone chooses a community or relationship that precludes later opting out - for example, a marriage that is indissoluble, or a community that requires a pledge of lifetime fidelity as condition of membership? This is the problem of freely entered con­tracts of slavery with which liberals from J. S. Mill on have wrestled. On the one hand, “liberal enslavement” seems oxymoronic. On the other hand, all contracts restrict future choices; and the refusal to enforce freely chosen restrictions of future choices would not be liberating but would instead limit liberty, given that the ability to enter into liberty-restricting enforceable covenants is itself liberty-enhancing.

Second, if the ability to opt out of total communities preserves liberalism, how is that ability to be gauged? It is insufficient merely to provide a legal permission to exit. For exit from a group might be very costly. One might have to forfeit property. One might have to move. One might need to acquire information about available alternatives to one’s current way of life, and what sacrifices one would have to make to choose them. (If, for example, all land belonged to gated communities of different flavors, exiting one would require joining another.)

Interestingly, American constitutional law doctrines take these concerns into account. For example, not only state governments, but also all sub-units thereof - counties, cities, school districts, water districts - are bound to adhere to liberal constitutional tenets. One cannot have a municipality or a school district restricted to members of a particular religious creed or race.^[320]

51

Moreover, in the United States, constitutional and other legal doctrines limit the powers of private communities and organizations. Company towns cannot restrict free speech.^[321] Private racial zoning cannot be legally enforced.^[322] Large organizations can ordinarily be statutorily forbidden to be racially or sexually restrictive.^[323] Moreover, it is doubtful that one can, by contract, give private communities or associations the power to punish rule violators with fines or imprisonment, given that states themselves are constitutionally forbidden to punish for breaches of contracts, even if the contracts provide for punishment.^[324]

Some liberals criticize at least some of these constitutional doctrines. Mark Rosen, for instance, would permit municipalities, possessed of ordinary police powers, to be organized along religious lines.^[325] And indeed it is at least plau­sible to argue that many of these constitutional limitations are fundamentally inconsistent with liberalism as neutrality. Perhaps we should assume, then, that if constitutional doctrine were aligned with a proper conception of liberalism as neutrality, individuals could consent to all sorts of illiberal restrictions on their liberty, and the government could and would permit such restrictions, at least through self-help, and perhaps with the backing of its own coercive mechanisms.

Let us put aside the single total community and ask how liberalism as neu­trality bears on a world of more than one illiberal group. First, different illiberal groups might make conflicting claims regarding land, resources, and activities. They might, for example, claim the same mountain as a sacred site leading to conflict over who may use it, how, and when. Or one group might believe in the moral necessity of hunting eagles, while another believes in the moral necessity of protecting them. And, of course, one group might believe that abortion is murder, and that it has a moral duty to prevent it, by force if necessary, while another believes that its members are morally at liberty to have abortions.

Is there a “neutral” way for government to adjudicate these intergroup controversies? None that I can see. After all, government must at the end of the day side with one group or the other, at least if neither of the opposing groups has been convinced of the error of its beliefs and values. And to side with one group on the matter in dispute is the antithesis of neutrality, at least from the point of view of the losing group. Put differently, because there is no view from nowhere, the “neutral” position will be as partisan as any other. Or put still another way, all views appear neutral to those who hold them and partisan to those who do not.^[326]

Second, as stated above, liberalism as neutrality requires that no illiberal group exercise control over a person’s liberty or property without that person’s consent. This means that government must monitor the quality of people’s consent to the terms of entry into and exit from illiberal groups.^[327] Such consent can only be valid if the person giving it has (morally) adequate alternatives to submitting to an illiberal group’s discipline and morally adequate information about alternatives to such submission. That means that government must ensure a level and breadth of education sufficient for understanding the stakes in joining an illiberal group and for making rejecting that group a viable alternative.^[328] For example, if group A speaks language A, and the only alternative to joining group A is joining group B, whose language is B, X’s consent to join A is arguably vitiated if X has not been educated in B, or has not been taught of B’s existence or its creed, and so forth. Or if the only alternative to A is B, but B is a long way away or has poorer resources or is otherwise extremely costly to join, consent to joining it may be vitiated.

It is commonplace, however, that no educational program can be neutral among groups.^[329] The only “neutral” truth of liberalism as neutrality is that liberalism as neutrality is correct. All other truths - all possible curricula - will be partisan along some axis or another. Because we are finite beings, with finite minds, finite attention, and finite time, our education must perforce be selective. And the criteria of selection can never be neutral. Just to take one well-known example from American constitutional law, one cannot educate Amish children about the world beyond their community sufficiently to make their consent to remain in that community meaningful without instructing them in values and information that the Amish will view as quite partisan.^[330] Indeed, the very worldliness required to satisfy the liberal of the validity of the Amish’s consent to the norms governing their way of life might be fundamentally inconsistent with that way of life.^[331] How are you going to keep them down on the farm, once they have seen the world beyond? But how can they give morally meaningful consent to being down on the farm if they have never seen the world beyond?

Liberalism as a neutral umbrella for illiberal persons, associations, and com­munities therefore faces a predicament that appears to undermine it completely. On this conception of liberalism, government must permit illiberal choices and communities and relegate its own role to that of neutrally umpiring the con­ditions of entry and exit. However, the role of neutral umpire is completely chimerical.

b. Liberalism as Cosmopolitanism. The second conception of liberalism ad­mits to being itself a way of life, a vision of the Good, a partisan view among partisan views. On this conception, a liberal society is one that has within it a diversity of religions, associations, occupations, ideas, and so forth to pro­vide its members with a rich menu of ways of life from which to choose.^[332] The liberal on this conception values pluralism, diversity, openness, and toler­ance. He may rule as out of bounds certain values and associated ways of life because he thinks they are degraded or counterfeit; but in general his motto is “Let a thousand flowers bloom.”

On this conception, liberalism is not the above-the-fray values of a neutral umpire but is rather a particular way of life, namely, that of the cosmopolitan.^[333] There are, however, both conceptual and constitutional problems with liberal cosmopolitanism.

Conceptually, one cannot be a liberal cosmopolitan without there being illib­eral choices and ways of life toward which to be cosmopolitan. As stated at the outset, at the level of individual choice, no one is a “liberal.” Everyone prefers value A over value B or value B over value A. Cosmopolitanism then must en­vision individuals’ tolerating and appreciating ways of life but not being deeply committed to them. Deep commitment to A would imperil tolerance and appre­ciation of B. But this has two unwelcome implications. First, cosmopolitanism as a way of life is shallow, denatured, bereft of deep commitments. Second, cosmopolitanism as a way of life is parasitic on noncosmopolitan ways of life; for without the latter, there is nothing for the cosmopole to be cosmopolitan about.

One might think that both of these points are overstated, and in some sense, they are. After all, there are “liberal” religions, religions that teach that there are many paths to salvation, that tolerance of other creeds is a virtue, and so forth. And many ways of life are not so total and exclusionary that they cannot coexist with and, indeed, foster appreciation of alternative ways of life. The farmer and the urbanite, the artist and the doctor, the Francophone and the Anglophone need not seek dominance over or eradication of the opposing way of life but can coexist and even relish one another.

Nonetheless, cosmopolitanism inevitably tends to homogenize and shallow out the various ways of life. If there are many paths to truth or salvation, then little is at stake in finding a path. After all, the path you are on is probably just as good as any other. And if you had chosen another path, the same would have been true of it. Without “fighting faiths,” faiths worth fighting for, there will be little spice in the variety that the liberal cosmopole celebrates. As the Boy Scouts’ brief to the United States Supreme Court in Boy Scouts of America v. Dale^[334] stated, “A society in which each and every organization must be equally diverse is a society that has destroyed diversity.”^[335] That is, indeed, the indictment of liberal cosmopolitanism’s greatest engine of conquest, the global commercial culture, namely, that it produces a shallow homogeneity and supplants a richer, deeper, but more antagonistic diversity.

Liberal cosmopolitanism also fits uneasily with the American constitutional tradition, which, in line with liberalism as neutrality, is protective of illiberal groups. If Catholics believe that priests should be male, or if some funda­mentalist sect believes that its congregation should be racially segregated, the American constitutional right of religious free exercise would likely trump an­tidiscrimination laws. And the constitutional right to freedom of association that the Supreme Court believes is implicit in First Amendment freedom of speech protects the Boy Scouts’ right to exclude gay scout masters and Irish-American groups’ right to exclude gay Irish groups from their St. Patrick’s Day parade.^[336] If liberalism is a way of life, then in the United States, it is not constitutionally enshrined to the extent that would permit government to outlaw illiberal ways of life.

2. yale university, coed dormitories, and orthodox jews: a case study in the problems of liberalism.^[337] Yale University is a private university that adopted policies requiring all students to reside in univer­sity dormitories and making those dormitories coeducational. Yale undoubtedly believed that liberal education is best promoted if men and women, as well as students of different races, ethnic groups, religions, geographical origins, social classes, and so forth, live side by side. In a very real sense, Yale’s vision of a good educational community is one of liberal cosmopolitanism.

A group of Orthodox Jewish students at Yale objected on religious grounds to being forced to reside in coeducational dormitories. Eventually Yale relaxed its rule and allowed these students to live elsewhere. But the incident pro­vides a lens on the two versions of liberalism and how they deal with illiberal groups.

The proponent of liberalism as neutrality would ask whether the Orthodox Jewish students could be deemed to have consented to Yale’s policies. That question would lead to examining the students’ information about their alterna­tives and the viability of those alternatives. But unless Yale possessed an unfair share of the resources required for a first-rate liberal arts education, the students would almost surely be regarded as having all the alternatives to Yale to which they were entitled, which in turn would mean that the costs of rejecting Yale would not vitiate their consent to Yale’s rules. And unless one believes that the students cannot validly consent to live off-campus unless they have experienced the coeducational alternatives - which would argue against their being allowed to live off-campus rather than in favor of it - there seems to be no problem under liberalism as neutrality with permitting Yale to run its dormitories as it wishes. The liberal neutralist’s response to the Orthodox students’ complaint is that they are free to reject Yale but not to change it.

Of course, if Yale itself is a liberal society, then the liberal neutralist’s re­sponse must change. Now the Orthodox students have a right to secede and form their own illiberal society within the greater liberal society that is Yale. So, too, might the black students or the gay students be entitled to their own housing. Only if Yale is a unit within liberal society rather than a unit of liberal society may it call the shots.

The cosmopolitan liberal’s take on the Yale controversy is different. By re­quiring integrated dormitories and rejecting group secession, Yale is furthering cosmopolitanism. By living together, all will come to tolerate and even appre­ciate the diverse viewpoints, values, and traditions that the students bring to the table. A Yale education, including its living arrangements, will be transfor­mative and will produce cosmopolitan liberals out of the parochial, sectarian students who enroll.

The problem, of course, is that to the extent Yale succeeds in transform­ing its students into cosmopolitan liberals, there will be little for them to be cosmopolitan about. Perhaps the Orthodox Jews will, after experiencing coed dormitory life, change their beliefs about sexual modesty and the like and come to think about such matters much as the rest of the students do. But if they do so - and other students lose their parochialisms - the result will be not diversity, but homogeneity. Thin gruel for a cosmopolitan.

Indeed, it is no coincidence that the move to greater diversity on college campuses has proceeded in lockstep with the move to orthodoxy known as “political correctness.” Real cosmopolitanism is unstable and self-undermining. It must domesticate and mute the diversity it celebrates. As Nomi Stolzenberg puts it:

The fundamentally agnostic attitude that underlies [the cosmopolitan’s] principle of tol­eration strains against the respect for individual beliefs that the principle also requires. The inescapable tension between agnosticism and respect for religious beliefs is another manifestation of the opposition between the fundamentalist principle of biblical in­errancy and modern, critical approaches to religious claims. The doctrinal treatment of a [religious proponent’s] beliefs as that individual’s beliefs and nothing more denies them respect, or serious consideration, on their own terms.^[338]

3. whither liberalism and illiberal groups? If liberalism is an outgrowth of and response to the European wars of religion - the quintessential conflict among illiberal groups - has it succeeded? In one sense, surely the answer is that it has. Religious freedom and toleration are well established in the West and are at least aspirational for a great many outside the West.

In another sense, however, liberalism’s victory is more doubtful. For it has been accomplished largely through liberalizing and privatizing religions. Most Christians and Jews differ on matters that have few public implications. They are now taught to respect or at least tolerate those who hold different beliefs, and that the ethical prescriptions of their religion are mainly private and in any event not worth fighting about. Even the issue of abortion, the issue with perhaps the most potential for toppling liberalism’s peace, has not done so.

Liberalism has thus solved the problem of illiberal groups - to the extent it has - mainly by eliminating illiberal groups, not by the sword, but by offering peace and prosperity as powerful temptations to abandon illiberal outlooks. The homogenization brought about by commercialization, globalization, and the relegation of religion to once-a-week private observance has been liberalism’s main instrument in achieving its victory.

So liberalism has not so much solved the problem of illiberal groups as it has eliminated the problem by reducing their number and membership. In doing so, however, liberalism has not truly triumphed. For without “illiberalism” - differences in beliefs and mores that are deep and conflicting - liberalism has nothing about which to be liberal. On the other hand, neither liberalism as neutrality nor liberalism as cosmopolitanism has the conceptual resources to deal with illiberal groups. Both versions of liberalism try to stand above the partisan fray, but neither can succeed. There is only partisanship - illiberalism if you will - at both the bottom and the top of the stack of turtles on which liberalism attempts to rest.

John Gray, on the other hand, believes that because of the truth of value­pluralism, there are diverse and incompatible good lives, and that this rules out any political theory, including liberalism, that posits one fundamental set of freedoms, rights, and principles of justice.^[339] The best we can aspire to, argues Gray, is a modus vivendi among different ways of life, the diverse and incompat­ible goods they make possible (and impossible), and the freedoms, rights, and principles of justice they require.^[340] Even John Stuart Mill and Isaiah Berlin, the two political philosophers most appreciative of the plurality of real values, erred in believing that those values could be reconciled at some fundamental level such as utilitarianism (for Mill) or negative liberty (for Berlin).^[341] The modus vivendi that Gray advocates is one achievable not philosophically but only po­litically.^[342] For Gray, liberalism is not a unified, coherent political philosophy but a family of political systems that resemble one another in certain respects but not others and that carve out to a greater or lesser extent room for different, antagonistic ways of life.^[343] On Gray’s account, when society is faced with con­flicts among groups, there is no liberal algorithm available for resolving them, nor is there any resolution that does not sacrifice some values for the sake of others that are incommensurable.^[344] We are all members of illiberal groups, and there are no principles for sorting out our differences. It’s politics all the way down.