Equal Law in an Unequal World

“And are there other sorrows, beside the sorrows of poverty?

And are there other joys, beside the joys of riches and ease?

And is there not one law for both the lion and the ox?”

                                            –Blake, Visions of the Daughters of Albion1


 “One Law for the Lion & Ox is Oppression”

      –Blake, The Marriage of Heaven and Hell2


The principle that the law must be general—that it must apply equally to all—is a fundamental demand of legal morality, associated with the ideal of the rule of law.3 But many worry that this generality, the “formal” equality of law, props up substantive inequalities in a hierarchical world in which people have different capacities, endowments, and fundamental interests. For example, if the law is forbidden to recognize that there are a dominant race and subordinate races, and respond to those facts (e.g., with affirmative action), it can reinforce that hierarchy, and it has the chutzpah to do so under the very name of “equality.” Thus, Derrick Bell has claimed that “despite law school indoctrination and belief in ‘the rule of law’—abstract principles lead to legal results that harm blacks and perpetuate their inferior status.”4 Morton Horwitz has said that the rule of law “creates formal equality” but in doing so “promotes substantive inequality by creating a consciousness that radically separates law from politics, means from ends, processes from outcomes.”5 Catharine MacKinnon has decried the conventional rule-of-law command that like cases be treated alike for “tell[ing] women that they are entitled to equal treatment mainly to the degree they are the same as men.”6 Robin West has characterized the standard conception of the rule of law as “a serious threat to progressive, egalitarian, and identity-based politics.”7 Nor is this a purely contemporary worry. It goes at least as far back as Rousseau, who blamed the law itself for entrenching and legitimating inequality: “all the inequality which now prevails . . . becomes at last permanent and legitimate by the establishment of property and laws.”8

This is a theoretical problem at the heart of liberal democracy. The rule of law is ordinarily understood to be a basic condition for the legitimacy of liberal states.9 Its demands are expressed in provisions of the U.S. Constitution like the Due Process and Equal Protection Clauses.10 And its ideals are deeply embedded in the Anglo-American legal tradition, from Magna Carta through the common law.11 Indeed, West has suggested that the “formal equality” model of reasoning about law, the practice of categorizing cases with respect to similarity and dissimilarity with the object of treating “like cases alike,” is just part of what makes legal reasoning legal reasoning, as opposed to some other kind of reasoning.12 If the rule of law is hostile to genuine equality, then that hostility is at the heart of our constitutional order.13

In this Article, I work toward relieving the worries about this putative hostility. I develop an egalitarian conception of the rule of law, which I began in previous work.14 Here, my main task is to answer the complaint about formal equality by showing how the rule of law actually promotes genuine, substantive equality.

I respond to these important concerns of Bell, MacKinnon, Horwitz, and many others by clarifying how we determine whether a society does or does not respect the rule of law, and using these clarifications to show that the regulative idea of the rule of law can ground an objection, rooted in the principle of generality, to social inequalities such as racial injustice and poverty. This theoretical work is critical for practical concerns at the heart of the legal system: since the rule of law can help political communities in their struggles to address social inequality, by extension, so can constitutional ideas like equal protection and due process.

Part I describes the egalitarian conception of the rule of law, and then gives an account of what it means for a law to be general. For a law to be general, for purposes of the rule of law, it must be justifiable by public reasons, understood as an expressive idea (Part I.A.). To be justifiable by public reasons is to be justifiable by reasons that each person affected by the law can reasonably accept, conceiving of him or herself as an equal member of the political community. To determine whether this criterion is satisfied, we look to the social meaning of a law—this is the “expressive” part of the account (Part I.B.). The Part concludes by intervening on some recent debates in legal philosophy with a novel account of how the social meaning of a law may be found (Part I.C.).

Part II applies all this theoretical work to a concrete case: the postbellum literacy tests. The literacy tests were inconsistent with the rule of law, and with the equal standing of freed slaves in the political community, because there were no public reasons available for them. Importantly, this judgment depends on nonlegal social facts about postbellum America, particularly, that freed slaves and their descendants had been denied an education.

Part III generalizes that last point and teases out its implications. Our rule-of-law judgments depend not merely on legal facts, like the language of legal rules or the actions of public officials, but also on the nonlegal social facts that give meaning to those legal facts. Moreover, this implies that the demands the rule of law generates can be disjunctive (Part III.B.): when some law, taken in its social context, violates the rule of law, the state can comply with the rule of law either by changing that law or by changing the underlying social context. Finally, sometimes the state may be foreclosed for other reasons from changing the law; under such circumstances the rule of law requires the state to change the underlying social facts.

Part IV applies the ideas generated in the previous Part to the problem of economic injustice. In societies where there is avoidable extreme poverty, many of the basic laws structuring economic rights and private property violate the rule of law for essentially the same reason as did the literacy tests. In the face of prima facie compelling objections to abolishing those laws, it follows from the logic of Part III that the rule of law forbids states from permitting avoidable extreme poverty to exist in their territorial bounds. The rule of law requires something like a sufficientarian distribution of resources, wholly independent of the freestanding theories of distributive justice that dominate the literature.15

Part V pauses to address a family of objections about the demandingness of the conception of generality advanced here. I show that the rule of law is still distinct from all-things-considered justice, and is coherent as a freestanding legal value.

Part VI discusses the broader implications of the substantive conception of generality. Many on the left have criticized the rule of law by arguing that its emphasis on “formal equality” either fails to contribute to the fight against real social inequality or actively impedes it.16 On the right, critics have concurred in this judgment and argued that the rule of law prohibits, inter alia, social welfare states.17 I argue, against both positions, that the rule of law is consistent with the demand for social justice.18

I.     The Idea of General Law

A.     The Egalitarian Conception of the Rule of Law, a Synopsis

In other work, I have given an account of the rule of law.19 From that account, the rule of law is a regulative principle of political morality governing the way in which officials who exercise the state’s coercive power relate to those over whom they exercise their authority.20 It may helpfully be understood by dividing its demands into two levels.

The first level, the “weak version of the rule of law,” comprises those formal principles that prevent state officials from ruling by terror and arrogantly claiming a superior position in a status hierarchy.21 A state satisfies the weak version of the rule of law when it satisfies the principles of regularity (which requires that official coercive power only be used against individuals when authorized by law) and publicity (which requires that officials justify their use of power by appeal to law that is available to citizens, in that they can know the law and use it to defend themselves).22 In the Anglo-American constitutional tradition, these principles are largely instantiated by the rules that have been developed under the rubric of due process of law (such as the prohibition against vague law) and related requirements (like the requirement of public trials, the prohibition against ex post facto laws, etc.).23

The weak version of the rule of law marks out the difference between a thug state, like Haiti under the Duvaliers, dominated by swaggering officials whose unconstrained power intimidates citizens into cowering submission, or a Kafkaesque state, like the Soviet Union, where law is something done to the public, the tool of a hostile officialdom rather than the political community as a whole, and a state in which power is effectively regulated by legal form and those subjected to state coercion have some opportunity to stand up for themselves.24

However, the weak version leaves off a major part of the conventional conception of the rule of law. Ordinarily, rule-of-law scholars say that the law must be general; that is, it must apply to all on equal terms, rather than carving out special privileges for some citizens over others. The strong version of the rule of law captures this principle of generality. In the United States, the Equal Protection Clause expresses this principle.

Nothing in this Article depends on the weak version, the development of which comprised the brunt of The Rule of Law and Equality. However, to reach the conclusion of the present argument, it is necessary to take the strong version substantially further than before. The remainder of Part I.A draws on the account of generality in The Rule of Law and Equality, and is consistent with it, but reformulates the arguments and develops them further. Subparts I.B and I.C are wholly new (as is the rest of this Article).

B.     Against the Formal Conception of Generality

The idea of general law can be conceived as either formal or substantive. Define a formal conception of generality as one according to which an observer can determine whether a law is general purely by examining properties of the law itself, including its text, the process by which it was enacted, and/or the actions and motivations of legislators. A formal conception of generality corresponds to a “classification” or “antidiscrimination” approach to Equal Protection Clause jurisprudence, one in which a law is subject to equal protection scrutiny to the extent it draws facial classifications between groups of citizens or is motivated by a classificatory intent.25

By contrast, a substantive conception (corresponding to a “class” or “anti-subordination” approach to equal protection) requires an observer to examine nonlegal social facts and/or appeal to normative values (such as “liberty” or “equality”) in order to determine whether a law is general. The conception I described in The Rule of Law and Equality and develop here is substantive.

This Part argues that the formal conception of generality is necessarily incoherent. In order to do so, I distinguish between, and reject, three different sub-types of the formal conception.

On the minimal conception of generality, the law is not allowed to pick out particular people. This conception forbids things like the bill of attainder, or the law with a proper name in it.26 In addition to proper names, this minimal version of the principle must (on pain of absurdity) also forbid laws that incorporate other rigid designators that refer to people, such as indexicals used in the right context.27 For example, it would prohibit a king from pointing at someone and saying “you are hereby outlawed.”

On the epistemic conception of generality, laws are forbidden to the extent that those who enact them know (i.e., can pick out) to whom they are to apply. This conception is distinctively associated with Hayek.28

Finally, on the similarity conception of generality, law must be cast in general (or “abstract”) terms, or treat every citizen the same.29 These conceptions propose to police the extent to which the law classifies citizens into different groups in order to ensure that it “treat[s] like cases [and citizens] alike.”30

The minimal conception fails because it is unstable along the dimensions of both uniqueness and rigidity, which are the only two plausible criteria by which we might distinguish the laws it forbids from the laws it permits. First: if the law may not contain rigid designators referring to one person, it would be irrational to permit it to contain rigid designators referring to multiple people. That is, if the rule of law forbids the legislature from enacting “Thomas Wentworth may not work as a lawyer,” it must also forbid “Thomas and Margaret Wentworth may not work as lawyers,” and if it forbids that, it must also forbid “Thomas, Margaret, Sarah, John, Phillip, etc. Wentworth may not work as lawyers,” or “none of you people whom I am addressing right now may work as a lawyer.”31

Second, if the law forbids rigid designators, it must also forbid at least some non-rigid designators that, in the actual world, are extensionally equivalent to rigid designators.32 This is clearest in the individual case: if the legislature may not enact “Thomas Wentworth may not work as a lawyer,” it also may not enact “the person who lives at 1640 Attainder Lane on July 30, 2012 may not work as a lawyer.” Otherwise, the prohibition against rigid designators would be practically meaningless, since the legislature could always find a sufficiently precise non-rigid designator that would pick out exactly those whom the legislature wished to attaint.

The instability of the minimal conception along the dimension of number and its instability along the dimension of rigidity can combine: from the above, it follows quite naturally that the rule of law forbids the legislature from enacting “nobody in the family of the person who lives at 1640 Attainder Lane on July 30, 2012 may work as a lawyer.” And after taking that step, we’ve lost both of the candidate principles (uniqueness and rigidity) by which we might distinguish those descriptions the minimal formal conception of generality forbids and those it permits. If the state cannot pick out the class of people who live at a given address for special [mis]treatment, can it pick out nobles as a class, or even citizens as a class? The minimal conception of generality offers us no answer.33

To see that the epistemic conception fails, simply ask: “knows under what description?” If the legislature passes a law “all redheads must serve in the army,” each legislator knows exactly to whom the law will apply, under the description “redheads,” even if none know each individual by name. The same is true if the legislature enacts “everyone who lives at 1125 Attainder Lane is to be shot,” just in case legislators are not quite sure of the names of the residents. Either the epistemic formal conception just reduces to the minimal formal conception (and collapses for the same reason), that is, to the demand that the legislature must not know those to whom a law can apply by name (or other rigid designator), or it fails to constrain laws, because legislatures always know to whom a law applies under the description written into the law.

The best contemporary liberal legal theorists have endorsed the similarity conception, in the form of the command that the law “treat like cases alike.”34 I argued against the similarity conception in The Rule of Law and Equality, and simply summarize here.35 All legislative acts are formally non-general in the sense that they indicate conditions for the application of law, only some of which any given people will meet. All legislative acts except for those that actually contain rigid designators are also formally general in that they specify in abstract terms (for some level of abstractness) the criteria for their application. To put it differently, all cases, and people, are alike in some respects and different in some respects.36 The demand to “treat like cases alike” requires a non-formal criterion by which we may pick out the features of the cases that are relevant for determining whether they are “like,” for generality purposes, or not.37

 Consider an example: People with disabilities are dissimilar from people without disabilities; black people are also dissimilar from white people. Yet, taken in a formal sense, the command “treat like cases alike” cannot help us understand why it is permissible to enact the law “the seats at the front of the bus are reserved for people with disabilities,” but impermissible to enact the law “black people must sit at the back of the bus.” Intuitively, we know that disability is relevant to bus seating in a way that race is not, but that relevance judgment comes not from some formal idea of what it means to treat like cases alike but from our deeper moral and political commitments to making the world accessible for the disabled and to avoiding racial segregation.

In addition to the problems we have in figuring out which cases are “like,” it is also often difficult to figure out what constitutes “treating” them alike in any formal sense. Consider taxation, and compare three laws: the first institutes a progressive tax structure in which the richest have the highest percentage of their income taxed. The second institutes a “flat tax” or “proportional tax” in which everyone pays the same percentage. The third institutes a “head tax,” in which everyone pays the same absolute dollar amount.

Someone might argue that the head tax is general while the proportional tax is not, because everyone pays the same amount only under the head tax. But we might also argue that the proportional tax is general while the head tax is not, because the proportional tax inflicts a similar amount of disutility (or pain) on everyone, considering the diminishing marginal utility of wealth, while the head tax inflicts more disutility on the poor. Put differently, the proportional tax (arguendo) inflicts on everyone the same disutility, while the head tax inflicts on everyone the same dollar cost. Alternatively, we might argue that the proportional tax is general while the head tax is not because the proportional tax takes the same percentage of income from everyone, while the head tax takes a higher percentage of income from poorer citizens. It is not obvious which of these arrangements should count as “treating like cases alike.”

Similarly, Hayek argued that the proportional tax is general while the progressive tax is not.38 But, in absolute dollar terms, compared to the head tax, both the proportional tax and the progressive tax are graduated systems in which the rich pay more than the poor, so it is rather puzzling that one is allegedly general while the other is not. The point is that which tax counts as formally general under the similarity conception shifts under different descriptions of the same taxes.

Ultimately, the judgment of generality is ineluctably substantive and normative: when we say a law is general, we mean that it doesn’t pick out its classes of application in a way that offends the values that lay behind imposing the requirement of generality in the first place.39 The same goes, mutatis mutandis, for the Equal Protection Clause; this is why our constitutional jurisprudence does not just decree that the law is not to treat people differently, but sets up a substantive kind of scrutiny that evaluates different kinds of classifications (race, gender, etc.) by different standards and judges them with respect to the overall social ends (“compelling government interests” and the like) to be served. More on this below.

Since we ordinarily say that the principle of generality captures the idea of equality under law, and since the rule of law as a whole is best understood as a regulative principle aimed at preserving equality in political communities, the relevance criterion that allows us to apply the requirement of generality should capture the idea that the subjects of law are to be treated as equals.40 Thus, I argued in the previous paper that a law is general, for the purposes of the rule of law, only to the extent the distinctions it draws between citizens are justifiable by public reasons, in something like John Rawls’s sense.41 That is, such a law is general only to the extent that it is justifiable by reasons that those whom it picks out might be able to accept, conceived of as equal citizens rather than inferiors.42 Here, I say some more on behalf of this position.

This reinterpretation of the idea of general law as law that is justifiable by public reasons captures a high-level similarity between the two ideas. Public reasons are reasons that can be addressed to all citizens. The law, in turn, is general when it genuinely is addressed to all.43 This mode of address comes in the form of reasons, or, in the eloquent words of John Finnis:

The rule of forms of law is far removed from the rule of law. Law is at its strongest and most real when it actually shapes and spans and links the thoughts of the decision-makers to the thoughts of all whom it addresses, including the decision-makers themselves in their subsequent actions. Then law is public reason which is most public and most reasonable precisely because it is shaping the private reasoning of every decision-maker, “public” or “private”. It is most factual when its empirically palpable, “positive” manifestations are understood as evidence—albeit also formative—of its more actual reality as the set of normative propositions (reasons for action) made true by reasonable decision, constantly reaffirmed, to consider the social facts of legislation, adjudication, etc., as having an intelligent and reasonable relationship with the strategic moral truths that we need rule by law in order to preserve and promote our common good as persons all needing friends in the face of both anarchy (private force and fraud) and “official” force and fraud.44

I would add to Finnis’s exposition that this linking of the thoughts of those who make decisions to those who obey them, through the medium of reasons that are accessible to all (public reasons) expresses the equality of all citizens. This remainder of this Article is devoted to filling out that claim.

I am not the first commentator to suggest that the idea that like cases should be treated alike be reinterpreted to command treatment consistent with the underlying equality of the subjects of law. Robin West argued as much in advancing a “humanistic” account of the ideal of formal equality, one that holds that to declare that someone else is “like” the legal decision-maker or like some privileged class is to recognize that the person under consideration is an equal member of the legal community.45 In this way, West’s account might be described as an ideal of non-exclusion: we violate the principle of generality when we allow the law to treat someone as a nonmember, to exclude them from the community of shared interest and mutual respect represented by the political state. The account in this Article is consistent with West’s, and develops it by giving a decision procedure by which we can sort out whether someone has been so excluded.

Before moving on, note that the conception of public reason appropriate for the rule of law is more capacious than that offered by justificatory liberals like Rawls, who are concerned with elucidating the appropriate relations of citizens in a liberal democracy.46 The rule of law can lend moral value to nonliberal as well as liberal societies, and non-democracies as well as democracies, and the correct conception of the rule of law must be compatible with nonliberal societies and non-democracies in order to capture that feature of the concept.47

I am also not the first to suggest that something like public reason can be applied even in non-democratic states to give the grounds by which they may distinguish between their citizens. A prototype of the idea goes at least as far back as Hobbes, the great defender of monarchy, who nonetheless insists that legal distinctions between classes of citizens be grounded on the public good:

      The safety of the People, requireth further, from him, or them that have the Soveraign Power, that Justice be equally administred to all degrees of People . . . .

      . . . The honour of great Persons, is to be valued for their beneficence, and the aydes they give to men of inferiour rank, or not at all.48

By contrast, the full-blooded Rawlsian version of public reason is designed to describe a fundamental value of liberal democracies, and does not work outside that context.49 For example, Rawls suggests that laws subject to the criterion of public reason may not be justifiable only by comprehensive religious reasons (e.g., “God commands it”).50 That requirement works for pluralistic liberal societies, but in a nonliberal religious society offering such a religious reason may be consistent with the equality of each citizen. If that is so—if in the society in question, it is understood by all that laws are to be justified in the terms of the state religion, and that doing so offers no disrespect to nonbelievers—then, for the purposes of the conception of the rule of law I have been developing, and continue to develop here, such a reason would count as public.51 That clarification in hand, I now turn to an analysis of how we figure out whether a law is consistent with public reason.

C.     Public Reason: Expressive

The requirement of public reason as it applies in the rule of law context is helpfully understood as expressive, in the sense given by Anderson and Pildes.52 To see this, consider that the standard formulation, given by Rawls, is that a public reason is “at least reasonable for others to accept . . . as free and equal citizens, and not as dominated or manipulated, or under the pressure of an inferior political or social position.”53 However, it is unclear what it might mean for it to be “reasonable” to so accept.

This reasonableness requirement might be understood in the first-person sense, from the point of view of the person offering the reason (the sovereign or its representative(s)). However, this is under-demanding: it would entail that a law is general whenever those who enact it think that those whom they regulate ought to agree, without regard to what the regulated actually think. Alternatively, it might be understood in the second-person sense, from the point of view of those to whom the reason is offered. But this is over-demanding. It would amount to giving those regulated by a law a veto over that law, since if they reject the reasons for it they will naturally think that it is not reasonable to demand they accept those reasons.54 Nor is there likely to be some kind of objective “view from nowhere” third-person source of the judgment about whether it is reasonable to demand someone accept the reasons for a law.55

Instead, we should understand these reasonableness judgments as conventional, drawn from the understandings shared by the members of a political community. It is unreasonable to demand that someone accept a reason if, in the community shared by the reason-giver and the reason-taker, demanding that reason be accepted is not something one does to a free and equal citizen, and accepting that reason is not something one does when one sees oneself as a free and equal citizen. That is, to fail to offer public reasons is one way in which one might fail to treat the one to whom reasons ought to be offered with the respect owed to a free and equal citizen, as that status is understood in the community in question.56

This is an expressive standard of behavior in Anderson’s sense57: it sets up a valuation of (an evaluative attitude toward) an object (“equal” attached to the reason-taker), and generates the demand that one behave in the way appropriate to that valuation (by giving only reasons consistent with it).The match of appropriate reasons to valuations is given by the social meaning of those reasons and that valuation. And as Anderson explains, to take an appropriate evaluative attitude to something is, in part, to act in the way that, in one’s social world, one acts when one holds that attitude.58

As I will argue in a moment, this exercise amounts to finding the social meaning of a law: does it express the equality of the citizens it regulates, or does it not? To determine the social (or “expressive”—I use the terms interchangeably) meaning of a law is to determine the valuation to which it is appropriate. The expressive meaning of a law is comprised of those attitudes, about those regulated, which members of the relevant community must attribute to the relevant agent in order to rationalize that law. In the remainder of this Part, I will give an account of how to identify those attitudes (and the agent).59

D.     Finding the Expressive Content of a Law

1.     Reasons and Meanings

In the rule of law context, we need to use the expressive content of a law not only to figure out whether the reasons under which a law is justified are consistent with conceiving of all members of the community as free and equal, but also to determine what those reasons are in the first place.

We are not engaged in a mind-reading exercise in which the object is to sort out what the legislature was thinking. We are engaged in a justificatory exercise in which the object is to sort out whether a law can be justified in the right sort of way to each member of the community. The inquiry is about whether a law could, in principle, be publicly justified, not about whether some legislators said the right magic words or subjectively held an attitude of respect toward those regulated. If a public reason is available, even if not actually in anyone’s brain, for a law, then that law is general.60

But reasons, for present purposes, are agent-relative. In saying this, I consciously steer clear of a philosophical morass that has mired many commentators: the controversial question of whether there are agent-independent reasons.61 Instead, I assume that what Rawls has called “the fact of reasonable pluralism” is true with respect to the sorts of alleged agent-independent reasons typically raised in political or legal dialogue, such as comprehensive religious or moral theories: any political/legal actor who claims that some such reason applies will not be able to demand or count on the assent of others in the community to that reason.62 Even in a nonliberal society, alleged agent-independent reasons, such as the claims of the dominant religion, will only count as public reasons to the extent they are embedded into the public political culture of a legitimate state worthy of the adherence of members of minority religions, in virtue, in part, of the fact that it treats them as equals; that is, they only count as public reasons because they can be redescribed as agent-relative reasons like the interest of each person in the community in equal treatment and a legitimate state.63 I conclude that all public reasons will be agent-relative reasons. And since the inquiry at hand is a search for public reasons for a law, we need not bother with looking for any other sort.

As I will argue in a few paragraphs, the inquiry into the expressive meaning of a law is rationalistic, in that it amounts to an inquiry into reasons associated with a law, and constructive, in that it attributes those reasons to the occupiers of several standpoints with respect to the law, based on the reasons that apply to people in those standpoints. That is, it attributes agent-relative reasons to legal actors. It follows that to attribute to all relevant agents the reasons they might endorse a given law is both to exhaust the logical space for expressive meanings of that law, and to exhaust the possible public reasons for that law. The public reason inquiries and the expressive meaning inquiries turn out, when the dust settles, to be the same thing.

But now the task of applying the principle of generality may begin to look very difficult. It is notoriously difficult to find the expressive content of a law, or, indeed, anything, and there are many who worry that it might be indeterminate, and the inquiry futile.64

The most useful recent attempt to give a method for carrying out this task (albeit in the course of a critique of the research program) is Blackburn’s.65 He proposes to resolve the indeterminacy problem by using a “credibility” conception of what counts as the expressive content of an act: an act (or set of acts) “P” expresses an attitude “A” when the only credible explanation for why the agent, conceived of as rational, carried out the act includes the proposition that the agent was committed, implicitly or explicitly, to that attitude.66

Making this more concrete: the law “black people sit on the back of the bus” could only credibly be understood by citizens in mid-twentieth century America if they attributed the attitude “black people are inferior” to the lawmakers. And since offering the inferiority of black people as a reason for the law would not satisfy the strictures of public reason (being inconsistent with the status of black people as equals), the law “black people sit on the back of the bus” would be non-general for rule of law purposes.

Unfortunately, Blackburn’s credibility conception doesn’t seem to give us much help. Different people, with different understandings of the social meaning of an enactment, might disagree about which attitudes may be credibly attributed to the legislature, or (as will be developed in a moment) to those called upon to obey a law, or to the community at large. Some acts (like bus segregation) have obvious social meanings, but others may be unclear or subculturally dependent.67

2.     A New, Distinctively Legal Approach

All is not lost. Blackburn fails to attend to the fact that laws, as distinct from all other expressive acts, implicitly make claims about the particular ways in which (a) legislators, (b) those called upon to obey the law, and (c) the community at large are supposed to relate to the law. By attending to these special properties of laws, from those particular standpoints, we can fill out their expressive content in a way that we cannot so easily accomplish for other acts.68

Specifically, legislators are supposed to enact laws for rational, purposive, and collectively oriented reasons: a law, to not be arbitrary, has to be rationally directed at some ostensibly public end.69 We can understand the expressive content of a law from the first-person perspective of the legislature enacting it in terms of the end at which it implicitly claims to be directed.70

As to those whom a law commands, the law demands it be taken as authoritative, that is, as giving exclusionary reasons for actions. And that claim to authority in turn depends on the claim that the law helps them act according to reasons that already apply to them.71 We can understand the meaning of the law from the second-person perspective of the one called upon to obey a law in terms of the reasons that it implicitly claims to help them apply.

Finally, as to the general members of a community for which a law is enacted, the law claims to be enacted in their names.72 As such, it claims to be consistent with their self-understanding as a political community and the relationships with one another that self-understanding instantiates. We can understand the meaning of the law from the third-person perspective of the general member of the community in terms of the self-understanding with which it implicitly claims to be consistent.

Moreover, the language of law is the language of reasons; those who participate in legislation and law-obedience do so with the presupposition that there is a rational connection between the reasons for a law and the law itself.73 For this reason, interpreting a law is analogous to interpreting a language in the rationalistic approach associated with Donald Davidson. Davidson has developed a theory known as “radical interpretation,” which gives an account about how we interpret the utterance of any speaker.74 In doing so, we rely on a “principle of charity” which assumes that the speaker holds true beliefs and speaks honestly, including a “principle of coherence,” which requires us to take the utterances of the speaker as logically consistent, and a “principle of correspondence,” which requires us to attribute to the speaker beliefs that we take to be true. Accordingly, “[s]uccessful interpretation necessarily invests the person interpreted with basic rationality.”75 I propose to apply these concepts to understand how communities can interpret the expressive meaning of their laws.

In sum, I claim that we may find the expressive content of a law by bringing four theses together:

(1) Expressive meanings are conventional. The expressive content(s) of a law is the content that it has in the community in which it is enacted, from the standpoint of that community, and cannot be determined apart from the social facts of that community, including its history and the way its members currently relate to one another. The inquiry is about social facts, not psychological facts about legislators or anyone else.

(2) Laws have meaning from three points of view. Laws have expressive content from the first-person, second-person, and third-person standpoints, corresponding to the point of view of the legislator, person regulated, and ordinary member of the community. However, the content of each of these standpoints is to be interpreted in light of the first principle, that is, we must understand the expressive content of a law as the meaning that the community at large must attribute to the law from the first-, second-, and third-person standpoints—not the subjective content of the brains of the legislators, people who are called upon to obey, and ordinary citizens.

(3) Law makes distinctive claims. The expressive content of a law is distinct from the expressive content of any other act, because laws make special demands on those who interact with them. As discussed above, laws claim to be rationally directed at public ends, to give exclusionary reasons to those who are to obey them, and to be enacted in the names of the members of the community in general.

(4) Expressive meanings of laws are rationalistic. This act of interpretation must be carried out, per Davidson’s principle of charity, by attributing true, rational beliefs to the occupiers of each standpoint, where those beliefs are the reasons for the occupier of each standpoint to interact with the law in the way appropriate to each standpoint (enact it for that reason, obey it for that reason, etc.).76

Using those principles, I can specify the expressive content of a law from each of the three standpoints. From the first-person standpoint, the members of a community may attribute expressive content to a given law by answering this question: What attitudes must a legislator in our community hold in order to rationally enact this law for some public purpose?77

From the second-person standpoint, the members of a community may attribute expressive content to a given law by answering this question: What attitudes must those whom the law commands hold in order to rationally take this law as helping them to act according to reasons that already apply to them?

From the third-person standpoint, the members of a community may attribute expressive content to a given law by answering this question: What attitudes must we hold in order to rationally take this law as enacted in our names and expressing our self-understanding as a political community?

This account borrows techniques from ethical constructivism to give the content of expressive values. In ethics, constructive views idealize (to a greater or lesser degree) human interests and reasons, from standpoints specified by the view and/or by people’s actual positions in the world, and derive moral claims from them.78 This theory of law’s expressive meaning takes idealized interpretations of the reasons that apply to people, from the three standpoints relative to the law which they may occupy, plus the claim that laws must be rational to people in each of those standpoints, and uses those building blocks to make claims about what the law must mean, expressively, to the occupiers of each standpoint. There are two distinct idealizing steps. The first is to attribute reasons to legislators, those called upon to obey a law, and those in the political community in whose name the law is enacted. The second is to attribute beliefs about those reasons (the reasons discovered in the first idealization) to members of the community at large. The point is that those meanings need not correspond to actual thoughts held by any of those people. A law can have (say) insulting meaning even if, empirically, nobody in the community actually thinks the law is insulting, just so long as the interpretation of the law according to which it is insulting is the correct way to interpret it in its social context.79 We don’t take an opinion poll to find out the expressive meaning of a law, we reason (from an external standpoint) about what community members should think.

Nonetheless, expressive meanings are social facts—observers do not get to just make them up. Rather, the reasons that observers may attribute to a law depend on the obligations and interests of, and constraints on, those in a given community at a given time. Moreover, while the expressive meaning of a law does not depend on how people in the community actually interpret it, ordinarily, the best evidence for the expressive meaning of a law will be the interpretation that actual people in the community give to a law. If the constructed interpretation of a law differs from the actual interpretation in the community, that is a reason to worry that the constructed interpretation is wrong, and to investigate it further with additional evidence or argument.

Also, there are some situations where the expressive meaning of a law depends (in a non-evidentiary sense) on the actual interpretations given it in the community. This is particularly likely where the law commands some symbolic or communicative behavior with a preexisting meaning.80 For example, if the state requires redheads to wear dunce caps, we should attribute an opinion about redheads’ intelligence to the legislature just because dunce caps are actually understood in the community to signal stupidity. If people had understood pointy caps to mean something else, then the law would have a different meaning. But this is not true of all symbolic laws; sometimes the state can confer meaning on a symbol where none existed before. For example, the yellow star did not, as far as I know, denote any kind of inferiority independent of the laws, but it nonetheless carried that meaning when the Nazis began requiring Jews to wear it.

I also note that a law may have multiple expressive meanings. This is not a problem for the account. If there is any public reason available for a law, then that public reason should correspond to an expressive meaning for that law which incorporates reasons consistent with the equality of each citizen. If any such meaning is available from each standpoint, the law is general.81

Finally, the expressive meaning of a law may change over time, because the social facts underlying that meaning may change. This entails that the correct rule-of-law evaluation of a law may also change over time: a law may be general at one moment and non-general at another.82 That’s not a problem: there are many acts and institutions whose moral evaluation may change over time, as understood by ordinary moral and political theory. For example, a utilitarian will accept or reject a law depending on the extent to which that law maximizes well-being or preference satisfaction; this evaluation may change over time as people’s preferences or needs change.83

E.     Proof of Concept

Consider a concrete example: the law “black people sit on the back of the bus.” This is a very easy case: de jure segregation is non-general if anything is. But going through the analysis will help clarify how generality works. The law will satisfy the principle of generality if and only if some public reason can be plausibly offered for it from each of the three standpoints. Each standpoint is necessary, because all laws serve a triple function, as purposive public policy (corresponding to the first-person standpoint), obligation-generating legal command (the second-person standpoint), and expression of the community’s self-understanding (the third-person standpoint). If a law cannot serve one of those functions without making use of the idea that some members of the community are of superior or inferior status, the law as a whole expresses the inequality that the principle of generality is meant to forbid.84

Considering the first-person standpoint, those in mid-century America would attribute expressive content to it as follows: “There’s no obvious public purpose for this law, except to express something about how black people and white people are to relate to another. In our social world, black people are ordinarily treated as inferiors, so a rational legislator, in the world in which we live, must accept that black people are indeed inferior and intend to reinforce that existing hierarchical treatment in order to enact this law.”85

Considering the second-person standpoint, they would attribute expressive content as follows: “Why should a black person sit on the back of the bus? There is no obvious reason that applies to black people except for reasons about their relative status and some duty to behave in accordance with it. Given our social environment, in which black people are understood as inferiors, the only reason that could be being served by such a law is a duty on behalf of black people to act in accordance with this inferior status. Therefore, to rationally take this law as authoritative, a black person must accept his or her own inferiority.”

Considering the third-person standpoint, they would attribute expressive content as follows: “Why would we, as a political community, have a stake in bringing it about that black people sit on the back of the bus? What reasons apply to us such that we might rationally endorse the social arrangements that this law brings about? Since the only effects of the law are to separate the subordinate caste from the dominant caste and to physically manifest underlying social relations, we must believe that it is right for black people to be subordinate in order to endorse this law.”

Unsurprisingly, the law “black people sit on the back of the bus” expressed the inferiority of black people from all three standpoints. And since the inferiority of black people is not a public reason, and, within the social context of mid-century America, no other reasons could plausibly be assigned to such a law, the law was not justifiable by public reasons.

Before moving on, it is important to clarify that although the expressive meaning of the bus segregation law was set by the way that those in the community should have understood it, our moral evaluation of that meaning is set by universalistic standards. That is, the law “black people sit on the back of the bus” can only be said to express the inferiority of black people in the social context in which it was enacted. In a different social context, it might have a different meaning. (We might imagine a culture in which the rear of a seating area is a symbolic position of esteem.) But the moral evaluation of a given socially determined expressive meaning does not itself depend on social facts. Once we determine that some law expresses the inferiority of some members of the community, that law is to be condemned on rule-of-law grounds whether or not anyone (or, indeed, everyone) in that community endorses this message. Even if both black and white people agreed that black people were inferior, and that it was appropriate to express this inferiority through segregation, that would not make the laws acceptable from the standpoint of the rule of law.86

F.     Public Reasons and Insults

The example just given has a feature often seen in non-general laws: it insults some members of the political community. The antimiscegenation law at issue in Loving v. Virginia is the classic case of such an insult. According to the Supreme Court, the law was “obviously an endorsement of the doctrine of White Supremacy.”87 As an endorsement of white supremacy—and, by necessary inference, nonwhite subordination—it violated the Equal Protection Clause. For the same reasons, it violated the strictures of public reason, and thus the rule of law.88 I will use the concept of “insult” in the remainder of this Article as shorthand for laws that are justifiable only by non-public reasons that express the inferiority of some members of society.

It is worth taking a moment here to distinguish my conception of generality from Hellman’s account of the grounds for criticizing legal discrimination. Hellman argues that legal distinctions are morally wrong when they demean those picked out by the distinctions.89 I agree. However, demeaning, or insulting, is sufficient but not necessary for a legal distinction to be unjustifiable by public reasons. A legal distinction can be unjustifiable by public reasons for reasons wholly unrelated to the insult it offers—if, for example, it simply disregards the interests of some citizens, or is unjustifiable by any coherent reasons at all.

For example, a law taking ten percent of the income of people living on even-numbered street addresses and giving it to those on odd-numbered street addresses would not be justifiable by public reasons, even though it degrades nobody. Although it does not degrade those robbed, it does fail to be justifiable by reasons consistent with treating them as free and equal citizens, for two reasons. First, it represents a total disregard of the interests of those living on even-numbered streets—if not even a minimally plausible argument is available for the proposition that their interests are protected by the law, it cannot treat them with the respect due to equals. (From the second person standpoint, what reasons could obeying the tax possibly help those on even-numbered streets follow?) Second, the law is not in fact justifiable by any reasons at all, beyond the whim of the lawmaker. “I enact this law because I can make any laws I please” is not a reason consistent with seeing those over whom the laws are to be enforced as equals.90

Under such circumstances, the public-reason conception of generality imposes something like the rational basis test from American constitutional law. To see this, compare the house numbers law to a law taxing those who do not own the means of production, and giving the money to those who do own the means of production. Such an unjust and regressive tax could nonetheless be justified by public reasons, in that it does not necessarily represent a disregard of the proletariat, but could be explained by a (foolish but coherent) theory of how their interests may be served (encouraging investment from the capitalist class).91 From the first-person standpoint, the community at large can imagine a legislator enacting the law for that reason; from the second-person standpoint, of a citizen obeying the law for that reason, and so forth.

However, I emphasize that even if insult/degradation is not necessary for a law to fail to be general, it is sufficient. This is so even if those who are insulted are not materially disadvantaged.92 For example, suppose the federal government ceased demanding an oath of office from any elected official except Muslims. This wouldn’t seriously disadvantage Muslims, because the oath is a de minimis burden. Nonetheless, the community at large, considering the law from the first-person standpoint, could only understand the limitation of the requirement to Muslims by attributing to the legislature a presumption of Muslim disloyalty. That would explain our rule-of-law and equal-protection objections to any such law.93

From the foregoing, it should be evident that U.S. equal protection doctrine tracks the public reason conception of generality fairly well. Through the heuristic device of suspect classes and levels of scrutiny, it applies a demanding test to those laws most likely to express an insult, and tests other laws only to see whether they are totally unreasoned.94

II.     The Rule-of-Law Critique of the Literacy Tests

Now we are in a position to apply the public-reason conception of generality to a more difficult case. Consider the postbellum literacy tests. Let us leave aside the fact that they were implemented unfairly, and consider what might be said of them, by way of moral evaluation, in the abstract.

As a first pass, we might say that they were unjustified because they were motived by the wicked intent to disenfranchise black people. But simple motive-based condemnation is unsatisfactory for several reasons. First, one might (and I do) think that the literacy tests would have been morally wrong even if they had only incidentally excluded black people from the polls. Second, there are good reasons to deny that our moral judgments of an act or an institution depend on the subjective state(s) of mind of those carrying out the act or implementing the institution.95

It is especially inadvisable to base our moral evaluation of a legal institution on the wicked intent of those who created it for two reasons. First, legal institutions are ordinarily created by the combined actions of a number of agents, who may not all have been acting for the same reasons. Laws, and the institutions they create, are products of politics, which is in turn a domain of compromise; the fact that any individual legislator voted for a bill does not entail that either that legislator shared any intentions with others who voted for it or desired the consequences.96 Moreover, social choice theorists have shown that legislative outcomes may not reliably track even the bare preferences of legislative majorities.97

Second, basing moral evaluations of legal institutions on intent confuses the moral evaluation of people (the racists in the legislature) with the moral evaluation of institutions (the racist laws). Those two judgments ought not be combined; it is perfectly plausible for them to come apart—bad people with bad intentions can enact good laws, and good people with good intentions can enact bad laws. And when confronted with wicked social institutions, we ought to be able to criticize those institutions independent of those who imposed them. For one reason, institutions may survive their creators, and future generations ought to be able to criticize them without anyone being at hand to blame.98

Nor should we criticize them in virtue of their racially disparate effects alone. Not every law that disadvantages some group within society is morally blameworthy—progressive taxes, for example, disadvantage the rich, but are not in virtue of that disadvantage subject to moral criticism. We would have to appeal to deeper ideas in political philosophy to criticize the literacy tests in virtue of their effects—to say, for example, that they were wrong because they violated democratic principles, or because they violated the principle of equal treatment of citizens.99

In this Part, I appeal to a distinctively legal, deeper principle. The rule of law can ground our moral objections to the literacy tests.

A.     A First Pass at the Argument, and the Neutrality Objection

The public-reason conception of generality explains what was wrong with the literacy tests. They could only be understood as insults to the freed slaves and their descendants, whose equal citizenship had supposedly been acknowledged by the reconstruction amendments. By making this acknowledgement into a lie, the literacy tests implied that no matter what the Constitution said, they would never be full members of the political community.

Because the enactment of the literacy tests unavoidably carried that message, the distinctions they drew between the literate and the illiterate were unjustifiable by public reasons.100 Consequently, the laws were not general.

Thus, at least, is the intuition. Someone might object and deny that the literacy tests carried any such message. While unfortunate, our objector might suggest, that the literacy tests just so happened to pick out freed slaves for disenfranchisement, they need not carry any social meaning about their less-than-full citizenship.

Further pressing this objection, one might point out that we could imagine all kinds of benign literacy tests. And if there was a non-insulting reason for the literacy tests from each of the first-, second-, and third-person standpoints, then they would not have unavoidably carried the insulting expressive content—and could have been justifiable by public reasons.

Indeed, in the abstract, there are many plausible arguments that lead to the conclusion that one ought to have literacy tests for the exercise of the franchise. For example, there are some epistemic arguments for democracy in which democratic institutions are justified by their propensity to reach better decisions, along the lines of the Condorcet Jury Theorem.101 But for the Condorcet Jury Theorem to entail that democracies make better decisions than plausible alternatives, individual voters must tend, on the whole, to have a greater than .5 probability of coming to the right answer.102 A literacy test could usefully eliminate the voters with the lowest probability of correctness, since voting well may depend on access to complex (written) information about public policy.

This sort of argument is not limited to those who think democracy is justified by its alleged decision-improving properties. Anyone who thinks it matters, morally, whether political decisions lead to non-foolish policy has at least prima facie reason to want the least competent citizens to stay home, even if that reason is ultimately outweighed by countervailing considerations relating to, for example, autonomy, pluralism, the civic educational value of political participation, or the like.103 Jason Brennan has plausibly argued that those unwilling to put in the effort to become educated about public policy have a moral duty not to vote.104 Literacy is a plausible proxy for minimal effort and competence.

Moreover, literacy tests may have useful indirect effects:

We must never lose sight of the truth, that the suffrage for a member of Parliament is power over others, and that to power over others no right can possibly exist. Whoever wishes to exercise it, is bound to acquire the necessary qualifications, as far as their acquisition is practicable to him. I have expressed my conviction that in the best possible system of representation, every person without exception would have a vote; but this does not imply that any one should have it unconditionally; only that the conditions should be such as all could fulfil. The greatest amount of education which can be fairly regarded as within the reach of every one, should be exacted as a peremptory condition from all claimants of the franchise.

      . . . .

. . . To make a participation in political rights the reward of mental improvement, would have many inestimable effects besides the obvious one. It would do more than merely admit the best and exclude the worst of the working classes; it would do more than make an honourable distinction in favour of the educated, and create an additional motive for seeking education. It would cause the electoral suffrage to be in time regarded in a totally different light. It would make it be thought of, not as now, in the light of a possession to be used by the voter for his own interest or pleasure, but as a trust for the public good. It would stamp the exercise of the suffrage as a matter of judgment, not of inclination; as a public function, the right to which is conferred by fitness for the intelligent performance of it.105

That argument seems plausible. Moreover, all of the reasons Mill offers for an educational qualification are, apparently, consistent with public reason, that is, they are all reasons for imposing a literacy test to verify that qualification that could be offered to every citizen, even those thus excluded from the franchise, consistent with understanding them as equal citizens.

Note in particular how this point works from the second-person standpoint. A disenfranchised citizen, considering whether the literacy test helps her act in accordance with reasons that already apply to her, could reason as follows: “I have a duty to vote, if at all, competently. The literacy test forbids me from voting if I’m not competent in this particular fashion (being unable to read), therefore, it keeps me from violating a duty I already have.” This is an interpretation of the law that does not require the self-attribution of inferiority.

Observe, however, the condition precedent for Mill’s argument: the state may impose such educational qualifications as “can be fairly regarded as within the reach of every one.”106 In Mill’s preferred society, I take the following to be the answer that we might offer to a disenfranchised citizen who claims to have been offered insult by being turned away from the voting booth:

You are not being turned away for anything about you that is out of your control, and we don’t intend, by turning you away, to make any claims about your personal inferiority. Moreover, as a political community in which you have a legitimate and recognized stake, we offer you the means to become qualified to vote: here, take this reading course, which will offer you numerous other benefits besides.107

That explanation sounds a little paternalistic, perhaps, but not, on the whole, insulting.

It is exactly that explanation that was not available to those who implemented the antebellum literacy tests in the United States. And the reason is that the political community did not recognize the freed slaves’ stake in it by offering them the means to become qualified to vote. Quite the contrary, freed slaves had been denied an education.108 And their descendants continued to be given inferior educations in segregated schools. It was that fact that changed the literacy tests as implemented in the south from a Millian effort to promote universal education and competent voting into an insult, incompatible with the rule of law.

We find this expressive content from the second-person standpoint, by following the method described above. Because they had been denied literacy, freed slaves could not understand the literacy test as helping them fulfill their duty to be competent voters. And the community at large could not attribute any such understanding to them. This is so thanks to the basic normative principle of “ought implies can” (with a qualification to be described in a moment).109 The freed slaves had been deprived of the means of becoming literate, therefore they had no such duty, and therefore the law cannot have helped them fulfill that duty.

All this left was an attribution of inferiority. In a world in which they could not be charged with a duty to be literate, they could only have taken the laws as helping them act in accordance with reasons that already applied to them if those reasons were derived from some intrinsic duty to not vote—not because of their contingent failures to satisfy a duty of literacy but because of their inherent unsuitability for the franchise—a belief that, of course, was already quite present in the political culture.

Before moving on, I pause to describe the qualification to “ought implies can” promised a couple paragraphs ago. Suppose that there were a literacy test in a society in which a decent education were offered to all, but in which some people were nonetheless unable to acquire literacy, for example, because of intellectual disabilities. The argument thus far seems to suggest that the rule of law would forbid literacy tests in that society, because it cannot make literacy available to some of its members. And this, in turn, seems to make the demands of the rule of law far too strong, forbidding legal privileges from being conditioned on any capacity that not every member of society is able to acquire.

The argument in this Part does not warrant that inference. Not every law that creates a legal privilege, for which not everyone can qualify, gives insult. Consider the law forbidding the blind from driving. That law does not give insult to the blind; the public reason from each of the three standpoints is obvious. In particular, the public reason from the second-person standpoint is obvious: blind people can take the law forbidding them from driving as authoritative because they have reason not to put the lives and property of others at risk by carrying out a dangerous activity when they do not have the physical capacity to do it safely. Similarly, voting is a dangerous activity. Those who vote badly can contribute to immense human suffering by permitting ruinous wars, environmental destruction, economic collapse, and many other evils; those who lack the capacity to do it safely have reason to stay away from the polls.

But one might worry that this argument also applies to freed slaves: they simply lacked the capacity to vote safely. The difference between the freed slave and the blind driver or the intellectually disabled voter, however, is that the inabilities of the blind driver and the intellectually disabled voter are the result (ex hypothesi) of nothing more than bad luck. By contrast, the state and the society at large were attributively responsible for the illiteracy of the freed slaves. And this matters, in turn, for the reasons that apply to the freed slaves. We might fairly say that I have a reason to sacrifice my own interests (in voting, in driving) to spare the rest of society from the ill-effects of my bad luck, but it is much less fair to demand that I sacrifice my interests to spare the rest of society from the ill-effects of a disability that society has imposed on me in the first place.110

In philosophical terms, this point amounts to the denial that society has the moral power to impose a reason not to vote on a citizen by denying that citizen the necessary means to vote in a permissible fashion. We can understand it in economic terms as well: society as a whole is in ordinarily a better position to avoid the risk of poor voting by those who have been denied literacy than are individual citizens. This is so because individual citizens must pay a very high cost to avoid this risk: they must sacrifice their individual political autonomy and sense of membership in the community, and, if literacy has been denied to an entire class of citizens, as it was in the slave states, they must also sacrifice the collective well-being of their group. Therefore, when society at large has the capacity to avoid the risk of poor voting, we should impose the duty on it, rather than on individual citizens.111 It follows that freed slaves did not have a duty not to vote, even though those who are illiterate as a result of sheer bad luck (arguably) might.112

III.     The Rule of Law and Social Facts

A.     Rule-of-Law Judgments Depend on Nonlegal Social Facts

The foregoing reveals that we cannot read a state’s compliance with the principle of generality off of the face of its legal texts, or even from observing the relationship between its legal texts and the practices of its officials. In the most abstract statement of this point, because generality is an expressive ideal, and because the expressive content of any legal act is at least partly conventional, and depends on social meanings which themselves depend on social facts, our evaluation of whether or not a law is subject to criticism will depend in part on prevailing social conditions when it is in effect.113

Moreover, because the correct evaluation of a law depends on social conditions, it can change when those social conditions change. Imagine that the states after the civil war had suddenly become much more liberal, and had begun to offer a decent education to freed slaves. No longer being denied literacy, the freed slaves would no longer have had reason to object to the literacy tests (assuming, counterfactually, that they were administered honestly and fairly). This point runs the other direction too: suppose Brown v. Board of Education were overruled today, and black children were again ghettoized into inferior schools; this would make the literacy tests (if any still existed) into rule-of-law violations again.114

Nor need the social facts upon which our rule-of-law evaluations are premised be so stark as the systematic denial of an education to some members of the community. Sometimes, existing attitudes toward some community members are enough to make a law nongeneral. Consider the French prohibition on religious symbols in schools.115 Outside of its social context, one might think that this law was justifiable by the distinctive French tradition of secularism in public life (laïcité). But the law was passed in 2004, in the wake of 9/11 and similar conflicts between Muslims and non-Muslims in Europe, and in an environment in which Muslims were (and are) subject to severe social stigma. In that context, nobody could believe that the law was passed in order to maintain the secular schools. That could not serve as a public reason because, in context, it was a sham.116 At the time, in the community, it could only be understood as targeting Muslims and expressing social disapproval of their faith.117

B.     The Disjunctive Character of Rule-of-Law Commands

The rule of law, as a regulative principle for political states, has the power to generate at least defeasible (if not absolute) demands for its obedience. The discussion thus far has revealed a perhaps unintuitive property of this demand: it need not be a demand for the state to change anything about its legal system. Rather, it might be a demand for the state to remedy the unequal social circumstances that make its laws objectionable from the standpoint of the rule of law.

More formally, I have argued that a law is criticizable for violating the rule-of-law requirement that the laws be general when that law, in the social circumstances in which it is found, is not justifiable by public reasons. It follows from the foregoing that in order for a state to make an objectionable law comply with the rule of law, it can either abolish the law in question, or it can remedy the social circumstances (usually injustices, social hierarchies) that make it objectionable. Thus, the rule of law issued a disjunctive demand to the postbellum south: get rid of the literacy tests, or provide freed slaves with a decent education.

IV.     The Rule of Law and the Criminalization of Poverty

A.     Majestic Equality?

Having set out my account of the principle of generality in the abstract, I turn now to applying it to the law’s treatment of the poor. Consider Anatole France’s sarcastic jibe against the law’s vain pretense of equality: “The majestic equality of the laws, which forbid rich and the poor alike to sleep under bridges, to beg in the streets, and to steal their bread.”118 France’s point seems to be that the laws forbidding vagrancy, begging and theft are not general.

It is easy to agree with France. In the context of extreme poverty, those laws leave some citizens no realistic choice but to commit the very acts that have been made criminal. Hence the widespread contemporary literature today on “the criminalization of poverty,”119 a phrase which accurately expresses the fact that a law that forbids the very poor from sleeping on the street, in a community in which some are unable to acquire anywhere else to sleep and a species in which one must sleep, is really the same thing as a law making it a crime to be poor—much like the literacy tests were really laws making blackness a disqualification for citizenship.

A law that leaves some members of the community no choice but to assume the status of criminals is a law that necessarily insults those so condemned, as it picks them out as criminals no matter what they do, just for who they are. The connection between criminal punishment and moral responsibility is broken, and the state expresses the scolding disapproval that goes along with criminal punishment against those citizens for something that is wholly out of their control.120 “No matter what you do, you’re a criminal just for being alive” is the message sent to the very poor.121 This is a “status crime” in two senses: it criminalizes the status of poverty, and it uses criminalization to reinforce the subordinated status of those who are poor.

This insult is reaffirmed in the minute daily interactions of the poor with the state. The case of the homeless is instructive. If one is homeless, and there are vagrancy laws, then merely to go to sleep is to be in constant danger of being turfed out of the “bed” one has made in some public space and chased away like a stray dog or a disobedient child. The homeless are perhaps exceeded only by those in total institutions like prisons, military barracks, and mental hospitals in the extent to which they can be ordered about.122 And every such interaction is a further blow to the dignity of the citizen who is constantly told that his very presence in a public place and his satisfaction of the basic needs of human existence are objectionable to his fellows.

The structure of the second-person expressive content of the laws criminalizing poverty is identical to that of the literacy tests. In each case, an otherwise potentially permissible law becomes non-general because the regulated citizen cannot understand it as authoritative without accepting his or her own inferiority. And in each case, this failure of authority comes as a result of the fact that the reasons to which the law would ordinarily respond have become unavailable to that citizen. Thus, in the case of the criminalization of poverty, in a society in which some citizens were not reduced to destitution, she could understand the law against sleeping on the street as directed at a preexisting duty not to inconvenience one’s fellow-citizens in public places by doing so. But “ought implies can” again arises, where some citizens are destitute, to vitiate the preexisting duty not to sleep on the street. She must sleep to exist; if she is to sleep at all, it must be on the street. Or, evidently, in jail.

Consequently, the only way that a destitute citizen can understand the vagrancy laws as helping her satisfy reasons that already apply to her is if she assumes that her existence itself violates a duty toward her fellow citizens—that is, if she believes that she is contemptible, unfit to be a member of the community. That is the belief the community at large must attribute to her in order to understand the expressive meaning of the vagrancy law in a social context where extreme poverty exists.

Not only does the criminalization of poverty offer insult to the poor, and thus violate the rule of law on the grounds that it is not general, but it violates perhaps the most uncontroversial and fundamental rule-of-law requirement as well: the principle that the law’s commands must be capable of being followed.123 Seen in this light, it is surprising that the literature does not contain a sustained discussion of the relationship between the criminalization of poverty and the rule of law.

Perhaps the implicit assumption behind the dearth of literature on this subject is similar to the explicit assumption behind Mill’s case for literacy tests: the rule of law does not ground objections to laws against vagrancy, begging, and theft if the community offers each citizen a real opportunity to not be subject to those laws, that is, to raise her economic condition to the point where she has alternatives to engaging in the forbidden behavior.

Alternatively, it may be that some things that I have been claiming are consequences of poverty are actually consequences of a misfortune that the state is not responsible for and does not have the capacity to avoid. This might be true if, for example, homelessness is largely a result of mental illness rather than poverty, and overriding civil-liberties concerns prevent the state from forcibly treating or institutionalizing those whose mental illnesses render them homeless.124

In Part II, I argued that the illiteracy of freed slaves gave them no reason not to vote, because society at large was responsible for their illiteracy, not the freed slaves. Likewise, here, the argument depends on the proposition that society at large bears responsibility for extreme poverty in virtue of the way it shapes the economic constraints under which citizens live (the “basic structure,” in Rawls’s terms125), and could do something to alleviate extreme poverty. Those are empirical questions; we need only note for present purposes that they are fairly debatable.126 The point is that if the state gives some members of the community no practical choice but to sleep under bridges, beg in the streets, and steal bread, then the laws against vagrancy, begging, and theft violate the rule of law.127

B.     The Rule-of-Law Critique of Economic Injustice

The foregoing confers a new dimension on our economic justice discourse. Recall the disjunctive character of rule-of-law judgments. If France and I are right that the rule of law prohibits things like vagrancy and theft laws in the social context of extreme poverty, then it demands an end to one or the other.

States probably ought to have laws against theft. On many accounts, the definition and protection of private property is a defining purpose of political states, and this is one of the chief lines of argument against the anarchist who denies that they are justifiable in the first place.128 Even a socialist society, though it will not have private property rights in the means of production, might defensibly have private property rights in personal goods—like food.129 And laws against theft are constitutive of private property rights—without a law forbidding everyone else from taking away my belongings, it would be silly to say that I have a private property right in them in the first place.130 A state with no law against theft of food may fail so completely in doing what states ought to do that it cannot defend itself against the anarchist critique.

The final step now presents itself. If the rule of law is inconsistent with a law against stealing food in a society in which some must steal in order to eat, and if in consequence the rule of law generates a disjunctive demand to either repeal the law against theft of food or correct the unequal social conditions in which some are starving, and if it would be impermissible for independent reasons to repeal the law against theft of food, then, one fork in the disjunctive road must be blocked off. The rule of law generates the demand to put a stop to extreme poverty or abolish the laws against theft; we cannot abolish the laws against theft, therefore the rule of law generates the demand to put a stop to extreme poverty.

The rule of law forbids every political state from permitting anyone within its territorial jurisdiction to be so poor that they have no practical option but to steal to eat. The rule of law, qua distinctively legal regulatory principle, generates this prohibition. Not some controversial theory of distributive justice, not Rawlsian justice as fairness, not utilitarianism, and not human rights. The same value that picks out the difference between a minimally legitimate state and a tyranny also sets a hard floor to the amount of poverty that it may permit.131

This claim has surprising implications for political philosophy and constitutional theory. Recently, Michelman has considered whether the elimination of poverty should count as a “constitutional essential” in Rawls’s framework—that is, as a minimum legitimacy condition for obliging all citizens to participate in the political community.132 Rawls accepts a relatively weak constitutionally essential antipoverty commitment: a “social minimum” guaranteeing everyone the satisfaction of their “basic needs.”133 For Rawls, more demanding principles of fair economic distribution are not to be constitutional essentials, in part because constitutional essentials are to be enforced by judges, and judges do not have the “full understanding of how the economy works” necessary to do so.134

The account of the rule of law given in this Article expands upon Rawls’s point by specifying the content of the social minimum. Citizens are to be ensured enough resources to satisfy the presuppositions embedded in the law, whether or not those resources are needed to satisfy their basic needs.

Ordinarily, the law will presuppose that citizens can satisfy their basic needs just in virtue of the fact that it gives other citizens private property rights in the instruments by which they may be satisfied. That is the essential point of the discussion of the law against theft of food: since citizens must eat to live, a law that prohibits theft of food must presuppose that all citizens have legal means of acquiring sufficient food to survive, otherwise the law becomes a status crime that violates the rule of law. And since laws against theft are partially constitutive of property rights, it follows that any society that creates private property rights in the resources necessary to meet the basic needs of life must count those resources as part of its social minimum.135

But if the laws presuppose something more, the social minimum must expand. For example, if the laws establish a property qualification for voting, then the state must bring it about that everyone is at least capable of meeting that qualification. Otherwise, it has established a non-general status qualification for the franchise, one that offends against the rule of law, and hence cannot coexist with a legitimate constitutional order. Put differently, a state that establishes a property qualification for voting without making it possible for everyone to comply with that qualification has ejected those who do not qualify from the political community.136

Michelman offers a different argument for finding poverty elimination in the fundamental commitments of constitutional liberalism.137 Rawlsians typically say that the demands of justice are greater than the demands of bare legitimacy (constitutional essentials).138 Michelman cogently explains the Rawlsian case for the justice-legitimacy gap, then offers a convincing argument for being quite skeptical of it.139 Accepting Michelman’s argument, like accepting mine, entails that the commitment to poverty elimination is a condition of a legitimate constitutional order.

Perhaps unsurprisingly in light of that conclusion, on the whole, I agree with Michelman. However, while there might be good reason to reject the justice–legitimacy distinction in the ideal theory framework of Rawls, there is more reason to accept it in the nonideal world in which we live.140 In this world, we cannot realistically expect states to achieve anything like full-scale distributive justice. But there are still important moral reasons to cooperate with less than fully just states while trying to improve them—the perfect ought to not be the enemy of the good. Thus, in our nonideal world, we need to know which shortfalls from the demands of justice permit us to altogether reject a state; those are the conditions of legitimacy.

Because the rule of law, to at least some extent, is basic to a civilized and functional state, this paper shows that some level of poverty elimination can serve as a legitimacy condition in nonideal theory, even if there is good reason to accept the justice–legitimacy gap (and even in nonliberal political communities that do not accept Rawlsian theories of distributive justice). Thus, we can more aggressively demand it of actual states in the real world, and potentially reject, or threaten to reject, the authority of states that do not provide it, even as we may not reject the authority of minimally decent states that fail to attain full-fledged distributive justice. However, the price for this greater scope of applicability is a weakened demand: as further discussed below, there are some economic inequalities that are compatible with the rule of law but not with full-fledged justice.

V.     A Principle Run Out of Control?

The discussion thus far suggests a family of objections related to the seeming capaciousness of the moral demands of the principle of generality. One objection is a worry that it amounts to a theory of all-things-considered justice. “Can’t we say,” this objection might go, “that any unjust social arrangement will not be justifiable by public reasons?”141 Call this the “runaway public reason” objection. It is troubling, within the theoretical framework of this Article, because the rule of law is a principle of basic legitimacy for legal systems, and is supposed to be compatible with societies that are not fully just liberal democracies; part of the aspiration of giving an account of the rule of law is to be able to say something about how even imperfect societies can be more or less legitimate.

A similar objection is that this conception of generality may be overdemanding. Suppose the rule of law commands universal health care. This is a goal that some ostensibly rule-of-law states are unlikely to meet. Must we say such states fail from the rule-of-law standpoint in addition to that of general political morality or justice? And what would that add to the moral understanding of such states?142

A related objection asks why these claims have to be attached to the ideal of the rule of law. I want to claim that those who think themselves committed to the rule of law have also committed themselves to the strong ideas about social equality developed in this Article, but perhaps those who want to resist that argument may simply deny that the strong version of the rule of law is part of what they are committed to. Why not reject generality altogether, or at least treat the strong and the weak version of the rule of law as two independent principles that need not be accepted or rejected together?143 (Call this the independence objection.)

I now answer these objections. The runaway public reason objection is answered if there are realistic social orders that satisfy public reason but are nonetheless unjust. To see that there are, consider that the (weak) version of public reason on which I rely depends on the social meanings of laws within a particular social context, while justice is (presumably) universal. In effect, I propose a social fact thesis about the rule of law similar to the social fact thesis advanced by positivists about law itself: whether a legal-social arrangement is consistent with seeing all in society as equals is a social, not a metaphysical or moral fact. If, in a given society, an unjust social arrangement can be accepted by all citizens consistent with seeing themselves as equals, then the arrangement in question will satisfy public reason.

Such a society is analogous to what Rawls calls a “decent nonliberal people”: a society that is not fully just, but is nonetheless legitimate and worthy of adherence.144 I say “analogous,” because Rawls describes decent societies as hierarchical, but hierarchical societies are ruled out by the principle of generality. A nonhierarchical society can still be unjust if, for example, there are morally unjustifiable distributive inequalities that do not lead to status inequalities. Such a society may be unjustly divided into rich and (moderately, not extremely) poor, yet if the rich nonetheless respect the poor, and have not too egregiously corrupted the laws in the process of their resource hoarding, such that those laws are still compatible with public reason, it may comply with the rule of law. For example, such a “decent unjust society” might choose a tax system that prefers general economic productivity and efficiency over distributive justice, based on the (debatable, but not obviously false) economic theory that holds that lowering taxes on capital will produce general gains; because general productivity counts as a reason for the poor, such a system is unjust but justifiable by public reasons.145

Now, I address the over-demandingness objection. The limit of the distributive demands of the principle of generality is given by the principle that the law must ensure that each has enough to satisfy the presuppositions embedded in the law itself. For example, if the law presupposes that each member of the political community has property (i.e., makes the exercise of some legal rights or avoidance of some legal disabilities depend on property), then the rule of law requires the state to make available to each person enough property to satisfy the presupposition, at least to the extent it can practicably do so and any unavailability is attributable to it.146 But the necessary condition of any rule-of-law demand of this form is that the law itself must generate the presupposition that the state is obliged to satisfy. The rule of law does not demand universal health care in a country like the United States because there are no legal rights or disabilities that presuppose access to health care. Moreover, if the laws accommodate the diversity of resources in the community, the rule of law may permit a more unequal, and potentially more unjust, distribution of resources.147

That begins to answer the over-demandingness objection. But we might worry that the objector still has more resources. “To what extent,” the next step of the objection goes, “does this argument assume that there is some domain of human activity independent of the law? If the law structures the market all the way down, can we not say that all goods that may be acquired through the market are legal goods, and hence that the rule of law demands that all people have access to everything produced by the market?” This sounds like a demand for absolute equality.

This objection erroneously fails to distinguish between relationships structured by or caused by the laws and rights and duties commanded by it. Even if one thinks (as I do) that much of the economic inequality in the United States and the world is attributable to legal structures like corporate limited liability, the free movement of capital across borders combined with the restricted movement of labor, regulatory capture, the legal suppression of labor organizing, and so forth, the demands of the rule of law only come into play when those inequalities are themselves made the basis of legal rights and obligations.

“But,” the objection persistently continues, “property rights themselves are legal rights and obligations. Therefore, access to a given piece of property—that is, a legal right—requires the scrutiny of the principle of generality.” It seems to follow that each individual property endowment in a society must be justified by the principle of public reason—an extraordinarily demanding claim.

To this, I respond that we need to distinguish between a general legal right or obligation and a specific legal entitlement. There is a difference between the right to hold property in general, or the right to vote, or the right to not be put in prison, and the specific entitlement to a discrete piece of property (or, say, to vote in a particular election). This distinction is analogous to the difference between general legislation and the (non-discretionary) application by officials of law to specific cases. The principle of generality speaks only to the former, and it judges the property system as a whole, not each specific instance of its application.148 If the property system complies with the rule of law, we need not examine individual endowments under it, which are simply consequences of the general rules of the system and the actions of private individuals within it.

To be clear, large-scale social phenomena like poverty are features of the system as a whole. To criticize poverty is not to say that the endowment of each poor person is objectionable from the standpoint of the rule of law, but that the overall legal–economic structure of society is objectionable for permitting there to be a caste of poor people and then imposing legal penalties on that caste.149

Finally, I turn to the independence objection. The structure of the problem is as follows. Typically, commentators have thought that the requirement that the law be general is a demand of the rule of law. But, as discussed in Part I, they have thought that this was a formal requirement, involving ideas like law being written in abstract terms or not containing proper names. And there seems to be a natural affinity between the formal conception of generality and the rest of the rule of law, which I have described as the “weak version”: the weak version is distinctively concerned with keeping state officials from abusing their power over citizens, such as by using it to retaliate against those who cross them, and hence forcing citizens to live in fear.150 The formal conception of generality can contribute to that end. For example, the rule against proper names can help keep officials from being able to legislate against their enemies. The demand that officials be subject to the same law as everyone else forbids them from giving themselves the right to ignore the personal and property rights of their fellows.

I have argued that the formal conception of generality is incoherent. I have offered a thicker, substantive conception to replace it, but I have not offered an argument against someone who thinks that the appropriate response to the failure of the formal conception is not to thicken it but to drop the generality principle (the strong version of the rule of law) altogether. Now I will. There is also a close connection between the substantive principle of generality and the weak version of the rule of law. I have already pointed out one dimension of this connection in this Article: the criminalization of poverty not only violates the principle of generality, but it also violates the principle traditionally associated with the weak version of the rule of law that the law must be capable of being followed. This is not a coincidence: all law that creates status crimes will both be impossible to follow and will express the social inferiority of those who are subject to it, since none will be able to understand such law as corresponding to reasons that apply to them except insofar as they see themselves as inherently criminal.

Moreover, the weak version also partially answers a question that the strong version asks. The strong version demands we give public reasons for all legal distinctions, including the fact that officials have special powers that non-officials do not have. And the weak version constrains those powers to ensure that public reasons are, in fact, available for them. There are obvious public reasons for having officials with the power to do things like adjudicate civil disputes and put lawbreakers in jail; it is much more difficult if not impossible to find public reasons for having officials with the power to coerce at whim. In The Rule of Law and Equality, I argued that such unconstrained power carries with it a message that the one who has it is of hierarchically superior status to those who do not have it.151 It carries this message in part because it is not justifiable by public reasons, such that those subject to unconstrained power cannot attribute anything other than hierarchical superiority to those who hold it.152 Anything that violates the weak version also violates the strong version; we can conceive of the weak version as a special case of the strong version.153

Another point of contact between the strong and weak versions of the rule of law goes through the idea of arbitrariness. What it means for a decision, or decision maker, to be arbitrary is quite undertheorized.154 However, one promising candidate for an interpretation of the concept is as a failure of decisions to be independent of decision makers (i.e., judges).155 If nothing about a case changes except the identity of the decision maker, and the result changes, we could call the result arbitrary; we would also suspect that it is a failure of generality.156

I have suggested that the principle of public reason gives us a test for generality of decisions within and across judges just as well as for generality of laws.157 Arbitrariness, in the sense I have just given, may violate the principle of generality, as applied to decisions. The worry about a ruling that varies with the identity of the judge is that the judge is importing some inappropriate reasons, independent of the law, into her decision-making process. Accordingly, the quintessential judicial violation of the principle of generality is the judge who hands out higher sentences to a criminal because of his race, or who throws out a case because she and her spouse had a fight that morning. However, this is also an act of hubris: such a judge expresses that her power is a personal possession, which she is entitled to use to carry out her idiosyncratic preferences or prejudices, without regard to her obligation to have reasons for her decisions consistent with the equality of those over whom she holds power. From this, it can be seen that a judge who issues rulings without public reasons offends both the strong and weak versions of the rule of law.

There is also a historical connection between the strong and the weak versions. One of the earliest demands for legal equality came from the Levellers of seventeenth-century England.158 The Levellers demanded legal equality together with the procedural protections that fall under the weak version of the rule of law, and that we currently associate with due process.159 The connection between these two ideas at the time was obvious: substantive legal inequality went through procedural inequality, as, for example, when commoners were prohibited from prosecuting nobles.160 Equal access to judicial resolution of disputes can be seen as the most basic form of legal equality.161 Not incidentally, the Levellers also went further, in the direction I go in this Article, to include demands for socioeconomic justice, such as free schooling and access to subsistence resources in the commons.162

Finally, the weak and the strong versions of the rule-of-law appeal to the same higher-level normative idea of respect for equals through reason giving. The weak version demands that officials give legal reasons—that is, reasons that can be found in the law—for their use of state power.163 And I have argued that the giving of legal reasons amounts to a kind of respect for the general public.164 The strong version requires the law itself to be consistent with giving reasons that respectfully address the public at large. Both the strong and the weak versions of the rule of law in this way express the same basic idea, in its most abstract form: no use of state coercive power without giving the right (respectful) kinds of reasons for that use.

It is wrong—inconsistent with the value of equality—to use the state’s monopoly of force to coerce someone without being able to offer reasons for that coercion that are consistent with treating the one coerced with respect. Both versions of “the rule of law” denote the set of regulative principles that tell states what they have to do to avoid that wrongness. Similarly, it is wrong because it is inconsistent with the value of freedom to coerce people without giving them some say in the matter; “democracy” denotes the set of regulative principles that tell states what they have to do to avoid that.165 “Distributive justice” denotes the set of regulative principles that tell states how to run a system of economic cooperation for mutual benefit consistent with equality.166 And so forth.

These considerations suggest that when we discuss the weak and the strong versions of the rule of law, we are discussing one thing, not two things. They are the same principle applied to different chronological stages of the law: the strong version to the enactment of law and the use of discretion in its interpretation; the weak version to its execution.

VI.     The Rule of Law for the [Wo]man of the Left

Morton Horwitz has given us a classic statement of the left-wing critique of the rule of law, in the course of criticizing E.P. Thompson’s claim that the establishment of the rule of law was “an unqualified human good”:

I do not see how a Man of the Left can describe the rule of law as “an unqualified human good”! It undoubtedly restrains power, but it also prevents power’s benevolent exercise. It creates formal equality—a not inconsiderable virtue—but it promotes substantive inequality by creating a consciousness that radically separates law from politics, means from ends, processes from outcomes. By promoting procedural justice it enables the shrewd, the calculating, and the wealthy to manipulate its forms to their own advantage. And it ratifies and legitimates an adversarial, competitive, and atomistic conception of human relations.167

Things look pretty grim already for Horwitz’s argument, and for all those on both the left and the right who have thought that the rule of law and socioeconomic (and racial, gender, etc.) justice are incompatible.168 For I have just argued that the rule of law positively demands intervention into a state’s class system, potentially including outright economic redistribution, at least in cases where the property laws and underlying socioeconomic circumstances bring it about that there are some in the community whose poverty is incompatible with the state’s treating them as equal subjects of the law. And it makes this demand without relying on those in power to use it “benevolently,” or requiring the poor to come begging to some official vested with lawless discretion over the well-being of the citizens under his supervision.

Nonetheless, we ought to directly confront Horwitz’s objections. In this Part, I will answer the left critique of the rule of law.169 I start (Part VI.A.) with the notion that the rule of law “creates formal equality” at the cost of “promot[ing] substantive inequality.”170 From there, I consider (Part VI.B.) the proposition that the rule of law “restrains power” only at the cost of “restrain[ing] power’s benevolent exercise.”171 (Most of Horwitz’s other claims need not detain us very long, and I address them in passing, in footnotes, at appropriate places.) Finally, I consider (Part VI.C.) the proposition, advanced by some Marxists, that the rule of law arises from unjust capitalist production relations, and ought to be rejected on those grounds.

A.     Two Concepts of Equality

Egalitarian discourse has tended to revolve around two distinct but related conceptions of equality: equality as identity and equal status. On the conception I will call equality as identity, A and B are equals in some respect if they are identical in that respect. In the distributive justice literature, for example, some have argued that citizens ought to have identical endowments of resources, or that any variations from identical endowment are to be justified by some independent moral idea.172 This also appears in Rawls’s difference principle, the proposition that any differences in resource endowments are permissible only to the extent they are necessary to maximize the endowments of the least well-off.173 The opposite of equality as identity is different treatment, or discrimination.

Equality as identity also goes by the name “formal equality.” We say that people are treated formally equally when the law treats them the same. But this formal inequality can, as Horwitz says, mask substantive inequality, or even prop it up. The law that forbids both the rich and the poor from stealing bread entrenches the underlying inequality of the two classes.

Horwitz’s mistake is to suppose that the rule of law commands formal equality. But I have argued (Part I.A) that the rule of law can make no such command. Instead, the rule of law requires the other kind of equality, equal status. We act in accordance with the demands of equal status when we conceive of people as of identical worth and importance, and then treat them as people of identical worth and importance (“equals”) ought to be treated. Equal status sometimes requires identical treatment. It would be inconsistent with the equal status of all citizens to give some more votes than others. But sometimes it does not. It is consistent (arguably) with the equal status of all citizens to use affirmative action policies to prefer those who have been the victims of injustice. Indeed, equal status may actually require such non-identical treatment. The opposite of equal status is hierarchy, the treatment of some people as of greater or lesser worth than others.174

The academic literature across several disciplines is pervaded by a confusion between these two conceptions. In the political philosophy literature on distributive justice, many had operated under the assumption that the relevant conception of equality was equality as identity, until Anderson pointed out that this conception misses the normative motivation underlying the demand for equal distribution of resources, namely the idea that people ought to be treated as of equal status.175

In the legal literature, this confusion appears prominently in Westen’s oft-cited The Empty Idea of Equality.176 Westen argues—for reasons similar to those given in my critique of the formal conceptions of generality—against the egalitarian demands to treat people identically. He concludes that the very concept of equality is, as the title suggests, empty.177 But Westen misses the work that equal status does in the egalitarian literature. For example, even though Rawls concludes that justice includes a requirement of identical treatment (hence, the difference principle), that theory of justice is at bottom an account of what society is obligated to do in order to treat people as “free and equal persons,” that is, as of equal status.178

At this point in the Article, it should be obvious that the rule of law is an important tool in the fight against substantive inequality.179 I turn now to the benevolent exercise of power.

B.     The Dangers of Power

It is true that the rule of law restrains “power’s benevolent exercise”—the rule of law does not attend to the goals that unconstrained officials might want to pursue, but simply demands that they be kept on a leash. I have argued at length for this proposition, and for its egalitarian bona fides, elsewhere.180 Here, I simply address Horwitz’s specific point.

The first thing to note is the alarming implication that the “benevolent exercise of power” is the way to bring about socioeconomic justice—as if what the critique of class really needs is some kind of universal technocrat or Platonic guardian who can directly observe the condition of everyone in society and give to each according to his or her needs.181 Can Horwitz—or any “Man [or Woman] of the Left”—possibly think that the worst-off ought to have their basic needs met as a matter of the benevolent exercise of power? I take it to be a fundamental commitment of the political left that the very worst-off are to have their basic needs met as a matter of right, not as an act of grace.182

The presupposition seems to be that unconstrained official discretion is necessary to correct unjust class hierarchies and resource distributions. Horwitz shared this narrow vision of the potential of law with many on the right, like Hayek, who thought that contemporary welfare states were incompatible with the rule of law, on the ground that they require broad official discretion.183

Here is the version of the welfare state that the rule of law forbids—and rightly:

      We finally received a letter summoning us to the Welfare Department. . . . A pimply-faced white man in a short-sleeved shirt stood behind the counter chomping on a candy bar. As he approached Miss Dora, he glanced at Mike and me. Then he bombarded her with questions.

“What you doin’ with two white boys?

“Where’s their mother?

“Why doesn’t their daddy work?

“How much money do you make?

“Isn’t twenty-five dollars enough to live on?

“Do all you people think there’s free money up here?

“Why don’t you sell your house to take care of them?

      Feeling a mixture of rage and shame, I stood next to her. My fists clenched below the counter, and I wished I could kill the man. Miss Dora didn’t seek anything for herself. She was willing to be degraded for our benefit. Finally, the clerk disappeared behind a partition. Forty-five minutes later he returned with a check. He slid it quickly onto the counter, carefully avoiding Miss Dora’s touch.

      Outside the office she showed it to me. I couldn’t believe my eyes. Humiliated for a mere five dollars and fifty cents!

      . . . .

      The following week we returned to the welfare office. We waited for almost an hour. Finally, the same administrator approached. He said simply, “No check!”184

That grim scene violates the rule of law not because the system it reveals was meant to remedy social inequality, but because it perpetuated that inequality by subjecting society’s worst-off to arbitrary administrators who used their unconstrained power to humiliate and degrade.185 This rule-of-law critique of the welfare state belongs to the left, not the right.186

But redistribution need not work that way. Goldberg v. Kelly showed that welfare officials can be controlled by law.187 And liberal theorists have produced countless ways of satisfying the demand to improve the condition of the worst-off without demanding the kind of constant discretionary meddling in people’s lives exemplified by the worst of the welfare state.188 Welfare laws look very different when the officials who administer them are constrained to not use their powers disrespectfully or arbitrarily. Then, they can support the rule of law, as a means to preserve the status of equals for all, even the poorest.189

As it turns out, Marx himself, at least in his early years, recognized the relationship between the rule of law and substantive equality. In 1842, Marx criticized the Prussian censorship laws in rule-of-law terms, claiming that such “laws without objective norms, are laws of terrorism, such as those created by Robespierre” and “positive sanctions of lawlessness.”190 Going on, he criticized the law as “an insult to the honor of the citizen” and “a mockery directed against my existence,” in virtue of the fact that it is “not a law of the state for the citizenry, but a law of a party against another party” which “cancels the equality of the citizens before the law.”191

In the same year, in an essay against a law criminalizing the gathering of fallen wood, Marx criticizes the customary rights of the elite, which he says are “formations of lawlessness” because “their content is contrary to the form of law—universality and necessity,” such that they must be “abolished and even punished.”192 He then scolds the assembly for “represent[ing] a definite particular interest” and “trampl[ing] the law under foot,” memorably declaring that “[p]rivate interest is no more made capable of legislating by being installed on the throne of the legislator than a mute is made capable of speech by being given an enormously long speaking-trumpet.”193

C.     The Genetic Fallacy in the Marxist Critique of the Rule of Law

Perhaps, in light of what the young Marx said about the rule of law, contemporary Marxists have just lost their way? But here the Marxists might raise an objection. Marx’s objections to the Prussian government’s rule-of-law violations were embedded in a capitalist society; perhaps the rule of law is an appropriate ethic for a capitalist mode of production but it is inappropriate for (ex hypothesi) more advanced economic relations.

This is what Robert Fine thought. On his interpretation, the late Marx turned into a critic of the rule of law.194 And he argues that this turn was correct, because the rule of law is an artifact of the capitalist mode of production—its function is to support capitalist property rights, and it arises out of a conception of the individual as property-owner in the first instance.195 He concludes that the rule of law is inseparable from capitalism.196

But that conclusion does not follow. It may be that capitalism creates the rule of law, but that the rule of law can exist, and can serve a different function (such as protecting citizens from officials who might otherwise be tempted to abuse their power) even under communism. (The chicken comes from the egg, but we can transpose it to a world completely devoid of eggs, and we can still identify it as a chicken and find a use for it.)197

Conclusion: In Praise of “Mischievous Results”

One law for the lion and for the ox is indeed oppression, if that law requires everyone to eat meat or to grow a mane. And in a world where some people have a lot, and others are destitute, treating people how equals ought to be treated means taking from the former and giving to the latter.

As it turns out, this is the correct interpretation not just of the ideal of distributive justice, but of the proposition that all are to be equal under the law. And this is a valuable discovery independent of distributive justice theory. Many who reject most of egalitarian distributive justice accept the rule of law and endorse the principle of equality under law. Indeed, it is often seen as a fundamental principle of the modern capitalist economy.198 This Article shows that they, too, are committed to at least some of the social reforms to which distributive justice egalitarians are committed.199

Of course, those who see themselves as committed to the rule of law, but not to the claims about socioeconomic justice that I have made in this Article, have some potential answers to this argument, but those answers are not as robust as they might seem. They might deny that I have the correct conception of what generality requires. But I take the argument I offered in the first Part to show not just that the public reason conception of generality is viable and plausible, or a reasonable alternative to the conventional formal conception that currently dominates the literature, but that the formal conception is incoherent. It is not within the realm of reasonable disagreement.

If that is right, the onus is on anyone who would deny that generality forbids states to allow avoidable poverty to develop an alternative non-formal conception of what generality requires. Alternatively, they might deny that the rule of law requires generality. Some have done so; the most prominent is Raz.200 However, I have already argued that the principle of generality is closely tied into the rule of law as a whole (Part V). And that is not the only cost of Raz’s move: the idea that law must be general is a deeply rooted part of our intellectual and moral culture, as well as our understanding of constitutional ideas that, in the United States, go under the names of equal protection and due process.201 To sacrifice that idea is to bite a very large bullet.

The demanding claims I have made about what legal equality requires are not wholly foreign to American jurisprudence. As I noted earlier, the Supreme Court’s use of a tiered levels of scrutiny framework in its equal protection jurisprudence tracks, and can be justified by, the public reason conception of generality.202 And while the Court has (erroneously, in my opinion, for the reasons given above) rejected the notion that socioeconomic class is a suspect classification,203 it has flirted with the ideas expressed here in at least one case. In Douglas v. California, the Court held that to deny indigent defendants publicly funded counsel on appeal was unconstitutional not because it violated the defendants’ Sixth Amendment right to counsel, or procedural due process, but because it violated equal protection.204 Justice Douglas, writing for the court, pointed out:

There is lacking that equality demanded by the Fourteenth Amendment where the rich man, who appeals as of right, enjoys the benefit of counsel’s examination into the record, research of the law, and marshalling of arguments on his behalf, while the indigent, already burdened by a preliminary determination that his case is without merit, is forced to shift for himself. The indigent, where the record is unclear or the errors are hidden, has only the right to a meaningless ritual, while the rich man has a meaningful appeal.205

Striking in this passage is that the Court looked beneath the formal equality of the law, which provided an appeal for rich and poor alike, and attended to—dare I say it?—the disparate impact of that law, when taken in the context of underlying socioeconomic inequality. Had Justice Douglas read (and agreed with) this Article, I imagine him further reasoning that in a world in which all had the resources to litigate their appeals, the state’s policy of not providing counsel on appeal would be permissible. But in a world in which this is not the case, and in which the state is partly responsible for that fact (e.g., because of the way the legal system is structured to require the assistance of counsel to have a fighting chance in the court of appeals), the policy is not permissible. The broad language of Douglas’s majority opinion captures the core political intuition of the latter half of this Article: that the state has no business propping up socioeconomic inequality with its laws.

Unsurprisingly, given the broad implications of the majority opinion in Douglas v. California, the other Justices recognized the dangers it posed to socioeconomic inequality. Justice Harlan, joined in dissent by Justice Stewart, declared that “the Equal Protection Clause is not apposite, and its application to cases like the present one can lead only to mischievous results.”206 The “mischievous results” he feared are essentially the conclusions of this Article:

      The States, of course, are prohibited by the Equal Protection Clause from discriminating between “rich” and “poor” as such in the formulation and application of their laws. But it is a far different thing to suggest that this provision prevents the State from adopting a law of general applicability that may affect the poor more harshly than it does the rich, or, on the other hand, from making some effort to redress economic imbalances while not eliminating them entirely.

      Every financial exaction which the State imposes on a uniform basis is more easily satisfied by the well-to-do than by the indigent. Yet I take it that no one would dispute the constitutional power of the State to . . . . impose a standard fine for criminal violations, or to establish minimum bail for various categories of offenses. Nor could it be contended that the State may not classify as crimes acts which the poor are more likely to commit than are the rich . . . .

      Laws such as these do not deny equal protection to the less fortunate for one essential reason: the Equal Protection Clause does not impose on the States “an affirmative duty to lift the handicaps flowing from differences in economic circumstances.” To so construe it would be to read into the Constitution a philosophy of leveling that would be foreign to many of our basic concepts of the proper relations between government and society. The State may have a moral obligation to eliminate the evils of poverty, but it is not required by the Equal Protection Clause to give to some whatever others can afford.207

I write here in defense of the “mischievous results.” Although, like so many Warren Court precedents, the reasoning in Douglas v. California has not been carried through in other cases, Justice Douglas gave us a map to a genuinely equal understanding of the Equal Protection Clause, and, with it, the rule-of-law demand of generality. Justice Douglas and E.P. Thompson were right. Horwitz, Harlan, and Hayek were wrong. The rule of law is both an unqualified human good and a tool in the fight against social injustice.

  1. William Blake, Visions of the Daughters of Albion 9 (Robert N. Essick ed.,
    Huntington Library & Art Gallery 2002) (1793).

  2. [2]. William Blake, The Marriage of Heaven and Hell 83 (Michael Phillips ed., Bodleian Library, Univ. of Oxford 2011) (1793).

  3. [3]. See generally Paul Gowder, The Rule of Law and Equality, 32 Law & Phil. 565, 603–04 (2013) (summarizing literature on generality).

  4. [4]. Derrick Bell, Racial Realism, 24 Conn. L. Rev. 363, 369 (1992).

  5. [5]. Morton J. Horwitz, The Rule of Law: An Unqualified Human Good?, 86 Yale L.J. 561, 566 (1977) (book review) (emphasis omitted).

  6. [6]. Catharine A. MacKinnon, Reflections on Sex Equality Under Law, 100 Yale L.J. 1281, 1290–91 (1991); see also Toni M. Massaro, Empathy, Legal Storytelling, and the Rule of Law: New Words, Old Wounds?, 87 Mich. L. Rev. 2099, 2100 (1989) (providing another feminist critique).

  7. [7]. Robin L. West, Re-imagining Justice: Progressive Interpretations of Formal Equality, Rights, and the Rule of Law 5 (2003).

  8. [8]. Jean Jacques Rousseau, What Is the Origin of Inequality Among Men, and Is It Authorised by Natural Law?, pt. II (G.D.H. Cole trans.) (1754), available at

  9. [9]. See Brian Z. Tamanaha, On the Rule of Law: History, Politics, Theory 1–2 (2004) (recounting universal acceptance of rule-of-law ideal); Jonathan Rose, The Rule of Law in the Western World: An Overview, 35 J. Soc. Phil. 457, 464–66 (2004) (discussing the special relevance of the rule of law for liberalism); Jeremy Waldron, The Rule of Law in Contemporary Liberal Theory, 2 Ratio Juris 79 (1989) (same).

  10. [10]. See generally Ronald A. Cass, The Rule of Law in America 23–33 (2001) (discussing traditional American commitment to the rule of law). In U.S. law, procedural due process captures part of what I have called, in The Rule of Law and Equality, the principles of regularity and publicity. See generally Gowder, supra note 3; see also infra Part I.A (discussing equal protection and its relationship to the principle of generality).

  11. [11]. See generally A.V. Dicey, Introduction to the Study of the Law of the Constitution 116–22 (reprt., 8th ed. 1915) (arguing that the rule of law is distinctively secure in England because of its close connection to common law); Steven G. Calabresi, The Historical Origins of the Rule of Law in the American Constitutional Order, 28 Harv. J.L. & Pub. Pol’y. 273, 276 (2004) (tracing U.S. commitment to the tradition of Magna Carta).

  12. [12]. West, supra note 7, at 118–19.

  13. [13]. See Frank I. Michelman, Poverty in Liberalism: A Comment on the Constitutional Essentials, 60 Drake L. Rev. 1001 (2012); see also Laurence H. Tribe, Revisiting the Rule of Law, 64 N.Y.U. L. Rev. 726 (1989) (arguing against formal conceptions of the rule of law that fail to attend to substantive inequalities).

  14. [14]. The starting point for that conception is The Rule of Law and Equality. This Article is step two in a project of which The Rule of Law and Equality was step one.

  15. [15]. By “sufficientarian,” I refer to those positions in the literature on distributive justice that hold that justice demands that everyone in society have enough to satisfy some specified standard. Sufficientarianism is contrasted with distributive egalitarianism (which itself is distinct from the legal egalitarianism that drives this Article’s account of the rule of law), which holds that justice demands that everyone in society have equal resources absent some special justification. Egalitarianism is the dominant position in the political philosophy literature on distributive justice, associated primarily with those working in the tradition of John Rawls. The classic defense of sufficientarianism is Harry Frankfurt, Equality as a Moral Ideal, 98 Ethics 21 (1987).

  16. [16]. See, e.g., Horwitz, supra note 5, at 566.

  17. [17]. See, e.g., D. Neil MacCormick, Spontaneous Order and the Rule of Law: Some Problems, 2 Ratio Juris 41 (1989) (criticizing (an interpretation of) Hayek’s right-wing rule-of-law critique of social welfare states).

  18. [18]. Before proceeding to the argument proper, a note about methodology. The theoretical part of the argument in Parts I and III and the concrete implications in Parts two and four have a bidirectional relationship. The theoretical claims ground the practical claims in the ordinary fashion, but the practical claims also support the theoretical claims: one reason to accept an abstract normative argument is that it coincides with our pretheoretic judgments about concrete cases. See John Rawls, A Theory of Justice 42–45 (rev. ed. 1999) (explaining the process of reflective equilibrium, by which theoretical judgments are tested against considered judgments of particular cases, and vice versa).

                    The claims of Parts II and IV develop, respectively, the pretheoretic judgments that there was something distinctively legally wrong about the literacy tests, and that there is still something distinctively legally wrong about the criminalization of poverty. To the extent the reader shares those judgments, the arguments in Parts I and III are valuable not just to the extent they satisfy the ordinary criteria for the success of normative arguments, but also because they coincide with and help explain those pretheoretic judgments. (By saying some
  19. [19]. Gowder, supra note 3.

  20. [20]. Id. at 574–82.

  21. [21]. Id. at 574–601.

  22. [22]. Id. at 574–85.

  23. [23]. See generally id. at 576.

  24. [24]. Id. at 574, 595–96.

  25. [25]. Cf. Owen M. Fiss, Groups and the Equal Protection Clause, 5 Phil. & Pub. Aff. 107 (1976) (distinguishing between “antidiscrimination” and “group-disadvantaging” interpretations of equal protection).

  26. [26]. See, e.g., Lon L. Fuller, The Morality of Law 47 (rev. ed. 1969) (describing the version of the principle of generality that forbids law with proper names); id. at 51 n.10 (describing the motivation for the bill of attainder clause in the U.S. Constitution as a commitment to the principle of generality).

  27. [27]. To simplify somewhat: a rigid designator is a designator that refers to the same entity in all possible worlds. By contrast, a non-rigid designator could mean something different in some other possible world; a quintessential non-rigid designator is a label like “the President of the United States”—it so happens that the President is Barack Obama, but that need not be the case. By contrast, it is necessarily true that Barack Obama, or “that person giving the State of the Union Address right now,” is Barack Obama. An indexical is, basically, a pronoun. Some philosophers argue that an indexical is a rigid designator when used, i.e., that the phrase “you go away” or “I want a pony” means the same thing in all possible worlds, essentially because the content of the utterance is defined by its immediate use. See generally Jason Stanley, Names and Rigid Designation, in A Companion to the Philosophy of Language 555 (Bob Hale & Crispin Wright eds., 1997) (explaining concept of a rigid designator and relationship to proper names); David Braun, Indexicals, Stan. Encyclopedia Phil.,
    indexicals/ (last visited Jan. 21, 2014).

                    The relevance of these concepts for rule-of-law purposes is that I take the prohibition against law containing proper names to be an attempt to prohibit special law for (or against) some specific people, but there are ways to refer to specific people other than by name, i.e., by some other rigid designator. To put a rigid designator in a law is to directly and specifically regulate the person so identified; by contrast, to put a non-rigid designator in the law—even one like “the President of the United States,” which only picks out one person—is to regulate
  28. [28]. F.A. Hayek, The Constitution of Liberty 152–53, 209 (1960) (explaining that those who enact general laws do not specifically know to whom those laws will apply).

  29. [29]. See, e.g., id. at 149–55 (giving an “abstractness” conception of generality).

  30. [30]. H.L.A. Hart, Positivism and the Separation of Law and Morals, 71 Harv. L. Rev. 593, 624 (1958); see, e.g., Rawls, supra note 18, at 237 (also giving “like cases alike” conceptions of generality). See also F.S. Royster Guano Co. v. Virginia, 253 U.S. 412, 415 (1920) (“[T]he classification must be reasonable, not arbitrary, and must rest upon some ground of difference having a fair and substantial relation to the object of the legislation, so that all persons similarly circumstanced shall be treated alike.”).

  31. [31]. When I say that the rule of law “must” entail those further prohibitions, I mean that if it does not do so, it is irrational and not worthy of endorsement.

  32. [32]. Two descriptions are extensionally equivalent when they mean (denote) different things but amount to the same thing. For example, “the occupant of 1600 Pennsylvania Avenue” and “the President of the United States” are extensionally equivalent. See generally Melvin Fitting, Intensional Logic, Stan. Encyclopedia Phil.,
    logic–intensional (last updated Jan. 27, 2011) (explaining difference between intensionality and extensionality).

                    I take a non-rigid designator that is extensionally equivalent to a rigid designator to be what Rawls meant by rejecting “rigged definite descriptions.” Rawls,
                    A slightly more direct way of killing off the minimal formal conception would be to show that all descriptions are formally equivalent to some rigid designator or set of rigid designators. Intuitively, this is true, but I am not aware of a proof.
  33. [33]. One might argue, in favor of the minimal conception, that it forbids legislatures from targeting people, e.g., out of malice. However, the minimal conception is insufficient to do this: even were it coherent, it might prohibit some official from enacting a malicious law against one person, but might not prohibit someone from enacting a malicious law against a whole disfavored class (such as anti-Semitic laws).

  34. [34]. See supra note 30 and accompanying text.

  35. [35]. See Gowder, supra note 3, at 604–05.

  36. [36]. This is a well-known philosophical problem. Quine describes it as one of defining similarity over an infinite set of possible kinds (like “red things” or “round things”). See W.V. Quine, Natural Kinds, in Ontological Relativity and Other Essays 118 (1969).

  37. [37]. See West, supra note 7, at 121–24 (describing similar critique of the similarity conception from the critical legal studies movement).

  38. [38]. Hayek, supra note 28, at 314–15.

  39. [39]. Hayek points out that, under the formal conception of generality, officials have to be subject to the same law as everyone else. Id. at 155. This seems intuitively important, and might make up a defense of the formal conception, except that it is impossible. Officials cannot be subject to the same law as everyone else, because they have official powers, given by law, which others lack. Since they must be subject to some different legal rules from the rest of us, we must have some non-formal criterion by which we can pick out which differences are acceptable and which are not.

                    We might say that officials are subject to special legal rules only while carrying out their public responsibilities, and subject to the ordinary law off the job. (I thank Elizabeth Anderson for suggesting this objection.) However, note that there is still a question as to which special rules may permissibly be applied to officials, and that requires some substantive value judgments. We typically only give officials the exemptions from ordinary law necessary to do their jobs; a law exempting police officials from the law against theft while on the job, for example, seems to me to be manifestly non-general just because it bears no relationship to policing. This intuition makes sense only in light of the theory of generality that I am about to give, according to which such exemptions are permitted to the extent they are justifiable by public reasons, here, the social importance of carrying out police functions.
  40. [40]. See Gowder, supra note 3, at 605–07.

  41. [41]. See id.; see also West, supra note 7, at 144 (describing prior attempts to resolve the incoherence of formal equality as “aim[ing] to provide a rule of decision that will enable judges to resolve questionable cases by reference to moral ideals themselves identified with the idea of law”).

  42. [42]. John Rawls, The Idea of Public Reason Revisited, in The Law of Peoples 136–37 (1999); see also Henry Sidgwick, The Methods of Ethics, 266–67 (1962) (arguing that the law may distinguish between classes of application, but not arbitrarily, where an arbitrary act is one “for which no sufficient reason can be given”); Gowder, supra note 3, at 605–11.

                    Also note here that there is no difference between saying “a law must be justifiable by public reasons,” and “the distinctions in a law (between cases and people) must be justifiable by public reasons.” At the highest level of abstraction, a law is nothing but distinctions; the work of a law is to carve out some situations, people, or acts, and declare that they have some legal status conferred on them distinct from other situations, people, and acts. (This, of course, is part of why any formal conception of generality is doomed from the start.)
  43. [43]. See Lawrence B. Solum, Constructing an Ideal of Public Reason, 30 San Diego L. Rev. 729, 738 (1993) (noting that the rule of law is respected when “laws and regulations are addressed to the public at large,” and suggesting that public reason captures this ambition to universal address).

  44. [44]. John Finnis, On “Public Reason” 19–20 (Mar. 26, 2005) (unpublished manuscript), available at

  45. [45]. West, supra note 7, at 149–50.

  46. [46]. See generally Steven Wall, On Justificatory Liberalism, 9 Pol., Phil. & Econ. 123 (2010) (explaining that a “justificatory liberal” is a liberal who thinks that the freedom and equality of citizens in a liberal democracy is to be protected in the first instance by constraints on the sorts of reasons that might be offered for a state’s laws and basic institutions, i.e., that they are to be limited to public reasons on some conception).

  47. [47]. See Gowder, supra note 3, at 570–73; see also Daniel H. Cole, ‘An Unqualified Human Good’: E.P. Thompson and the Rule of Law, 28 J.L. & Soc’y 177, 197–98 (2001) (describing historical independence of the rule of law from liberal democracy).

  48. [48]. Thomas Hobbes, Leviathan 237–38 (Richard Tuck ed., Cambridge Univ. Press 1996) (1651); see also David Dyzenhaus, Hobbes’s Constitutional Theory, in Leviathan 453, 461 (Ian Shapiro ed., 2010) (explaining that, for Hobbes, the laws “pretend the benefit of the individuals commanded”).

  49. [49]. Moreover, for use in an ideal like the rule of law, minimalism is appropriate: the rule of law is conventionally conceived of as a formal criterion for basically legitimate government, distinct from more capacious theories of democracy or justice. Although a formal conception of generality, as it turns out, does not work, we still ought to remain as faithful as possible to the pretheoretic understanding of what the rule of law is, and hence import as few substantive philosophical demands as possible. See Gowder, supra note 3, at 570–73 (for a defense of this approach).

  50. [50]. John Rawls, Political Liberalism 224–25 (1996).

  51. [51]. Critics might object that the original position, as given by Rawls, is meant to model the idea of what free and equal citizens can accept, and itself leads to the full-fledged ideal of public reason as given by Rawls. But that’s not quite right: the original position models free and equal citizens in a liberal democracy, the notion of a free and equal citizen might mean something else altogether in a nonliberal or nondemocratic society. See id. at 34–35. I thank Marcus Arvan for pressing me to answer this point.

  52. [52]. Elizabeth S. Anderson & Richard H. Pildes, Expressive Theories of Law: A General Restatement, 148 U. Pa. L. Rev. 1503, 1508 (2000) (“[A]n expressive theory . . . evaluates actions in terms of how well they express certain intentions, attitudes, or other mental states.”).

  53. [53]. Rawls, supra note 42, at 136–37.

  54. [54]. Rawls’s own elaboration of the requirement tends in this direction. For example, he suggests that controversial economic theories are not within public reason. Id. at 224–25.

  55. [55]. I borrow the phrase from Thomas Nagel, The View From Nowhere (1986).

  56. [56]. For rule-of-law purposes, “citizen” does not just mean those whom the state counts as members of the political community, but includes those whom the state coerces more generally. For more on those whom the rule of law protects, see Gowder, supra note 3, at 573–74.

                    Also, when I say “free and equal citizen” here, I do not mean to suggest that the principle of generality protects citizens’ freedom in the sense claimed by traditional rule-of-law scholars. (I think that claim is false, but space prohibits developing the point here.) Rather, a reason that can be offered to a free and equal citizen is one that a citizen may freely accept (whether or not she is free in some broader political or metaphysical sense), conceiving of herself as an equal. Put differently, a citizen who does not accept a given set of nonpublic reasons cannot both see herself as a social equal and freely accept laws justified under those reasons. I thank Marcus Arvan for urging me to clarify this point.
  57. [57]. See Elizabeth Anderson, Value in Ethics and Economics 22 (1993) (explaining that on expressive theories, “[a]ctions are ranked according to how well they express our rational valuations”).

  58. [58]. Id. at 18.

                    It is important to distinguish the philosophical theories on expressive meanings of law (and otherwise), associated primarily with Anderson’s work, from the expressivist theories in the “law and economics” tradition. McAdams has produced the most important work in the latter category.
                    I have no quarrel with the claim that law serves such functions. But it would be a serious error to limit the idea of the expressive content of law to that information that permits people to predict one another’s behavior.
  59. [59]. The possibility of making such a determination is the key point in favor of understanding public reason as expressive: it gives one some social facts on which to hang one’s evaluative hat in determining the extent to which a given reason is public.

                    There is also a deeper reason that I make this philosophical move. If public reason is supposed to be the reason of equal citizens, then it must be expressive, because equality itself is expressive. That is, the morally important kind of equality is an expressive kind, the sort of equality captured by the notion that we treat one another with the respect due to people conceived of as having the status of equals.
  60. [60]. At any rate, the legislature need not state its reasons for enacting a law, and different members of a legislature may support a law for different reasons. (Or legislators may utter sham reasons to disguise wicked or politically divisive intentions.) Under such circumstances, the attribution of reasons for some enactment will unavoidably be constructive: we actually attribute reasons to the legislature, we do not try to guess the beliefs and values held by individual legislators.

                    However, public reasons must be able to justify the actual law enacted; in order to do so, it must be possible to plausibly say that the reasons under consideration are the actual reasons for the law (sham reasons or insincere reasons are not justifying). This need not require that any actual legislator hold the reasons in question, just that it must be possible to say with a straight face that they did. For an example of a law–reason pair that fails this requirement, see
  61. [61]. See generally Bernard Williams, Internal and External Reasons, in Moral Luck (1981).

  62. [62]. Rawls, supra note 50, at 36; see also Joshua Cohen, Philosophy, Politics, Democracy 54 (2009) (explaining that political appeals to comprehensive reasons appear, to nonadherents, as “simply appealing to what we believe”).

  63. [63]. See discussion infra Part V. This is what I describe, in that Part of this Article, as a “decent nonliberal state.” Space does not permit a more fleshed-out account of the decent nonliberal state here. I hasten to note that, while we might describe the demand to treat others equally as an agent-independent reason rooted in the moral value of equal treatment, what matters for this argument it that it has an agent-relative description, and thus can be discovered by discovering the set of agent-relative reasons.

  64. [64]. See generally Heidi M. Hurd, Expressing Doubts About Expressivism, 2005 U. Chi. Legal F. 405, 422–28 (criticizing expressive theories that depend on conventional meanings—like the one given in this Article—as both “under- and over-determinate”). I aim, in the next Subpart, to answer Hurd’s criticisms.

  65. [65]. Simon Blackburn, Group Minds and Expressive Harm, 60 Md. L. Rev. 467 (2001).

  66. [66]. Id. at 483–84.

  67. [67]. A helpful recent discussion of the extent to which “cultural facts” are determinate is Tarunabh Khaitan, Dignity as an Expressive Norm: Neither Vacuous Nor a Panacea, 32 Oxford J. Legal Stud. 1, 9–13 (2012).

  68. [68]. This point was first noticed by C. Edwin Baker, Injustice and the Normative Nature of Meaning, 60 Md. L. Rev. 578, 593 (2001), who also makes similar points about what I call the conventional nature of legal meanings. This Article further develops a number of ideas first gestured at in broad brush strokes by Baker.

  69. [69]. I take this to be an axiom of legitimate governance. We call a law that is directed at some private end “corruption”; we call a law that is directed at no end at all “irrational.” Even Hobbes, long seen as a defender of absolutism, declares that the sovereign is bound by the public good. See Hobbes, supra note 48, at 231; Dyzenhaus, supra note 48, at 461 (explaining Hobbes).

  70. [70]. Because this is a constructive account of the expressive content of a law, it does not matter that real-world legislators often have corrupt or irrational reasons for the laws they enact. I thank Aziz Huq for pressing me to clarify this point.

  71. [71]. Here, I follow, and basically accept, Raz’s account of law’s claim to authority. For Raz, the law necessarily claims authority. Joseph Raz, The Authority of Law: Essays on Law and Morality 29–30 (1979). And authority, in turn, is justified if it helps those subject to it comply with reasons that already apply to them. Joseph Raz, The Morality of Freedom 53 (1986). For example, the law against running red lights helps people avoid crashes, which they have independent reason to do. The law against murder commands people to do that which they already have moral reason to do.

  72. [72]. On the notion of laws being enacted in citizens’ names, see generally, Thomas Nagel, The Problem of Global Justice, 33 Phil. & Pub. Aff. 113, 128–29 (2005) (arguing that sovereign states act in their citizens’ names, and impose moral responsibility for their acts on their citizens). Nagel’s argument, on its terms, is not limited to democracies, and it has long been suggested that even nondemocracies act in the name of their citizens. See, e.g., Hobbes, supra note 48, at 122 (arguing subjects even of a monarch are deemed authors of monarch’s acts).

  73. [73]. This truth, of course, is embedded deeply into our constitutional ideas. Thus, the Supreme Court applies rational basis review to every legislative act, even those not impinging on a fundamental right or protected class. See generally City of Cleburne, Tex. v. Cleburne Living Ctr., Inc., 473 U.S. 432 (1985). But the idea goes deeper: positive law is a purpose-driven activity (the legislature is trying to bring something about by its enactments), and purpose-driven activity is under a rational requirement to take the means necessary to achieve its ends. See Immanuel Kant, Groundwork for the Metaphysics of Morals 31–34 (Allen Wood ed., Yale Univ. Press 2002) (1785).

  74. [74]. Radical interpretation is set out in Donald Davidson, Inquiries into Truth and Interpretation ch. 9–10 (2d ed. 2001); see also David Lewis, Radical Interpretation, 27 Synthese 331 (1974) (providing a generally clearer explanation than Davidson’s exposition).

  75. [75]. The quote, along with the principles of coherence and charity, comes from Donald Davidson, Three Varieties of Knowledge, in Subjective, Intersubjective, Objective 205, 211 (2001).

  76. [76]. These last two features of the instant account make some progress toward answering the worries noted by Steven Smith, Expressivist Jurisprudence and the Depletion of Meaning, 60 Md. L. Rev. 506, 560–62 (2001). Smith worries that by relying on idealized interpretations of the meaning of a law, expressivists simply engage in unbounded fictional speculation. But by building the expressive meaning from a law out of the attribution of reasons and rational beliefs to the holders of the three standpoints, we have concrete questions to ask in order to attribute meanings to a law, and hence are saved from the ungrounded “byzantine disputes” Smith decries.

  77. [77]. Ordinarily, these attitudes will be beliefs, which I take, consistent with the standard philosophical account, to be a variety of propositional attitude. See generally Eric Schwitzgebel, Belief, Stan. Encyclopedia Phil., (last updated Nov. 21, 2010) (describing the standard philosophical account). However, I mean to leave open the possibility that the expressive meaning of a law could comprise some other propositional attitude. Nothing in the argument turns on this.

                    Beliefs might be about social, physical, or normative truths, or some combination of them—a law may require attributions of beliefs about, e.g., economic theories, moral claims, the character traits of particular persons or groups, etc.
                    Multiple sets of attitudes may supply the expressive meaning of a law—for example, we may attribute to the legislature that enacts a regressive tax the inclusive disjunction of hatred of the poor and/or a belief in supply-side economic theory. In such a case, both would count in the candidate reasons by which the law might be justified; if either is public, the law is general.
  78. [78]. See generally Sharon Street, What Is Constructivism in Ethics and Metaethics?, 5 Phil. Compass 363 (2010).

  79. [79]. However, members of the community must be able to give the laws the interpretation claimed for them—the interpretation should not be routinely met with a blank stare of confusion.

  80. [80]. Such meanings are established by convention. See generally David Lewis, Languages and Language, in 7 Minnesota Studies in Philosophical Science 3 (Keith Gunderson ed., 1975) (explaining the relationship between meaning and convention). In the case of symbolic commands embedded into law, the conventions that give meaning to the symbol will also give meaning to the law. If the state commands a symbolic utterance that symbolically means X (like redheads wearing dunce caps symbolically means “redheads are stupid”), that law will mean X simply because people in the community ordinary take the utterance the law commands to have that meaning.

  81. [81]. The consequence of this point is that Hurd’s under-determination criticism of expressivism in law, which takes the multiplicity of potential conventional meanings of an expressive act to be objectionable, does not apply to the argument in this Article. Hurd, supra note 64.

  82. [82]. That being said, it is plausible to think that social meanings will become stable over time. I thank Dustin Buehler for pointing this out. For example, some acts may acquire stable symbolic meaning from the expressive use to which they are put. The yellow star has come to acquire a stable symbolic meaning because of the use the Nazis put it to; if a contemporary state ordered someone to wear a yellow star, that act would now carry the taint of anti-Semitism regardless of the actual purpose of the contemporary legislation. Similarly, thanks to the Klan, there is no innocent way to burn a cross, even on one’s own lawn.

  83. [83]. This point is not limited to utilitarianism, but applies to all forms of moral reasoning in which evaluations of institutions depend on facts about the world. For example, one might think that the laws creating a military draft are permissible only because one’s state has institutions ensuring that it only fights just wars; if those institutions were to go away the draft would become impermissible.

  84. [84]. I thank Enrique Guerra for pressing me to clarify why public reasons are necessary from all three standpoints.

  85. [85]. Conceivably, a legislator might also have enacted the law to accommodate the racist attitudes of white private citizens. However, this amounts to the same idea of inferiority as that given in the text, since it requires the supposition that it is right that the members of the subordinate caste give way to accommodate the distaste held toward them by the dominant caste.

                    We must also reject the reason given for prison segregation raised in
                    Observe that this point further supports my claim that equal protection doctrine loosely tracks (in broad outlines) the requirements of generality.
  86. [86]. Here, I disagree with Hayek, one of whose formulations of the ideal of general law is that a law that carves out different classes of application is acceptable to the extent it is equally acceptable to those inside and outside of the relevant group. Hayek, supra note 28, at 154. Hayek fails to attend to the possibility of false consciousness, leading some to endorse their own social inferiority.

                    This also shows that the rule of law can, in some cases, be stricter than the ideal of democratic autonomy. I take it that some social institution must be acceptable from the standpoint of democracy if it meets with unanimous assent (unanimity may not be necessary for democratic legitimacy, but it surely must be sufficient); this need not be the case for the rule of law.
  87. [87]. Loving v. Virginia, 388 U.S. 1, 7 (1967).

  88. [88]. It may be suggested that this is an objectionably ahistorical interpretation of the law; perhaps when it was enacted the people of Virginia genuinely thought that real harm (e.g., birth defects) would ensue from racially mixed marriage and procreation. I thank Herb Hovenkamp for raising this point. But that’s not a problem for the argument: as noted above, the expressive meaning of a law can change; as it became clear that racial mixing did not cause biological damage, that potentially public reason for the law would become unavailable.

                    Moreover, to the extent any genuine medical worries were behind the anti-miscegenation law when enacted, those worries could nonetheless be insufficient to provide a public reason for the law, to the extent they were rooted in stigmatized representations of black citizens as biologically inferior.
  89. [89]. Deborah Hellman, When Is Discrimination Wrong? 29 (2008).

  90. [90]. See Gowder, supra note 3, at 586–89 (explaining that unconstrained official power is a form of hubris, that is, of arrogant status-claiming).

  91. [91]. See id. at 608.

  92. [92]. See generally Andrew Koppelman, On the Moral Foundations of Legal Expressivism, 60 Md. L. Rev. 777 (explaining moral significance of expressive harm independent of material injury).

  93. [93]. Recently, there has been litigation alleging that Muslims are subject to unusual questioning at the border between the United States and Canada. See Cherri v. Mueller, No. 12-11656, 2013 WL 2558207 (E.D. Mich. June 11, 2013). This real-world example is objectionable for the reasons given in the text.

  94. [94]. Why are race-, gender-, etc.-based laws most likely to express an insult? Foreshadowing: because of nonlegal social facts about race, gender, and the like. See infra Part III.A.

  95. [95]. See generally T.M. Scanlon & Jonathan Dancy, Intention and Permissibility, 74 Proceedings of the Aristotelian Society, Supplementary Volumes 301 (2000).

  96. [96]. See generally J. Morgan Kousser, Expert Witnesses, Rational Choice and the Search for Intent, 5 Const. Comment. 349 (1988). For discussion of a concrete example where some in Congress may have voted for a bill without intending its alleged purposes, see Michael Selmi, Interpreting the Americans with Disabilities Act: Why the Supreme Court Rewrote the Statute, and Why Congress Did Not Care, 76 Geo. Wash. L. Rev. 522, 526–27 (2008).

  97. [97]. See William H. Riker, Liberalism Against Populism: A Confrontation Between the Theory of Democracy and the Theory of Social Choice 233–38 (1982).

  98. [98]. Critics of contemporary racially unjust institutions often are told that they are trying to punish innocent majority-race citizens today for “past discrimination.” But that’s not true—those who complain about contemporary “punishment” for past generations’ discrimination make the error noted in the text—antiracists are not blaming individuals, they are criticizing institutions, independent of individual blame.

  99. [99]. One might think that the conjunction of intent and effect can ground moral condemnation where either alone cannot. But this still cannot answer the twin objections that: (1) it is notoriously hard to sort out the motivations behind a law; and (2) that approach confuses criticism of people with criticism of institutions.

  100. [100]. It may alarm the reader that I say that the law’s distinction between those who were literate and those who were illiterate were unjustifiable by public reasons because they insulted a different category of people, freed slaves. Shouldn’t the relevant groups for testing the generality of a law be those who the law actually picks out? Actually, this is no problem: it is quite plausible to think that a law that distinguishes between groups X and Y thereby insults group Z, particularly when group Z is (as in the case of the literacy tests) strongly correlated with one of the groups named in the statute. This is just a consequence of the fact that descriptions in the text of a statute might be intensionally different but extensionally equivalent, or might have a social meaning referring to a different group from that named in the statute. If a law said “people with Hebrew last names cannot work as lawyers,” we would know it was anti-Semitic even though it does not directly refer to Judaism and might incidentally pick up some non-Jews or miss some Jews.

  101. [101]. The Condorcet Jury Theorem is a mathematical proof that, under some circumstances, the probability of an electorate coming to the correct result in a vote increases with each voter added. David Austen-Smith & Jeffrey S. Banks, Information Aggregation, Rationality, and the Condorcet Jury Theorem, 90 Am. Pol. Sci. Rev. 34 (1996).

  102. [102]. See id. at 43.

  103. [103]. See generally Robert A. Dahl, Democracy and Its Critics (1989) (discussing various justifications for and critiques of democracy); Carole Pateman, Participation and Democratic Theory 43, 56–66 (1970) (defending “participatory society” in which individual democratic participation is radically expanded not only in politics but even in workplaces, largely on civic educational grounds).

  104. [104]. Jason Brennan, The Ethics of Voting 6 (2011). Brennan disclaims the further step of claiming that those who have a moral duty not to vote ought to be disenfranchised. Still, that step is arguably available to the supporter of a literacy test.

  105. [105]. John Stuart Mill, Thoughts on Parliamentary Reform, in The Collected Works of John Stuart Mill, Volume XIX - Essays on Politics and Society Part II 326–28 (John M. Robson ed., 1977), available at

  106. [106]. Id. at 327.

  107. [107]. I borrow this hypothetical letter method of drawing out the expressive implications of public policy from Elizabeth S. Anderson, What is the Point of Equality?, 109 Ethics 287 (1999).

  108. [108]. See, e.g., 1830–1831 N.C. Sess. Laws 11, available at
    document.php?CISOROOT=%2Fp249901coll22&CISOPTR=175790&CISOOP1=exact&REC=0&CISOBOX=figures (prohibiting the teaching of reading or writing to slaves). See generally Thomas G. Sticht, The Rise of the Adult Education and Literacy System in the United States: 1600–2000, in 3 The Annual Review of Adult Learning and Literacy 10 (John Comings et al. eds., 2002) (recounting southern laws against educating slaves).

  109. [109]. See generally Gideon Yaffe, ‘Ought’ Implies ‘Can’ and the Principle of Alternate Possibilities, 59 Analysis 218 (1999).

  110. [110]. In Kantian terms, to demand that someone who suffers from a socially imposed disability sacrifice important interests in order to spare that same society from the consequences of that disability is to disrespectfully use that person’s capacity to respond to reasons as a mere means for the ends of others. It is a form of moral exploitation. See Kant, supra note 73, at 46–47 (“Act so that you use humanity, as much in your own person as in the person of every other, always at the same time as end and never merely as means.”) (citations omitted); Allen W. Wood, Kant’s Ethical Thought 142–55 (1999) (explaining that the Kantian prohibition on treating the humanity in others as a mere means requires that we treat the rational capacities of others with respect).

  111. [111]. See generally Guido Calabresi & Jon T. Hirschoff, Toward a Test for Strict Liability in Torts, 81 Yale L.J. 1055 (1972) (explaining the idea of imposing duties on the least cost avoider).

                    Objection: I seem to be arguing the state avoids the problem of incapable voters at a lower cost than do individuals, but I do not say anything about the costs the state might have to incur to do things like make literacy available to all. This is not a problem for the argument: the sort of community we are concerned with is not the community where nobody is literate, but where literacy is a privilege of higher-caste citizens. The cost at issue is not the cost to create an educational infrastructure, it is the cost to expand the existing infrastructure to not exclude some members of society, which is likely to be somewhat lower.
                    Moreover a society in which nobody got an education would be importantly different from the standpoint of the reasons that apply to its members. Where all are equally illiterate, nobody in particular has a reason not to vote on grounds of illiteracy. And where literacy is limited to a very small elite, like an oligarchy, those who are not literate at least arguably have reason to vote generated by the need to broaden access to political power that overrides competence-related reasons not to vote. The upshot is that this argument is only meaningful in a society in which some very significant part of the population gets an education (white people), while some discrete group (black people) is excluded—that is, in a society in which it is least likely to be unreasonably costly for the community as a whole to provide an education to the excluded groups.
                    This was in fact the case in the nineteenth century United States. Public schools had existed in the United States since the founding, and, as of 1868, “[n]inety-two percent of all Americans . . . lived in states whose constitutions imposed [the duty to provide an education] on state government.” Sara Aronchick Solow & Barry Friedman,
  112. [112]. Actually, those who are illiterate as a result of bad luck may not even have a duty not to vote, if the (low) risk that their votes will do damage is outweighed by the social (as well as individual) benefit of their participation in the system. Nothing in the argument turns on this question.

  113. [113]. Even the law at issue in Loving was only objectionable because whites were seen as superior in society at large, such that a law forbidding other people from marrying them was correctly understood as an expression of that superiority and an attempt to preserve it.

  114. [114]. This is de facto the case today. See Sean F. Reardon et al., Brown Fades: The End of Court-Ordered School Desegregation and the Resegregation of American Public Schools, 31 J. Pol’y Analysis & Mgmt. 876 (2012).

  115. [115]. Loi 2004-228 du 15 mars 2004 encadrant, en application du principe de la

  116. [116]. See supra note 60 (explaining that sham public reasons are not justifying).

  117. [117]. See, e.g., Elaine Sciolino, French Assembly Votes to Ban Religious Symbols in Schools, N.Y. Times (Feb. 11, 2004), (recounting history of Muslim-school conflicts leading up to the law, as well as exceptions for Catholic and other institutions).

  118. [118]. Anatole France, The Red Lily 95 (Frederic Chapman ed., John Lane Co. 1914) (1894).

  119. [119]. See, e.g., Don Mitchell, The Annihilation of Space by Law: The Roots and Implications of Anti-Homeless Laws in the United States, 29 Antipode 303, 312–13 (1997).

  120. [120]. On the expressive facet of criminal sanctions, see the works cited by Hurd, supra note 64, at 417.

  121. [121]. This may sound like a polemic. So it is. Extreme poverty is a great moral evil. Polemic is called for.

  122. [122]. See generally Jeremy Waldron, Homelessness and the Issue of Freedom, 39 UCLA L. Rev. 295 (1991) (for an apt analysis of this problem).

  123. [123]. See Lon L. Fuller, The Morality of Law 39 (rev. ed. 1969). See also Gowder, supra note 3, at 576 n.32. On the account of the rule of law in The Rule of Law and Equality, the requirement that the law be capable of being followed is an implication of the principle that public officials must not have what I have called “open threats” against citizens to punish them whenever the officials desire; a society with de jure or de facto status crimes gives police and prosecutors absolute discretionary authority over those citizens with the forbidden statuses. See id. at 575.

  124. [124]. I thank Mark Osiel for raising this worry. However, even if it is true, surely there are some who are homeless because the economy has treated them poorly, especially in the wake of the recent housing market crisis.

  125. [125]. See Rawls, supra note 18, at 6–7.

  126. [126]. Incidentally, correlations between homelessness and mental illness do not necessarily imply mental illness causes homelessness. Homelessness could cause mental illness. Moreover, even if mental illness causes homelessness, poverty could, in turn, cause mental illness, or exacerbate the effect of mental illness on homelessness (to the extent treatment or social support for psychological treatment is more available to those with wealth, for example). See generally Jeffrey Draine, Mark S. Salzer, Dennis P. Culhane & Trevor R. Hadley, Role of Social Disadvantage in Crime, Joblessness, and Homelessness Among Persons with Serious Mental Illness, 53 Psychiatric Services 565 (2002) (criticizing the empirical literature on the relationship between homelessness and mental illness for failing to take into account the effects of poverty). The same points apply to other alleged causes of homelessness, like substance abuse.

  127. [127]. Another objection that might be raised to this argument is that there are many things that we legitimately criminalize even though there is a sense in which those responsible cannot help it. A kleptomaniac or a pedophile might (arguably) not be able to help stealing or molesting children, but we nonetheless criminalize their behavior, and the rule of law surely cannot forbid that. (Thanks to John Inazu for raising this point.) But this is a problem for all criminal law and has vexed commentators at least since Aristotle. See, e.g., Aristotle, Aristotle’s Nicomachean Ethics III bk. 3 (Robert S. Bartlett & Susan D. Collins trans., Univ. of Chi. Press 2011). I cannot, and need not, resolve the problem here. So long as there are some acts for which we, as a society, are willing to say the criminal is attributively responsible, there is a sense in which we fairly credit that criminal with the power to have done otherwise. See generally Henry E. Allison, Morality and Freedom: Kant’s Reciprocity Thesis, 95 Phil. Rev. 393 (1986) (defending Kantian argument for interdependency between responsibility and free will). Another way to think about the problem is that kleptomaniacs and pedophiles are subject to internal constraints and bad luck only (much like the blind driver). The homeless are subject to external constraints, rooted in the basic structure of the state (e.g., the existence of private property rights) and those make all the difference.

  128. [128]. See, e.g., John Locke, Two Treatises of Government §§ 123–127 (McMaster Univ. Archive of the History of Econ. Thought, 2000) (arguing that the state’s primary end is the protection of property rights).

  129. [129]. See, e.g., John Rawls, Justice as Fairness: A Restatement 138–39 (Erin Kelly ed., 2001) (describing a possible “liberal socialism,” which preserves private property rights in personal property in order to facilitate individual autonomy, but does not permit private property in the means of production).

  130. [130]. Cf. Hobbes, supra note 48, at 100–11 (arguing—though this is something of a simplification—that there is no political right in the absence of coercive enforcement).

  131. [131]. Cf. Steven A. Ramirez, Lawless Capitalism: The Subprime Crisis and the Case for an Economic Rule of Law 184–216 (2013) (asserting that the concept of the rule of law should be expanded to prevent economic despotism). This Article can supply some philosophical flesh for Ramirez’s conclusions.

  132. [132]. Michelman, supra note 13, at 1005.

  133. [133]. Id. at 1010; see also Rawls, supra note 50, at 228–30.

  134. [134]. Rawls, supra note 129, at 162.

  135. [135]. Cf. Locke, supra note 128, ch. 5, at § 27 (permitting original creation of property rights only “where there is enough, and as good, left in common” for others).

  136. [136]. At least, this is so where citizens’ property endowments are at least in part attributable to the state, including to its basic structure (such as the extent of property rights it recognizes).

  137. [137]. This portion of the argument has benefited greatly from informal discussions with Frank Michelman.

  138. [138]. To clarify the concepts: a just state is fully compliant with the demands of political morality, while a legitimate state may be less than fully just, but is sufficient to command the allegiance of its people.

  139. [139]. Michelman, supra note 13, at 1014–21.

  140. [140]. In the jargon of political philosophy, ideal theory is the theory of what obligations we have when everyone else is complying with their obligations, nonideal theory is when they are not. See generally Liam B. Murphy, Moral Demands in Nonideal Theory (2000); Zofia Stemplowska, What’s Ideal About Ideal Theory, 34 Soc. Theory & Prac. 319 (2008).

  141. [141]. I thank Sascha Somek for raising this point.

  142. [142]. I thank Eli Wald for raising this concern.

  143. [143]. I thank Aziz Huq for raising this worry.

  144. [144]. Rawls, supra note 42, at 59–70.

  145. [145]. Such welfarism may not be consistent with full-fledged public reason as elucidated by Rawls. See Lawrence B. Solum, Public Legal Reason, 92 Va. L. Rev. 1449, 1491–93 (2006) (arguing that welfarism is inconsistent with public reason, because it incorporates the sectarian view that preference satisfaction is the correct conception of the good). However, it is consistent with the less demanding version of public reason deployed here. We do not have to think anyone in the community is superior or inferior to accept welfarist justifications like “this law raises the general utility/wealth of the populace as a whole.”

  146. [146]. Obviously, some of the requirements of generality, such as those necessary to effectuate the right of suffrage, must be limited to those who are full-fledged members of the political community (“citizens” in the contemporary American sense, as opposed to “permanent residents” or “transients” and the like). I cannot consider, in this Article, the extent to which the rule of law regulates who must be counted as a member.

  147. [147]. By “accommodate the diversity of resources in the community,” I refer to things like subsidies and fee waivers for the poor which permit them to access opportunities given by the law on terms equal to the rich.

  148. [148]. For more, see Gowder, supra note 3, at 601–02, 609–10 (principle of generality applies to legislation and to discretionary official acts under general laws).

  149. [149]. For a defense of the claim that the principle of generality is concerned, in the first instance, with the abolition of caste, see id. at 611–13.

  150. [150]. For a defense of this proposition, see id. at 594–98.

  151. [151]. With respect to the principle of publicity, which requires officials to not only be constrained by legal rules but also explain their use of the state’s coercive power with reference to public laws that ordinary people can use to defend themselves, secret law or unexplained rulings violate the principle of generality because, from a second-person standpoint, the person called upon to obey cannot distinguish them from mere official whims.

  152. [152]. This is a gloss on the weak version that I did not develop in The Rule of Law and Equality. Gowder, supra note 3.

  153. [153]. In The Rule of Law and Equality, I had the strong version require the satisfaction of the weak version by simple definitional fiat. Id. at 566, 601–02. This paragraph explains why it must work this way. Put differently, the things the weak version demands and prohibits are proper subsets of the things the strong version demands and prohibits.

  154. [154]. Because the concept tends to obscure more than it illuminates, I have thus far avoided using it in developing the egalitarian conception of the rule of law.

  155. [155]. I thank Patricia Broussard for suggesting this interpretation.

  156. [156]. Note that we cannot be sure that a decision is nongeneral just because it fails the test of decision-maker independence—two decision makers may come out to different results because they interpret facts or exercise discretion differently, within a range of reasonable variation; by contrast, a single decision maker may vary in obviously non-general fashions (as by flipping a coin). For this reason, this does not work as a (formal) conception of generality. However, a decision that is not decision-maker independent is at least suspicious, because it raises the worry that a decision maker is making use of idiosyncratic reasons, drawn from his personal preferences or prejudices rather than the body of public reason.

  157. [157]. Gowder, supra note 3, at 609–10.

  158. [158]. See generally Michael Kent Curtis, In Pursuit of Liberty: The Levellers and the American Bill of Rights, 8 Const. Comment. 359 (1991). I thank Elizabeth Anderson for suggesting I discuss the Levellers.

  159. [159]. Id. at 371–74.

  160. [160]. Id. at 373.

  161. [161]. In the other direction, Fred Smith, Due Process, Republicanism and Direct Democracy, N.Y.U. L. Rev. (forthcoming 2014), argues that equality is a procedural due process value—that laws may violate procedural due process (the primary constitutional home for the weak version of the rule of law) in virtue in part of their disregard of the legal equality of the citizens to whom they apply.

  162. [162]. Curtis, supra note 158, at 374.

  163. [163]. Gowder, supra note 3, at 581–89, 598 n.78.

  164. [164]. Id. at 586–89, 598 n.78.

  165. [165]. See generally Jean-Jacques Rousseau, Of the Social Contract and Other Political Writings 70–75 (Christopher Bertram ed., Quintin Hoare trans., Penguin Classics 2012) (1762).

  166. [166]. See generally, Rawls, supra note 18, at 3–6.

  167. [167]. Horwitz, supra note 5, at 566.

  168. [168]. Some still make this claim today. See, e.g., Richard A. Epstein, The Perilous Position of the Rule of Law and the Administrative State, 36 Harv. J.L. & Pub. Pol’y 5 (2013) (arguing that administrative power in general, including that related to social welfare, poses problems for the rule of law).

  169. [169]. A different set of answers to the Marxist critiques of the rule of law can be found in Cole, supra note 47, at 195

  170. [170]. Id.

  171. [171]. Id.

  172. [172]. See generally Richard J. Arneson, Luck Egalitarianism Interpreted and Defended, 32 Phil. Topics 1 (2004).

  173. [173]. Rawls, supra note 18, at 76–80.

  174. [174]. This is a point repeatedly made by Elizabeth Anderson. See, e.g., Elizabeth Anderson, The Imperative of Integration 102–03 (2010).

  175. [175]. Anderson, supra note 107; see also Samuel Scheffler, What is Egalitarianism?, 31 Phil. & Pub. Aff., 5–6 (2003).

  176. [176]. See Peter Westen, The Empty Idea of Equality, 95 Harv. L. Rev. 537, 539–40 (1982). These arguments have also been conducted in the tax literature, with a different vocabulary but similar underlying ideas. See James Repetti & Diane Ring, Horizontal Equity Revisited, 13 Fla. Tax Rev. 135, 135–36 (2012) (describing longstanding arguments in tax scholarship about “horizontal equity,” or the (formal) treating of likes alike, and “vertical equity,” or the treating of different people differently, consistent with the normative demands that their differences generate). Repetti and Ring come very close to the public reason conception of generality when they point out that the idea of horizontal equity might mean a demand that “government should communicate the rationale for different treatment.” Id. at 151.

  177. [177]. A different answer to Westen, that foreshadows the discussion in the first Part of this Article, comes from Simons, who argues that the “treat likes alike” rule “is not empty if it expresses an egalitarian demand for a principled explanation of differences in treatment.” Kenneth W. Simons, The Logic of Egalitarian Norms, 80 B.U. L. Rev. 693, 761 (2000) (internal quotation marks omitted). I agree, and add that what counts as a “principled explanation” is given by the demands of public reason.

  178. [178]. See Rawls, supra note 18, at 12, 441 (“I now turn to the basis of equality, the features of human beings in virtue of which they are to be treated in accordance with the principles of justice.”); see also Rawls, supra note 50, at 29–35; Rawls, supra note 129, at 18–24.

  179. [179]. In this Article, I have not said that the principle of generality requires people to have identical resource endowments. Generality grounds an attack on social class because it demands that the condition of the poor be improved, not that it be made identical to the condition of the rich. In the jargon of political philosophy, I have defended a sufficientarian distribution principle on egalitarian grounds, where those egalitarian grounds are conceived of in the terms of equal status.

  180. [180]. Gowder, supra note 3, at 513–601. Incidentally, see id. at 582–85, which shows that separating processes from outcomes is a major part of the egalitarian virtue of the rule of law, because it enables ordinary people to defend their interests in the legal system.

  181. [181]. Perhaps Horwitz meant to suggest that the rule of law forbids all progressive law, whether or not that law endows officials with unbounded discretion. This seems to be the implication of his claim that rule of law “radically separates law from politics, means from ends, processes from outcomes.” Horwitz, supra note 5, at 566. But for a legal positivist, the content of law can be the tool of politics: so long as the state implements its policies in accordance with the rule of law, the content of those policies is left up to the political community. Horwitz seems to have illegitimately helped himself here to a natural law conception of what law is, or an organic, evolutionary, conception like Hayek’s.Cf. F.A. Hayek, Law, Legislation and Liberty (1976).

  182. [182]. Here, I follow Avishai Margalit, The Decent Society 240 (1996) (“I claim that a society which assists the needy on the basis of their being entitled to the assistance is less humiliating in principle—whatever the application might be—than a society based on benevolence.”).

  183. [183]. Hayek, supra note 28, at 261–62; see also id. at 314–15 (arguing that progressive taxation violates the principle of generality). Note, however, that Hayek’s views were fairly complex. Sometimes, he suggested that some social minimum was permissible. See generally Elizabeth Anderson, Thomas Paine’s “Agrarian Justice” and the Origins of Social Insurance, in Ten Neglected Classics of Philosophy (Eric Schliesser ed.) (forthcoming) (interpreting Hayek’s views on social insurance).

  184. [184]. Gregory Howard Williams, Life on the Color Line: The True Story of a White Boy Who Discovered He Was Black 79–80 (1995).

  185. [185]. Relatedly, the rule of law enables the rich and powerful to manipulate the system, as Horwitz suggests, only under those incorrect conceptions of the rule of law that do not take account of this possibility and generate a demand that the poor and weak be given the tools to defend themselves. See Gowder, supra note 3, at 582–85 (arguing that the rule of law requires legal rules be accessible to individuals). See G.W.F. Hegel, Philosophy of Right 223–24, 240–47 (S.W. Dyde trans., 1996) (1821) (insisting that all citizens be given at least basic access to legal knowledge in order that they may defend themselves against those with such knowledge). On a stronger egalitarian conception of the rule of law, institutions like functional low-income legal services are required. I plan to develop this point in future work.

  186. [186]. At that time, commentators like Harry W. Jones, The Rule of Law and the Welfare State, 58 Colum. L. Rev. 143, 154–56 (1958) were arguing that the law could and should expand to constrain the discretion of welfare officials.

  187. [187]. Goldberg v. Kelly, 397 U.S. 54, 266–68 (1970) (discussing the requirement of a hearing prior to revoking welfare benefits).

  188. [188]. See generally William E. Scheuerman, Between the Norm and the Exception: The Frankfurt School and the Rule of Law 211–17 (1994) (describing a variety of strategies for reconciling welfare provision with prohibition on administrative discretion); Phillippe Van Parijs, A Basic Income for All, Boston Rev., Oct./Nov. 2000 (proposing universal basic income).

  189. [189]. Anderson, supra note 107, at 289 (explaining “democratic equality” demands of distributive justice which are concerned with resource distributions that protect the equal status of democratic citizens); see Margalit, supra note 182, at 222–40 (discussing the ways that the welfare state can alleviate, or, if run badly, promote, the humiliation of the poor).

  190. [190]. Karl Marx, Comments on the Latest Prussian Censorship Instruction, in Writings of the Young Marx on Philosophy and Society 67, 79 (Loyd D. Easton & Kurt H. Guddat eds. & trans., 1967).

  191. [191]. Id. at 80; cf. Gowder, supra note 3, at 589 n.64, 594–98 (arguing that the rule of law serves equality between citizens and officials in part by preventing terror and preserving a separation between officials’ public roles and their private interests).

  192. [192]. Karl Marx, Debates on the Law on Thefts of Wood, in Marx & Engels Collected Works, Volume 1: Karl marx 1835–43, at 231 (Clemens Dutt trans.) (1975).

  193. [193]. Id. at 261; cf. Rousseau, supra note 8, at pt. II

  194. [194]. Robert Fine, Democracy and the Rule of Law: Marx’s Critique of the Legal Form 122–23 (1984).

  195. [195]. Horwitz seems to draw on this idea with his suggestion that the rule of law props up an “atomistic [etc.] conception of human relations.” Horwitz, supra note 5, at 566. But history puts the lie to the atomism claim: I have shown elsewhere that, at the very dawn of the rule of law and democracy, the Athenian democrats saw—accurately—the rule of law as a critical support to their class solidarity in defense against wealthy and powerful aristocrats. See Paul Gowder, Democracy, Solidarity and the Rule of Law: Lessons From Athens, Buff. L. Rev. (forthcoming 2014), available at

  196. [196]. Many, including social scientists, have erroneously assimilated the rule of law to private property protections. See, e.g., Daniel Kaufmann et. al, Aggregating Governance Indicators 30 tbl.1 (The World Bank Dev. Research Grp. Macroeconomics & Growth & World Bank Inst. Governance, Regulation and Fin. Policy Research Working Paper No. 2195, 1999) (counting “ability of state to protect private property” as part of rule of law). For more, see Gowder, supra note 3, at 572.

  197. [197]. Evgeny Pashukanis, a leading early Soviet legal scholar, argued that the rule of law arose from commodity fetishism and capitalist relations of production. He suggested it be replaced, particularly in the criminal context, with unfettered official discretion: “a measure of expediency for the protection of society,” which could vary “according to whether the purpose is the mechanical elimination of the dangerous individual, or his reform.” Evgeny Bronislavovich Pashukanis, The General Theory of Law and Marxism 185–87 (Chris Arthur ed., Barbara Einhorn trans., 2002). Stalin, doubtless without any sense of irony, applied Pashukanis’s theory to Pashukanis. In 1937, he was, as it were, mechanically eliminated by the state security apparatus. Dragan Milovanovic, Introduction to Pashukanis, supra, at vii, x. He was followed a year later by Nikolai Krylenko, a so-called “legal nihilist” who was accused of “the uncritical repeating of the ideas of Pashukanis”; Krylenko was given a twenty minute “trial” and then shot. Id. at xx–xxi. One wonders whether either objected to their treatment.

  198. [198]. See generally Tamanaha, supra note 9, at 4–5.

  199. [199]. This is a move I have borrowed from Pettit, who argues similarly that his neorepublican theory of liberty gives those who reject liberal egalitarianism a reason, based on their endorsement of the idea of freedom, to support distributive justice. See Philip Pettit, Republicanism: A Theory of Freedom and Government 11–12 (1997).

  200. [200]. Joseph Raz, The Rule of Law and Its Virtue, in The Authority of Law: Essays on Law and Morality, supra note 71, at 210, 210–229.

  201. [201]. See, for example, Bi-Metallic Inv. Co. v. State Bd. of Equalization, 239 U.S. 441 (1915), in which Justice Holmes distinguishes those state actions that require procedural due process from those that do not by appealing to the difference between general laws and the differential treatment of individuals.

  202. [202]. Supra text accompanying notes 92–95.

  203. [203]. San Antonio Indep. Sch. Dist. v. Rodriguez, 411 U.S. 1, 18–28 (1973).

  204. [204]. Douglas v. California, 372 U.S. 353, 356–58 (1963).

  205. [205]. Id. at 357–58.

  206. [206]. Id. at 361 (Harlan, J., dissenting).

  207. [207]. Id. at 361–62 (Harlan, J., dissenting) (several of Justice Harlan’s examples omitted).


Professor of Law, Northwestern University Pritzker School of Law.

For extremely helpful feedback, I’d like to thank participants in the Iowa Legal Studies Workshop, the Mid-Atlantic People of Color Legal Scholarship Conference, the Southeast/Southwest People of Color Legal Scholarship Conference, the Arizona State University Legal Scholars Workshop, ClassCrits V, the Midwest Junior Scholars Workshop at Washington University, St. Louis, and Albany Law School’s Sharing Scholarship, Building Teachers workshop. For yeoman’s service in helping this Article out on an individual basis and/or as part of the formal presentations noted above, I would also like to single out Elizabeth Anderson, Marcus Arvan, Patricia Broussard, Dustin Buehler, Enrique Guerra, Peter Halewood, Aziz Huq, John Inazu, Frank Michelman, Muriel Morisey, Eli Wald, too many of my colleagues and friends at The University of Iowa to mention exhaustively, but particularly including Herb Hovenkamp, Angela Onwuachi-Willig, Mark Osiel, and Alexander Somek, and my research assistants Shawn McCullough and Estiven Rojo. Thanks are also due to the editorial staff of the Iowa Law Review for extremely careful yet non-intrusive editing. The genesis of this paper came from a conversation in my grad school days, sometime in 2010 or 2011, when I was discussing my account of the rule-of-law principle of generality with my Ph.D. advisor, Joshua Cohen, and he said “what about class?” This is my belated attempt to answer that question.