I. Justice or Consent?

Often the existence of a theoretical impasse is a sign that a question has been badly framed. In this essay I argue that this is the case with the ongoing debate in liberal political philosophy about the merits and shortcomings of “ideal theory.” The trouble with this debate, as I see it, is that it is driven by two conflicting claims about the proper aims of political philosophy; claims that are laid out at the beginning of John Rawls’s seminal book A Theory of Justice. ^{Footnote 1} The first claim is that, as Rawls puts it, “justice is the first virtue of social systems, as truth is of systems of thought”; that just as “a theory however elegant and economical must be rejected or revised if it is untrue; likewise laws and institutions no matter how efficient and well-arranged must be reformed or abolished if they are unjust.” Justice, Rawls concludes, is an “uncompromising” virtue. Call this the “justice first” condition, or justice first for short. The second claim is that a just society is one in which “everyone accepts and knows that the others accept the same principles of justice,” and in which “the basic social institutions generally satisfy and are generally known to satisfy these principles.” ^{Footnote 2} This claim underwrites the “liberal principle of legitimacy” that became the focus of Rawls’s later work; the claim that “our exercise of political power is fully proper only when it is exercised in accordance with a constitution the essentials of which all citizens as free and equal may reasonably be expected to endorse in the light of principles and ideals acceptable to their common human reason.” ^{Footnote 3} Call this the “universal consent” condition, or universal consent for short. ^{Footnote 4}

Taken together, justice first and universal consent define the contractarian or contractualist position in political philosophy which Rawls more or less single-handedly revived, and which continues to define the mainstream of academic liberalism. Whatever their other disagreements, those who endorse these two claims assign priority ipso facto to two corresponding projects: first, the project of determining what justice requires, and thus what a fully just society would look like, and second, the project of demonstrating that the requirements in question are ones that all reasonable people could be expected to accept (or not to reject). ^{Footnote 5} These two projects are, however, in considerable tension with one another. On the one hand, justice first entails the further claim that questions about the meaning of justice must be settled before we can effectively pursue justice — or indeed any other political value, since other values cannot, by hypothesis, be pursued unless doing so is consistent with what justice requires. In other words justice is, from this point of view, “first” in both the moral and the temporal sense. On the other hand, universal consent entails a project of justification that, like all projects of justification, can only take place over time. Even if we grant that it is possible in principle for all people to agree about what justice requires, the process of actually reaching agreement can be expected to take generations at best.

Just as justice first and universal consent define the terms of contractarian liberalism, so too does the tension between them define the terms of the ideal theory debate. ^{Footnote 6} For example, many critics of ideal theory object that the contractarian commitment to justice first fails to provide us with any guidance about how we should act in the unjust world in which we actually live. As Amartya Sen puts it, “a transcendental approach cannot, on its own, address questions about advancing justice and compare alternative proposals for having a more just society, short of proposing a radical jump to a perfectly just world”: “we may indeed be willing to accept . . . that Everest is the tallest mountain . . . but that understanding is neither needed, nor particularly helpful, in comparing the heights of, say, Kanchenjunga and Mont Blanc.” ^{Footnote 7} Ideal theorists respond that we cannot coherently talk about “advancing justice” unless we know what a “perfectly just world” would look like; as John Simmons puts it, “[w]e can hardly claim to know whether we are on the path to the ideal of justice until we can specify in what the ideal consists.” From this point of view nonideal theory has an important but distinctly subordinate role to play in thinking about justice: “Where ideal theory dictates the objective, nonideal theory dictates the route to that objective.” Moreover, because this route may not be linear, it is neither necessary nor sufficient on this account that a theory of justice allow us to make comparative judgments about the relative justice of the intermediate steps: “which of two smaller ‘peaks’ of justice is the higher (or more just) is a judgment that matters conclusively only if they are both on equally feasible paths to the highest peak of perfect justice. And in order to endorse a route to that highest peak, we certainly do need to know which one that highest peak is.” ^{Footnote 8}

Taken in itself, this is a compelling response to Sen’s objection. However it is hard to know what to make of the claim that “we” need to know what the highest peak is in order to chart a route forward – or of Adam Swift’s related claim that “we need fundamental, context-independent, normative philosophical claims to guide political action even in nonideal circumstances” ^{Footnote 9} – because strictly speaking these are things that “we” have assuredly never known or had, and in all likelihood never will. Even if we take the word “we” in the distributive rather than the collective sense, so that the claim is that each of us needs an ideal of justice to guide our actions in a non-ideal world, we are left with a dubious piece of moral psychology and an unhelpful piece of political philosophy: social justice is, after all, something that “we” can only achieve (if at all) by acting together. In short, the effort to rescue justice first from the charge of unhelpfulness brings us immediately up against the demands of universal consent, and of what Jeremy Waldron calls the “circumstances of politics”: the existence of fundamental disagreement about what justice requires, coupled with the necessity of nevertheless making a collective decision about what should be done. “[W]hatever else we wish away in our elaboration of ideal models,” Waldron insists, “we should not wish away the fact that we find ourselves living and acting alongside those with whom we do not share a view about justice, rights or political morality.” ^{Footnote 10}

Contractarian liberals are of course perfectly well aware that there are few if any moral principles that would actually be endorsed by all people in a liberal society, even if we suppose (optimistically) that they are all reasonable in the sense of being sincerely motivated to identify and abide by such principles. ^{Footnote 11} They therefore propose to satisfy universal consent by idealizing the consenting parties and the conditions under which they offer their consent. Rawls’s famous “original position” and the various other thought experiments that contractarians have devised are designed to accomplish this task. As both defenders and critics of these devices have pointed out, the idealizations that they propose place a strain on the liberal credentials of the resulting theories of justice. That is, to the extent that we define consent counterfactually — as something that would or should have been given under the right conditions — it becomes doubtful that we have shown our actual, nonideal fellow citizens the kind of respect that contractarians insist they are owed, and difficult to see how or in what sense ideal justice can serve as a point of orientation for political action in the nonideal world. This, again, is the dilemma that underlies the contractarian position: on the one hand, taking the demands of universal consent seriously requires that we defer or dilute the demands of justice first — and thus risk giving undue deference to the benighted and selfish beliefs that people actually hold. On the other hand, taking the demands of justice first seriously requires that we defer or dilute the demands of universal consent — and thus risk running roughshod over those who happen to disagree with us. ^{Footnote 12} In light of this dilemma it is not surprising to find defenders and critics of ideal theory talking past one another. I don’t see any non-question-begging way to resolve the dilemma, and I therefore think that it is time to take a step back and rethink the assumptions that define the ideal theory debate in the first place.

I propose that we do so by returning to the etymological roots of the word “liberalism” in the word “liberty,” and treat freedom rather than justice as the “first virtue” of a social system. This may not seem like a promising strategy for avoiding the dilemma that I have identified, since people are at least as likely to be uncompromising about freedom as about justice — “give me liberty or give me death,” as the saying goes — and since the prospect of reaching agreement about what the pursuit of freedom requires, politically speaking, is at least as remote as the prospect of doing so with respect to justice. I respond to this objection in two stages. First, I point out that the liberal understanding of freedom is internally complex; that in attributing or extending liberal freedom to individuals we are in effect doing two things: on the one hand, defining the conditions under which they can properly be held responsible for what they do, and on the other, defining the conditions under which they are nevertheless not held responsible (Section II). ^{Footnote 13} Second, I argue that the pursuit of freedom so defined requires that we strike a balance between these two considerations, that this is irreducibly a matter of judgment, and that we therefore should not expect to reach agreement about what the pursuit of liberal freedom requires, at least for the foreseeable future. In short, a commitment to liberal freedom requires that we abandon universal consent (Section III). This raises the question of whether a freedom-centered liberalism can take into account the moral intuitions, rooted in a commitment to the value of equal respect, that lie behind the contractarian commitment to universal consent — intuitions whose force I don’t wish to deny. The key, as my epigraph suggests, is to treat agreement about justice as the endpoint rather than the starting point of political action, and thus to sever the link between legitimacy and consent (Section IV).

The aim of the following discussion is not of course to develop a full-blown alternative to contractarian liberalism — I will raise at least as many questions as I settle — but rather to try to show how the goals and challenges of liberal political philosophy, and indeed of liberalism itself, look when seen from a very different — and in many ways more traditional ^{Footnote 14} — moral perspective. As I have suggested, the theoretical impasse to which the ideal theory debate leads gives us reason to think that it’s time for a fresh start. This essay should be read a promissory note for future work along these lines. ^{Footnote 15}

II. Two Concepts of Liberty

When we ask whether someone is free to do something — or whether, having done it, they did it freely — we may be asking two different and seemingly contradictory things. On the one hand we may be asking whether it would be appropriate to hold them responsible for what they’ve done, or whether they can properly take responsibility (for example, in the form of credit or blame) for having done it. ^{Footnote 16} In other words, we may want to know whether they are properly subject to what P. F. Strawson calls the “reactive attitudes” on which our practices of holding responsible depend. ^{Footnote 17} This way of thinking raises two broad and difficult sets of questions. One set of questions has to do with the agent him- or herself: whether they are capable of making responsible choices in general (are they a child, an addict, or mentally deficient in some way?), whether they were in full possession of their faculties when they made a particular choice (were they intoxicated, sleepwalking, or hypnotized?), whether they were aware of the consequences of their actions (were they ignorant, manipulated, or deceived?), and so on. In short, when we ask whether someone did something freely, or whether they were free to do something, one thing that we could be asking is whether they are (or were) a responsible agent, however we define that term. It follows that one thing that we could be saying when we say that we want to make someone free (or more free) is that we want to make them into a (more) responsible agent: to ensure that they’re mentally sound, properly informed, not subject to manipulation, and so on.

A second and equally difficult set of questions raised by the idea of freedom as responsible choice has to do with the social context in which the agent acts. Most if not all of our decisions are shaped, explicitly or implicitly, by the expected costs and benefits of acting in one way rather than another. Do we act freely — are we responsible for our choices — insofar as our behavior is influenced by these kinds of considerations? Do I act freely when I pay my taxes, take a detour to avoid a traffic jam, use “proper” table manners, or call my sister on her birthday? What about when I pay off an extortionist, take a detour to avoid an unsafe neighborhood, observe “proper” gender norms, or refrain from expressing an unpopular opinion? In order to answer any of these questions in the affirmative we would have to tell some kind of story about the relationship that we stand in with respect to the constraints in question; in particular, one that establishes that we are in some sense responsible for their existence: because we played a role, or could have played a role, in creating them, because we endorse their content, or would do so on reflection, because we’re able to contest them when we disagree, and so on. In other words, we’re free in this sense if and to the extent that we’re self-legislating in one of the many possible senses of that term.

For the sake of convenience, and following much precedent, I will call this first kind of freedom republican freedom, because it’s concerned with the question of how we govern ourselves and thereby make ourselves into agents who can properly be held responsible for what they do. As Philip Pettit, the leading contemporary philosopher of republicanism, puts it, “there is an a priori connection” between freedom and responsibility: “[s]omeone who did not see why that connection had to obtain would fail to understand what freedom was or what holding someone responsible was.” ^{Footnote 18} Of course the position that I have sketched here leaves room for a much broader range of possible “republicanisms” than most actual republicans — including me and Pettit himself — would be willing to defend. What it means to be free in this sense will depend, after all, on the kind of story that we tell about how the conditions of self-legislation and responsible agency should be met, and not all such stories will be equally attractive or plausible. Indeed, as my use of the word “story” may suggest, any particular claim that a person or a people is self-legislating, in the sense of being responsible for (some of) the constraints that it faces, will be contestable in principle and almost certainly contested in fact. Trying to get this right is one of the main things that a liberal politics is about. And, as we will now see, the claim that freedom consists in the absence of such constraints is no less open to dispute.

As I have pointed out, it is rarely if ever the case that we act in the absence of constraints, even if we only consider constraints that are imposed directly or indirectly by other people, as most theories of freedom do. ^{Footnote 19} For example, I may be free to choose how to spend my money, but what I can afford to buy is constrained by the choices of all of the other buyers (including the buyers or potential buyers of my own labor and other resources), which add up to an overall pattern of prices that I do not control. I may be free to choose what route to take when I drive to work, but how long it takes me to get there will be constrained by the choices of all of the other drivers, which add up to an overall pattern of traffic that I do not control. I may be free to choose with whom to associate and how to act in a variety of social contexts, but I do not control how other people will react, for better or worse, to the company that I keep or the things that I do. In such cases I’m free in the sense that I am the one who decides what to do given what everyone else is doing, or given what I expect them to do; I decide how to respond to the pattern of potential costs and benefits that is created by other people’s choices. But I’m not free in the sense that I control the pattern itself; indeed, unlike in the case of republican freedom no one controls it: there’s no story to be told here about self-legislation even in principle.

The flip side of this line of argument is that my decisions affect other people in ways that they do not control: when I choose what to spend money on, which way to drive to work, with whom to associate, or how to act, my choices have a marginal effect on the overall pattern of prices, traffic, association, and behavior — and often an effect that I did not intend. The distinguishing feature of this kind of freedom is thus not the absence of constraint on choice: my choices are constrained, as I’ve pointed out, by everyone else’s choices. Nor is it the exercise of control over choice: I can act recklessly, compulsively, ignorantly, or out of sheer habit and still act “freely” in this sense. Freedom consists here rather in the absence of responsibility for choice: one is free in this sense if and to the extent that one is not accountable to anyone else for what one does, except by choice. To put the point in slightly more technical terms, one is free in this sense if one is allowed to impose costs on other people without being answerable to them for doing so. I decide what price I’m willing to sell my house for, and if I lower my neighbors’ property values or raise their property taxes in doing so, then that’s my business. I decide which way to drive to work, and if I waste gas or contribute to a traffic jam in doing so, then that’s my business. I decide how to respond to the various social pressures that I face — who to associate with, whose approval to seek, and whose disapproval to ignore — and if I offend some people or encourage what some people regard as bad behavior in doing so, then that’s my business.

For the sake of convenience, and again following much precedent, I will call this kind of freedom market freedom, because it is concerned with allowing people to decide for themselves how to respond to the various opportunities and hazards with which social life presents them. As we have seen, the word “market” in this sense can be used to refer to a wide range of social spaces that have little if anything to do with the exchange of commodities — it is similar in this respect to Friedrich Hayek’s similarly expansive use of the term “catallaxy.” ^{Footnote 20} From the standpoint of the individual actor, ^{Footnote 21} a market is, again, any domain of conduct in which we can impose costs on other people without being answerable to them for doing so. From the standpoint of society as a whole, a market is any domain of conduct in which the unsupervised, unregulated, and uncoordinated behavior of individuals gives rise to an overall pattern of outcomes that no one can predict, control, or take responsibility for; one that is, as Adam Ferguson famously put it, “the result of human action, but not the execution of any human design.” ^{Footnote 22} Every decision that I make in a “market” has a marginal effect on the overall pattern, but I enjoy market freedom not because I am (in some sense) responsible for that pattern, but rather precisely because I am not: the effects of my choices, and sometimes even the choices themselves, are morally (and often literally) invisible to other people.

These, then, are the seemingly contradictory answers that a liberal can give to the question of when a person is free, or under what conditions it becomes possible to do something freely: the first, which I have associated with the term republican freedom, says that we are free when we can properly be held responsible for what we do, and the second, which I have associated with the term market freedom, says that we are free when we are not responsible to anyone else for what we do, except by choice. This dichotomy, it seems to me, provides a more perspicuous framework for thinking about social and political freedom than the enormously influential but highly equivocal dichotomy between “negative” and “positive” liberty that Isaiah Berlin lays out in his famous essay “Two Concepts of Liberty.” ^{Footnote 23} I will now argue that, far from being contradictory, republican and market freedom are in fact complementary ways of thinking about freedom. The challenge is not to choose between them, but rather to define the domains of action within which they should each be applied, and this is the proper task of liberal political thought.

III. Liberal Freedom

Liberalism as I understand it — and, I think, as it was understood by the most admirable and influential liberals of the past — is the political ideology that holds republican and market freedom together in a single political vision. It follows that liberal freedom, like liberal society, has two distinct and complementary dimensions, which give rise to two distinct and complementary moral aims: on the one hand, to create the social conditions that make responsible agency possible, and on the other hand to carve out a social space within which the demands of responsible agency are relaxed or absent. Liberal thinkers, and thinkers who have been inspired by liberal ideals, have sometimes tried to collapse the tension between these two projects in one direction or the other — to treat either responsibility or non-responsibility as being not only a necessary but a sufficient condition for the enjoyment of liberal freedom. I want to argue, however, that there is no way to resolve the tension without stepping outside the non-utopian — or, if you like, the nonideal — bounds of a liberal politics. Our aim in theorizing about freedom should therefore not be to determine which of these conceptions is the correct or superior one, but rather to determine what their proper spheres of application might be. Liberal politics consists, in short, in an ongoing debate about what we should hold each other responsible for and how, and thus (conversely) about what the extent of market freedom should be.

It should be clear to begin with that the enjoyment of market freedom depends on a prior guarantee of republican freedom; that is, on the existence of a set of well-enforced rules that make orderly social behavior possible. It should also be clear (though here the details are more controversial) that the domain of market freedom can appropriately be limited in a variety of ways for the sake of republican freedom. Commercial markets depend for their proper functioning on the existence of well-defined property rights and of publicly enforced guarantees against theft, force, and fraud, and all polities regulate or otherwise limit the costs that market actors can impose on other people: for example, through minimum wage and maximum hours laws, workplace and product safety laws, anti-discrimination laws, environmental protection laws, and so on. The same is true, mutatis mutandis, of markets in the broader sense that I have described: our ability to participate effectively in any “unregulated” social space requires that we have certain guarantees regarding our personal safety and the reliability of our potential partners. We have therefore decided as a society that people should be held responsible — that we will hold each other responsible — for imposing certain kinds of costs on other people. I am free to decide what to spend my money on, but I am not allowed buy something that is not recognized as a legal commodity, such as a person, narcotics, or leaded gasoline. I am free to decide what route to take when I drive to work, but I have to stay on public thoroughfares and obey the relevant traffic laws. I am free to decide with whom to associate and which social norms to observe, but I am not allowed to join (what society has defined as) a subversive organization, or to engage in (what society has defined as) obscene or indecent behavior.

To expand the domain of market freedom — for example, by lifting trade barriers, loosening traffic laws, removing restrictions on association, or repealing obscenity codes — is to transfer decision-making power from an arena in which identifiable actors can be held publicly responsible, at least in principle, for the social outcomes that are realized to an arena in which this kind of responsibility does not exist. More precisely, since responsibility is never perfectly enforced and non-responsibility is never complete, to expand market freedom is to shift the balance of considerations from greater public responsibility toward greater private non-responsibility. One kind of freedom is lost in such a shift — republican freedom; the freedom not to be subject to irresponsible or arbitrary power — and another kind of freedom is gained — market freedom; the freedom to make choices without being answerable to others for the choices that one makes. Liberals don’t agree about how exactly the trade-offs between these two kinds of freedom should be made, but they do agree that republican freedom should have priority in the procedural sense that the trade-offs should be made in publicly visible and contestable ways and for publicly avowable reasons. In other words, not only is republican freedom a necessary condition for the enjoyment of market freedom, but a liberal society is one in which the rules that guarantee and limit market behavior are created and enforced in a way that is consistent with the demands of self-legislation, however those are defined.

It’s not immediately clear, however, that the enjoyment of republican freedom depends on the existence of market freedom, except of course to the extent that the enjoyment of republican freedom requires that we keep a certain amount of decision-making power out of the hands of the state. What grounds, then, do liberals have for giving independent weight to market freedom? Why shouldn’t we always hold people responsible for their choices, at least in principle, if those choices impose costs on other people without their consent? Why should we allow people to consume wastefully, say hurtful things, neglect their health, or spoil their children? Let me suggest thirteen different reasons that we might have for doing so – reasons that, while conceptually distinct, will often reinforce or conflict with each other in practice:

1. (1) Indifference: A given domain of conduct is thought to be too insignificant to be worth regulating.

2. (2) Tradition: There is a longstanding practice of letting people make their own decisions in a given domain, and we’re unwilling to disrupt those habits of private decision-making, or it simply does not occur to us to do so.

3. (3) Conservatism: We defer as a matter of principle to existing habits of private decision-making, with the thought that there must be a good reason why they exist even if we can’t articulate what it is.

4. (4) Infeasibility: The necessary administrative or coercive capacity to regulate a given domain of conduct does not exist, or it would be unacceptably costly to develop and exercise such a capacity.

5. (5) Epistemic modesty: We don’t believe that we would be able to make a good rule if we set out to regulate a given domain of conduct, because we lack the necessary knowledge or judgment.

6. (6) Stability: We disagree so deeply about what the rule should be that any effort to regulate a given domain would create an unacceptable level of social conflict.

7. (7) Prudence: Imposing a rule in a given domain would give the state too much power, or set a dangerous precedent for future cases.

8. (8) Diversity: Allowing people to make their own decisions in a given domain allows for a wider range of outcomes than a centralized or “one size fits all” approach.

9. (9) Efficiency: Allowing people to make their own decisions in a given domain leads to more efficient outcomes than a centralized or “one size fits all” approach.

10. (10) Dispersal of power: Allowing people to make their own decisions in a given domain would end or undermine a monopoly on certain goods by a dominant class or faction.

11. (11) Perfectionism: Allowing people to make their own decisions in a given domain would teach them a valuable lesson in self-discipline or self-reliance.

12. (12) Anti-paternalism: People should be allowed to make their own decisions in a given domain because they know their own interests best.

13. (13) Sovereignty: People have a right to make their own decisions in a given domain.

Some of these reasons cluster together in familiar ways: (2), (3), and (5), for example, are characteristically “conservative,” (8), (9), and (12) (at least) are characteristically “libertarian,” and (7), (10), and (11) are characteristically “republican.” The important point for present purposes is that the force of each of these reasons, and of whatever other reasons we might invoke, will vary substantially across cases and over time. Which decisions seem insignificant; which traditions seem worth respecting; what it is feasible or prudent to regulate; what we are willing to fight about; how important we think diversity, efficiency, and self-reliance are; how well we think people know their own interests; even what rights we think people have: we disagree about each of these issues now, and our views about them have changed significantly over time and are likely to continue to change going forward. It follows that we are not likely to agree about what the pursuit of liberal freedom requires, either now or for the foreseeable future.

Liberal freedom is thus not a condition to be met or a value to be maximized so much as a balance to be struck between the competing claims of republican and of market freedom. In this sense it functions more like an Aristotelian virtue than a contractarian side-constraint or a utilitarian maximand, and like an Aristotelian virtue its meaning is likely to be clarified and its aims advanced as much through exemplary action as through reasoned argument. This is not to deny that we have relatively fixed views on certain questions. There are domains of action in which we can agree that it would be illiberal to hold people publicly responsible for their actions — for example, with regard to their religious beliefs and practices (within certain limits) — just as there are domains of action in which we can agree that it would be illiberal not to do so — for example, with regard to their treatment of other people’s property (again, within certain limits). These are, for the time being, matters of settled judgment in liberal thought. But the caveat is crucial: judgment is settled through experience, and the lessons of experience are subject to revision through further experience over time. It would require a gross reification of terms to suggest that the property rights and religious freedoms that the citizens of liberal societies enjoy today are the same rights and freedoms, practically speaking, as the ones that were defended by the “classical” liberal thinkers of the seventeenth, eighteenth, and nineteenth centuries, and even the most nostalgic liberal today would probably not be willing on reflection to trade today’s bundle of rights and freedoms for that of 1690, 1790, or 1890.

This raises the question of how the exercise of political authority can be made legitimate — how it can be exercised in a way that is consistent with the equal respect that liberals believe all citizens are owed — when there is persistent disagreement about the practical meaning of the fundamental values on which a liberal society rests. As we have seen, this is the worry that gives rise to the contractarian commitment to justice first and universal consent. I will conclude with a brief consideration of how we might go about answering this question from the freedom-centered perspective that I have outlined here.

IV. Beyond Contractarianism?

I began this essay by pointing out that the contractarian commitment to justice first requires that we dilute our commitment to universal consent by defining consent in counterfactual terms. Contractarian liberals therefore spend a lot of time and energy on the metapolitical project of trying to show that a given way of constructing consent — of defining reasonableness — is in fact reasonable. A justice-centered liberalism is in this sense a justification-centered liberalism. Justification — the exchange of reasons with the aim of persuading others — is of course an essential part of any politics, liberal or otherwise. However, the contractarian approach raises the stakes of justification to (we might say) an unreasonable level. Those who refuse to accept one’s own favored principles of justice — who do not recognize themselves in the idealized conditions under which their consent is said to have been given — must after all be deemed “unreasonable” if their refusal is not to detract from the legitimacy of those principles. This is not a very promising rhetorical stance to take if one hopes to persuade those with whom one disagrees — nor is it a very liberal one. After all there are few if any interesting political questions — not even regarding “constitutional essentials” — on which all people of good faith can be expected to agree or converge. ^{Footnote 24} Liberal political thought, insofar as it follows the contractarian line, is therefore characterized by a strange mixture of diffidence and insularity — intensely preoccupied with the kind of immanent critique that one would expect from people who take universal consent as their ideal, but largely uninterested in engaging with political views that do not begin from liberal premises as it defines them.

I have argued that we should therefore abandon the contractarian commitment to justice first and universal consent and instead think of liberalism in freedom-centered terms. However, I have also argued that liberal freedom depends in part on the republican ideal of self-legislation. What could it mean, then, to say that the rules by which I am governed and the responsibilities that I have toward others are rules and responsibilities that I have imposed on myself, if not that I have in some sense consented to them? The key to answering this question, as I see it, lies in the fact that defining the limits of republican and market freedom is a matter of political judgment, and that the exercise of judgment always involves making claims about the future: moral claims as well as empirical ones. That is, in defending a given course of action we are not only predicting that a given set of consequences will follow, but also that we will evaluate those consequences in a certain way. Needless to say, both types of claims are inherently fallible: what we think is going to happen does not always happen, and even when it does we don’t always evaluate the results in the way that we expected. Things often look very different once we have them than they did when we imagined having them.

Because political judgment is prospective and fallible in this way, so too must be our understanding of political legitimacy. I therefore propose that we think about legitimacy in the following way. ^{Footnote 25} Whenever a politically decisive coalition acts in the face of objections from some of their fellow citizens — that is to say, whenever political action is taken — its members must hold, if their decision is to be legitimate, that the dissenters could be brought to accept their position if sufficient arguments and evidence were made available to them — and they must have made, and must continue to make, a good faith effort to provide the necessary arguments and evidence. They must, at the same time, concede — again, if their decision is to be legitimate — that this prospective claim of justifiability could in principle be rebutted if sufficient arguments and evidence fail to emerge, or if sufficient arguments and evidence to the contrary come to light — and, crucially, they must ensure that the necessary conditions for free and open criticism of their own arguments and evidence are in place. ^{Footnote 26} The dissenters assume the opposite set of practical commitments: pointing out the flaws in the acting coalition’s arguments, the gaps in their evidence, the ways in which things haven’t worked out as they predicted, and so on – and, of course, they must be prepared to admit that they too may be mistaken. In more personal terms, if you and I disagree about a matter of public policy, and if we intend to try to impose our beliefs on the polity as a whole (for example, by voting), then I have to believe that you sincerely believe that you could persuade me that your preferred policy is superior to mine if only you had enough time and information, you have to believe the same thing about me, and we must each have an effective opportunity to persuade the other. As fellow citizens of a liberal polity we are committed to continuing to engage with each other in this spirit for as long as we disagree: our (republican) freedom depends on it, because otherwise we cannot properly be held responsible for the various ways in which our actions are constrained by the exercise of political authority.

Unlike the contractarian view, which treats justice as the first virtue and so requires, at least in principle, that we settle basic political questions ahead of time, a freedom-centered liberalism is essentially forward-looking: it combines an appreciation of the fact that our judgment and understanding are always fallible with a commitment to improving them over time. Legitimacy is, on this view, not an established state of affairs but rather an ongoing project; one whose realization lies in the hands (and minds) of ordinary citizens. This view is not as starry-eyed as it may seem at first blush. Ordinary politics often puts a severe strain on this kind of relationship: it’s often hard to believe that our fellow citizens are people of good will who are acting according to their best lights, and it’s often hard to commit ourselves to doing the hard work of trying to justify ourselves to those who disagree with us, not least because this often requires that we step outside of our ideological comfort zones — and because we often have reason to doubt that our interlocutors are really listening. As with our views about the practical entailments of liberal freedom itself, the extent to which we are willing and able to maintain the necessary beliefs, and to engage in the correspondingly appropriate behavior, will vary substantially across cases and over time. But this is what we are committed to if we are committed to the project of building and living in a free society together. The theory of legitimacy that I have sketched defines the boundary conditions of a liberal politics; it tells us what we have to believe and do in order for such a politics to be possible. In that sense a freedom-centered liberalism is a “realistic utopia,” to use Rawls’s term: an ideal that it’s possible, if not entirely likely, that actual human beings could live up to ^{Footnote 27} — although this utopia is more “realistic” than Rawls’s in the sense that each of us has an opportunity to try to live up to its demands in our daily lives as ordinary citizens of a liberal polity.

From this point of view questions about the proper meaning of liberal freedom are the very substance of ordinary politics, not a matter to be settled in advance or at a higher level of abstraction. This is not to say that ordinary politics is everything. Sometimes the strain on our belief in the good faith of our fellow citizens is too great, and sometimes our fellow citizens don’t even pretend to be acting in good faith. At such times we are not self-legislating even in the indirect and long-term way that I have described here, and so we are not free in the republican sense: we’re just obeying force — or, if we happen to be members of the acting coalition, we are just compelling others to obey force. Whatever our political views may be, we’ve all had the experience of obeying laws with which we not only disagree, but that we believe have been imposed in bad faith. This is, as far as I can see, an irreducible fact of political life. And sometimes, if the stakes are high enough, force is the appropriate response to such a situation: there is a distinguished tradition in liberal thought, stretching from John Locke in the 1690s to Martin Luther King in the 1960s, of defending the use of extralegal means, whether violent or nonviolent, to advance liberal ends in certain circumstances. But of course the question of what those circumstances might be, and of when and how this kind of resistance can be justified, lies far beyond the scope of this essay.

This, then, is how I believe that we can satisfy the liberal commitment to equal respect without committing ourselves to the contractarian norm of universal consent. Might we still hold on to justice first? In particular, might we characterize the liberal effort to balance republican and market freedom as itself being a justice-centered project? Might we say, in other words, that to answer the question of how and to what extent we should be held responsible for our actions is simply to answer the question of what justice requires? I have no objection to this way of putting the point. However, Madison offers a useful word of caution in the epigraph to this essay. Justice, he writes, is the end of government and of civil society; it’s not something with which we begin but rather something whose meaning and implications we work out together. And as the broader context of Federalist 51 indicates, to give any authority, whether majoritarian or philosophical, the unilateral power to determine what justice requires, and thus to ignore, override or assume away the disagreements about justice that exist among ordinary citizens, is to ensure that “liberty [will] be lost in the pursuit.” Given the flawed and fallible grasp that each of us has on what justice requires, liberty remains for Madison, as it does for most liberals, the primary value.

A freedom-centered liberalism treats politics as an arena in which the proper ends of public life and the proper bounds of public authority are open to revision at any time and to substantial change over time. On this view liberal ideals are the proper objects of political contestation, and not just the background against which such contestation occurs. The ideals that liberal polities stand for today are not the same ones, practically speaking, that such polities stood for in the past, and there is good reason to expect that the ideals that we stand for today will not be attractive, or even fully intelligible, to future generations. The boundaries of republican and market freedom have changed significantly, and can be expected to change further, and a moment’s reflection will reveal that resistance to this kind of change has hindered the advancement of liberal ideals at least as often as it has promoted it. The only way to ensure that liberal ideals as we understand them today are not corrupted or abandoned in the future is therefore through our own efforts and those of like-minded people. If we can look to the past achievements of liberal polities — the widespread (but still imperfect) acceptance of religious toleration and of free speech and inquiry more generally, the relative (but still woefully incomplete) deconstruction of racial and gender hierarchies, the (possibly temporary) defeat of totalitarianisms left and right — to remind ourselves of the promise of liberal citizenship, we can look to our many remaining moral failings, and to the fact that even our achievements are preserved only through our own vigilance, to remind ourselves how fallible our ideals, and how fragile the institutions that we have built upon them, actually are.