Why not anarchism?

Robert Nozick (1974, p. 3) famously opens Anarchy, State, and Utopia by asking, “If the state did not exist, would it be necessary to invent it?” Whether the state should exist in the first place is, in Nozick's view (p.4), “the fundamental question of political philosophy.”

Recent debates over ideal theory have reinvigorated interest in the question of anarchy. Would a perfectly just society need—or even permit—a state? Ideal anarchists such as ourselves, G.A. Cohen, and Jacob Levy argue that strict compliance with justice obviates the need for a state. Ideal statists such as John Rawls, David Estlund, and Gregory Kavka think that coercive political institutions serve indispensable functions even in ideal conditions. This paper defends ideal anarchism.

At stake in the debate between anarchists and statists is the very concept of a just society. (Compare “If this were a just world, the government would really punish those Wall Street crooks instead of placing them in a cozy white collar prison” with “If this were a just world, Wall Street wouldn’t be crooked—or wouldn’t even exist—in the first place.”) Moreover, the debate spotlights the possibility that political philosophy has been beset by unnoticed moral complacency. Both Cohen and Estlund argue that philosophers and others make improper concessions to human nature. People could behave well, though predictably they won’t because they don’t want to. Rather than curving or dumbing down morality, we should instead conclude people are morally flawed and most societies fall far short of justice. ¹

Consider, for instance, that we have little expectation that those who are well off will willingly share enough of their spare resources with the poor; the state has to resort to force. Or, as Cohen notes, we’ve resigned ourselves to a world in which the talented demand higher wages to put their talents to work for the common good. But perhaps we should hold ourselves to a higher moral standard. We might have no choice but to tax the rich or compensate the talented extravagantly to motivate them to do what we want; still, this response is only required because the rich aren’t behaving as they should. To borrow Cohen's memorable analogy (2008, part 1), it's like paying a kidnapper's ransom to secure the release of your child. You ought to do it, but the kidnapper was wrong to take your child in the first place. In this paper, we examine how most arguments for the state take this form—they presuppose people act quite badly and that the state is a necessary response to their bad behavior. But if people were consistently decent, let alone ideally just, the state would not be necessary or valuable.

Although ideal theory asks which institutions would be best under a rather narrow and special set of conditions, it has a point. Many theorists insist that a central task of political philosophy is to identify which institutions could in principle fully realize justice. ² Political philosophy is normative, not descriptive. It tells us how societies ought to function, not how they in fact function. Moreover, that a perfectly just society is infeasible does not make it any less intrinsically desirable. As Cohen (1995, p.256) says, grapes do not become any less tasty just because they are out of reach. Contrary to Rawls, Estlund, and Kavka, we argue that a state is only needed in conditions where people are morally defective and is therefore, by its nature, a nonideal institution.

Previous defenses of ideal anarchism tend to be noncomprehensive or impressionistic. For instance, Cohen focuses his attention on the claim that ideal egalitarians would be motivated to share the wealth without taxation. Brennan argues that ideal capitalists would not exploit their workers or withhold help from those in need. Levy (2016, p. 319) notes that a strictly compliant society would be without crime and “the limited beneficence that sits at the base of theories of justice in property and in the coercive provision of social welfare, and more generally the failings that make politics and justice unavoidable.” Freiman's earlier arguments for ideal anarchism go into greater detail but leave a variety of objections unaddressed. In this paper, we offer a more comprehensive defense of the position.

Our preliminary argument for ideal anarchism begins by describing a case inspired by Cohen—call it “the anarchist camping trip”—that illustrates why fully just people would cooperate without coercion (§1). We then detail the assumptions of Rawls's ideal theory and criticize his claim that a state is needed to solve assurance problems in ideal conditions (§2). Next we explain Estlund's arguments against dumbing down the requirements of justice to accommodate people's bad moral motivations and show why a consistent application of Estlund's “non-concessive” assumptions undermines his own case for the state (§3). We then address Kavka's argument that a state is required to resolve moral conflicts among morally blameless but zealous people (§4). We conclude that none of the major arguments for why strictly compliant societies would need a state are successful and that the case for ideal anarchism stands (§5).

§1 The Anarchist Camping Trip

Cohen (2009) famously argues for socialism with a thought experiment that has us imagine a camping trip among friends. First, he has us imagine people living together while abiding by (purportedly) socialist principles of community and equality. Then he asks us to imagine people start behaving in ways common in real-life capitalism, such as refusing to use their talents unless they are paid more than others, or gloating over having more than others. Nearly everyone agrees the “socialist” version of the camping trip was morally superior to the “capitalist” version.

Cohen then asks whether it would be desirable if the entire world could be run like the socialist version—on the basis of community spirit and equality—rather than the capitalist version—on the basis of fear and greed. He tells us to be careful to distinguish what people won’t do from what they can’t do. That people are predictably selfish, callous, indifferent to others’ welfare, willing to exploit one another, and overly concerned with promoting their own status does not show that they could not act otherwise. Sure they could. Instead, it shows that they are morally flawed people. If people were good—not necessarily even ideal, but simply consistently decent—they would not act in these ways. Cohen thus concludes that the motivational impediments to socialism do not show that socialism is undesirable. Given what we know about how people will choose to act, we may have reason to not try to instantiate socialism in the real world. But on Cohen's view, socialism nevertheless remains the best way to live together.

In this section, our argument and our language will often mirror Cohen's. So too does our conclusion. Our claim takes no stand on the feasibility of anarchism. Rather, our claim is that fully just people would be anarchist If we should have a state, that's a reaction to our moral shortcomings. Later, we will consider the possibility that, pace this argument, even ideal people would need the state.

Now for the anarchist camping trip: Imagine that we go camping. We have no hierarchy and no one in charge. We have different goods, such as matches, fishing rods, and coffee, among us. Some of these might be collectively held; some might be privately held. We are largely free to do as we please, and no one tramples on anyone else's rights. Everyone treats everyone else with respect and kindness.

There is peace. Everyone respects each other. People observe rules that lead to the common good. No one steals. If anyone is in need, someone always pitches in. Everyone does their fair share to maintain common facilities.

People also abide by rules that make coexistence better. Some rules are just the traditional rules of the campsite, which evolved over time from experience and from people copying functional behaviors they witness in others. In some cases, rules are decided on by discussion. If there is insufficient time for everyone to discuss rules, then a few of the campers, with great care, will suggest good rules, and everyone accepts useful, functional rules. People recognize the value of cooperation and loathe the idea of violent conflict. Accordingly, there is no such conflict.

No coercion or threat of violence is needed to maintain order and community. Everyone understands that she would be able to violate others’ rights or the rules with near impunity, if she so chose. However, out of moral motivation alone, campers respect each other's rights, play fair, and abide by the rules.

Now suppose that people begin behaving as they do in real-life statist societies. Suppose that Sylvia says she needs assurance that others will do their fair share. She says we should not count on other campers’ good will. Thus, Sylvia demands that some people be armed with sharp sticks and charged with enforcing the rules. Harry and Leslie are tasked with the job and subsequently demand that all campers provide them a set amount of goods as compensation, to be collected at regular intervals.

Meanwhile, Morgan has been happily sharing her apples with the other campers whenever they ask for one; that is, until a bunch of apples goes missing. She sees Nina snacking on what looks like one of her Granny Smiths and, without any further evidence, publicly accuses Nina of theft. The rest of the campers ostracize Nina. Some even contemplate violent retribution. They are only deterred by the prospect of being stabbed by Harry and Leslie and locked in a makeshift cage.

Campers also start having significant disagreements about the social norms, and organize into factions. They give their factions names, such as “the Social Justice Party” or “the Honored Tradition Party”. They compete for the right to hold the sharp sticks and tell others what to do.

At this point, the campsite resembles a state. But notice that when the campers had no moral failings, they were anarchistic, while the introduction of the state here was motivated by their moral failings, distrust, and animosity. (Again, we will consider below whether even angels would need a state.)

Non-coercive anarchism may not be feasible given how bad people are, but our claim is not about feasibility. We claim that a fully just society would be anarchist rather than statist The campers’ proto-state was only needed when unjust behaviors were introduced—when the campers were unwilling to voluntarily do their fair share, entertaining the use of violence against other campers on the basis of a mere suspicion, or opting to resolve honest moral disagreements through war rather than peaceful compromise. However, as we will see below, these are precisely the sorts of behaviors that ideal statists believe give rise to a need for the state. Rawls, Estlund, and Kavka see themselves as arguing that even ideal people would need a state, but in fact their arguments rely on the assumption that people will act in decidedly nonideal ways.

Still, there is disagreement on how, exactly, we should expect ideal moral agents to behave. Perhaps situations similar to those described in the later stages of the camping trip will arise even within ideal conditions. In reply, we argue that nongovernmental solutions will work as well as governmental solutions. And since there is a moral presumption in favor of respecting people's liberty, noncoercive, nongovernmental solutions are morally preferable to coercive, governmental solutions.

Following Gregory Kavka (1995, p.5), we understand a government or a state to be the subset of a society which claims a monopoly on the legitimate use of coercion, and which has coercive power sufficient to maintain that monopoly. Many arguments for the state take the following form: “In the absence of a state, people will do X or fail to do Y, and so we need the state to stop them from doing X or help them/force them to do Y.” But any such argument which appeals to bad motivations or a morally blameworthy lack of good motivations is nonideal. For instance, if someone says the state is necessary to stop mutual predation, this presumes people will act badly. Statist arguments can also be nonideal when they appeal to people lacking information or certain skills, if the reason they lack such information or skills is that they make morally bad choices. Such arguments may well justify having a state in the real world, given people's actual bad behavior, but it does not thereby show that a just society full of just people would have a state. They at most show that a state is a just response to unjust people's pervasive injustice.

In the real world, we rely upon institutionalized threats of coercion because we think we must, given human limitations. We tax the rich and impose emissions caps because flawed people will act unjustly otherwise. But if people were good—not perfect, just good—they would do the right thing without coercive threats. They would act like people on the anarchist camping trip. To be clear, a morally good society would no doubt include institutions—roughly, rules structuring social cooperation—but these institutions would not include a coercive, monopolistic state. In this anarchistic situation, each person abides by norms and rules that serve the common good, promote justice, and solve coordination problems. No one free rides on public goods. No one engages in rent seeking. No one seeks to abuse institutions for private ends. No one takes advantage of, exploits, or preys upon others.

As noted, statists might rightly complain that cooperative anarchy would not succeed in the real world. ³ Yet at the level of ideal theory—in which everyone does what is right—it appears cooperative anarchy would work. As Levy notes (2016, p. 319), “Taking ‘strict compliance’ seriously would mean assuming away the crime that justifies the state's control of the means of violence.” Freiman elaborates (2017a, p.6), arguing that criminal justice institutions are unnecessary in a fully just society: “We need laws against, say, shoplifting—and not simply polite suggestions—because some people would be tempted to shoplift if not for the threat of punishment. The state's function is essentially remedial. But in a fully just society, there's nothing to remediate. A fully just person doesn’t want to shoplift.” ⁴

Similar observations apply to redistributive taxation. Cohen (2009, p. 211) says that “in a truly just society, with full compliance, taxation on behalf of equality would not need to be coerced.” Since citizens are assumed to be morally just, they will voluntarily organize and contribute to charitable endeavors without the need for state-enforced taxation. Labor market regulations would also be superfluous. As Brennan puts it (2014, p. 86), an ideal capitalist “would never exploit anyone, because she is too nice.”

Perhaps, despite this, there are other reasons why ideally just people need a state. In what follows we will examine arguments from Rawls, Estlund, and Kavka against cooperative anarchy. We contend that these arguments fail and that ideal agents don’t need a state. More specifically, we offer internal critiques of their arguments. That is, we criticize their cases for the state by showing that a consistent application of their own assumptions about the nature of an ideal society favors anarchism.

§2 Rawls

To clarify up front: Rawls understands “ideal theory” differently in different contexts. Sometimes ideal theory aims to discover which fundamental principles of justice would be justified were society to comply with those principles (Rawls 1999, p. 8, 126).

We set this sort of ideal theorizing about basic principles of justice to the side. Instead, we focus on the variation of ideal theory that concerns itself with institutional design. Here we seek to discover which institutions would be justified if, in Rawls's terms (2001, p. 13), everyone exhibits “strict compliance” with the basic principles of justice. In Justice as Fairness: A Restatement, Rawls (2001, p. 137) is explicit about evaluating different political regimes according to an “ideal institutional description,” that is, “the description of how it works when it is working well, that is in accordance with its public aims and principles of design.” This style of institutional analysis permits us to forgo empirical inquiry given that an ideal description of an institutional regime is one that will “abstract from its political sociology, that is, from an account of the political, economic, and social elements that determine its effectiveness in achieving its public aims” (Rawls 2001, p. 137). As Rawls puts it (2001, p. 137), his question is “What kind of regime and basic structure would be right and just, could it be effectively and workably maintained?”

To illustrate exactly what ideal theorizing about institutions involves, consider the following two questions:

Question 1:

Question 1 asks about what are in some way “ideal” circumstances; question 2 asks about realistic circumstances. When someone says she favors or disfavors a particular political or economic system, we need to know at what level of idealization she claims to favor or disfavor that system. We need to know whether she is asking question 1, 2, or something else.

Rawls must be asking question 1. He tells us to ignore, for instance, criticisms of the welfare state based on the idea that the safety net might lead to moral hazard. He has us imagine that people have such good motivations and so strong a sense of justice that so-called “conservative critiques of the welfare state” automatically fail, because people would never engage in free-riding or take advantage of one another's generosity or sense of fairness (2001, p. 137).

More specifically, Rawls asks which institutional arrangement would best realize justice under these four conditions:

People have as strong a sense of justice as humanly possible. They might not be perfect, and they might experience strains of commitment if social institutions demand too much of them. However, everyone is as good as real human being could be, if not as good as an angel. Assuming that the strains of commitment are sufficiently modest, people fully comply or nearly fully comply with the principles of justice.

Rawls argues that ideal moral agents need coercive political institutions to solve assurance problems. As conditional cooperators, ideal moral agents are motivated to contribute their fair share conditional on others doing the same. (This needn’t be because they are immorally selfish, but instead because they recognize certain goals require collective action.) Virginians are happy to spend a little more on hybrid cars to do their part for the environment so long as Arizonans buy them too. Alaskans want to send in their donations to the social justice redistribution fund—but only if Texans put their own checks in the mail. Since these two groups cannot monitor each other's activities, they cannot know whether the other is making their own contributions. So both refrain from contributing. Thus, even ideally motivated actors might fail to provide public goods. Rawls (1999, p. 236) points to situations like these to justify the need for a state in ideal conditions:

The sense of justice leads us to promote just schemes and to do our share in them when we believe that others, or sufficiently many of them, will do theirs. But in normal circumstances a reasonable assurance in this regard can only be given if there is a binding rule effectively enforced. Assuming that the public good is to everyone's advantage, and one that all would agree to arrange for, the use of coercion is perfectly rational from each man's point of view. Many of the traditional activities of government, insofar as they can be justified, can be accounted for in this way. The need for the enforcement of rules by the state will still exist even when everyone is moved by the same sense of justice.

Before we detail our own response, we’ll note a number of replies we could offer to Rawls that we nevertheless won’t pursue. ⁶ Perhaps Rawls's ideal theory is too concessive to human limitations. We could assume that ideal agents are unconditional cooperators and thus don’t need assurance of reciprocity. Or maybe they are conditional cooperators but don’t face the knowledge limitations that Rawls assumes they do and therefore know that all others will cooperate conditionally. These criticisms may be correct, although there are some reasons for doubt. For instance, one could argue that conditional cooperation is morally better than unconditional cooperation because morality shouldn’t ask you to allow your contributions to be exploited by others. Furthermore, knowledge limitations due to human finitude plausibly don’t count as moral failings and thus shouldn’t be imagined away by an ideal theory that assumes we’re as good as humanly possible.

In any case, our reason for focusing on Rawls's strict compliance view here is simply because it makes our job harder. Under more extravagant idealizing consumptions, the assurance problem doesn’t even arise in the first place and thus the objection can be dismissed easily. However, we aim to show that the objection can be addressed even if we assume Rawls's own more moderately ideal conditions.

Rawls's argument overlooks the ways in which noncoercive, nongovernmental institutions can solve the collective action problem. Many projects are now “crowdfunded” through assurance contracts—benefactors pledge money to a cause or campaign but are not charged unless sufficiently many others pledge as well. Thus, the Alaskan can pledge her share to the redistribution fund knowing that the money will only be taken in the event that Texans pledge their share, too. Such assurance contracts function fairly well in non-ideal conditions; presumably they would function even better under ideal conditions.

Dominant assurance contracts are even better. ⁷ To illustrate, consider Hume's (1978, p. 538) meadow case:

Two neighbours may agree to drain a meadow, which they possess in common: because it is easy for them to know each other's mind; and each must perceive, that the immediate consequence of his failing in his part, is the abandoning of the whole project. But it is very difficult, and indeed impossible, that a thousand persons should agree in any such action; it being difficult for them to concert so complicated a design, and still more difficult for them to execute it; while each seeks a pretext to free himself of the trouble and expense, and would lay the whole burden on others.

To obtain the contributions, she proposes the following arrangement. If you contribute the $100 and too few others contribute to meet the $100,000 threshold, your $100 is returned with a bonus of $10. Thus, if you expect too few others to contribute, you ought to contribute. Alternatively, if others contribute and you’re nearing the threshold, you ought to contribute to ensure the threshold is met and the good is provided given that good's benefit to you exceeds the cost of your contribution (by hypothesis). Contributing is thus the dominant strategy. You profit by contributing in the “no one gets it” scenario because of the refund bonus. You also profit by contributing in the “everyone gets it” scenario because the personal benefit provided by the public good exceeds the personal cost of your contribution. The public goods problem is solved without coercion.

Of course, dominant assurance contracts don’t work as smoothly for larger scale public goods. There is still an incentive to withhold a contribution if the pool of prospective contributors is unknown. You might not contribute if you expect the next person to contribute, thereby enabling you to receive the benefit without paying the cost To take an artificially precise case to fix ideas, suppose some ecologically-minded entrepreneur seeks contributions from five million people to fund widescale deployment of carbon capture technology to fight climate change. You see that 4,999,999 people have already contributed. Even though, let us suppose, the welfare gain to you of mitigated climate change is greater than the welfare loss brought about by contributing to the project, you suspect that another person will make that final contribution. In this case, you get the benefit of a healthier climate and you get to keep your money.

Yet dominant assurance contracts may self-correct for this sort of problem. If you think that contributions will stall as the threshold nears because of the free rider problem, then you have an incentive to contribute in the expectation of receiving the bonus refund that results from the failure of the project. But of course if the prospect of free riding incentivizes people to contribute in this way, the free riding problem is thereby solved. Furthermore, the contract could incorporate features that ameliorate this problem—perhaps the person who makes the threshold-meeting contribution receives a lucrative bonus prize. ⁸

In any case, this problem would not arise in ideal conditions, however much of a worry it would be in the real world. Remember: just agents prefer to contribute. They don’t want to free ride. They will do their fair share “to promote just schemes” when they “believe that others, or sufficiently many of them, will do theirs” (Rawls 1999, p. 236). So if they see others contributing and getting close to the threshold, they’ll gladly pitch in their fair share.

Maybe expecting millions of people to contribute a pledge to pay for a public good via a dominant assurance contract seems far-fetched. But once again, it's important to remember that we are considering ideal conditions that are themselves far-fetched—we’re considering a scenario in which hundreds of millions of people are assumed to be fully compliant with justice. More to the point, expecting millions of ideal moral agents to contribute a pledge to pay for a public good via a dominant assurance contract seems no more implausible than expecting millions of ideal moral agents to contribute a vote for a tax that will pay for that same public good.

Here's another objection: prospective contributors will withhold contributions out of a concern that the entrepreneur won’t make good on her plans and will take the money and run. ⁹ We have several replies.

First, the relevant agreements can be structured in a way that mitigates this sort of risk. For one, the entrepreneur could start with low-cost, small-scale projects where the potential loss is small (e.g. a single carbon capture device) and build trust with repeated interactions. It might be objected that prospective contributors won’t contribute even in this scenario because the possibility of any loss is enough to dissuade them from contributing. The problem with this objection is that it applies with equal force to ideal statists. They face the same assurance problem, but in the political realm. ¹⁰

To see why, consider the predicament of the Virginian at the dawn of a presidential election. She is happy to spend time and resources becoming an informed and unbiased voter so long as Arizonans do the same. But there's the rub: she has no assurance that Arizonans will do the same, just as she had no assurance that they will buy hybrids or donate to redistribution. Thus, she decides not to put in the work to cast a thoughtful vote out of a fear of having her contribution to a public good go unreciprocated and wasted. Alternatively, if she figures that enough other voters will contribute their thoughtful vote for effective climate change strategies, then she can reap the benefits of their contributions without contributing herself (just as the prospective dominant assurance contract contributor could withhold her contribution in an effort to free ride). Or she might withhold her informed vote because she worries that the candidates are untrustworthy and will not make good on their promises (that is, the same worry that applies to the entrepreneur). In short, for the statist solution to work, it must be the case that people are willing to accept some risk of a small loss, in which case the anarchist solution would work as well.

It could be the case that voting well in ideal conditions will not be very costly. ¹¹ After all, we assume that everyone is just and so there's little need to inspect the justness of any of the candidates. However, voters would still need to acquire empirical information to judge the comparative effectiveness of each candidate's proposals (e.g. whether nuclear power is a better bet than solar and wind for combatting climate change). ¹² Moreover, the assurance problem resurfaces here: if people are unsure about the motives of others, then voters would need to research the track record of candidates to ensure that they have a history of following through on their campaign promises. In any case, the ideal anarchist could simply stipulate that the cost of people's voluntarily contributions to a dominant assurance contract will be no greater than the cost of people's voluntary votes under ideal statism.

So, at most, ideal anarchists and ideal statists find themselves at a stalemate. But this uncertainty weighs against the establishment of a state in ideal conditions. The reason is because institutionalized coercion of the sort that characterizes the state is presumptively wrong. When all else is equal, we should prefer voluntary cooperation to state coercion. ¹³ Indeed, as Rawls himself argues (2001, p. 44), “There is a general presumption against imposing legal and other restrictions on conduct without a sufficient reason.” Samuel Scheffler (2010, p. 154) elaborates on this idea, writing, “Coercion always requires justification, and this requirement is particularly urgent with respect to the coercive political power of the state . . . [G]iven the status of individuals as free and equal, the establishment of coercive institutions poses a special justificatory problem.” Although this is not the place for a full-fledged exploration of the presumptive wrongness of coercion, we’ll suggest that it has something to do with the way in which coercion clashes with the basic moral equality of all persons in virtue of subordinating the will of the coerced to the will of the coercer.

In reply, one might argue that the presumption of liberty is only a tiebreaker if the severity of the assurance problem is roughly the same in anarchist and statist conditions. ¹⁴ If we expect that the problem will be significantly worse under anarchism than statism, then the presumption of liberty could be overridden.

However, there is good reason to doubt a significant difference in severity. For one, it's assumed that people will be equally trustworthy and motivated to cooperate in the two scenarios. Moreover, contributions in both scenarios risk going unreciprocated or exploited. Indeed, if anything, the prospect of a payoff in the event of inadequate contribution supplies actors in the anarchist condition with a stronger incentive to contribute than voters in the statist condition.

Lastly, we take seriously the possibility that the assurance problem wouldn’t even arise in the first place. In ideal conditions, the entrepreneur won’t be motivated to run off with the contributions—the objection is that prospective contributors won’t know this and thus won’t contribute. However, we think it's likely that in ideal conditions people would become aware that everyone is fully compliant with justice even if they don’t possess superhuman cognitive powers. It would be strange if every human being in society were fully compliant with justice but this fact was lost on everyone. By analogy, human beings with their existing cognitive limitations are aware that all other humans are mortal. We see little reason why human beings with their existing cognitive limitations wouldn’t become aware that all other humans are fully compliant with justice. Thus, it's plausible that the assurance problem that troubles Rawls would simply not arise in ideal conditions.

§3 Estlund

David Estlund has recently produced a systematic defense of non-concessive moral theorizing. (He avoids using the word “ideal theory”.) Estlund notes that most political theories make demands of people which we know are unlikely to be met. Just war theories ask leaders to forbear from offensive action, but we know ahead of time world leaders will initiate aggression and claim they were justified even when they are not. We know most people are too selfish to donate most of their income to the poor. We know that people will indulge racist biases rather than overcome them. And so on.

Nevertheless, Estlund argues, none of these foreseeable behaviors prove that our normative theories demand too much or that the theories are defective. As Cohen argues, everyone agrees that “ought implies can”; if a person cannot X, they are not obligated to X. But we should avoid confusing the claim that a person very much does not want to do X, or finds X demanding, or is inclined against doing X, with the claim that he cannot. “I do not want to X” rarely implies “I cannot”.

In Utopophobia, Estlund looks more carefully at the issue of motivation, trying to determine under what circumstances people's attitudes and motivations constitute genuine barriers which block or otherwise relieve them of what would have been motivations. He argues these situations are rare. Here, we will briefly review the cases he considers and his arguments against them.

For one, Estlund argues that positing human selfishness typical adds nothing in estimating people's moral requirements. Bill might be too selfish to clean up his mess, but that does not by itself excuse him from cleaning. His selfishness does not explain away his obligations; rather, it merely explains why he fails to meet his obligations (Estlund 2020, p. 102).

Estlund adds that nothing changes if we note such selfishness is common and typical. If most people are too selfish to clean up the mess they impose on others, it does not follow they have no obligation to clean. It follows that most people act badly, and the reason they act badly is that they are selfish (Estlund 2020, pp. 102–3).

Estlund admits that conflicting motives might be part of an explanation about why a person is relieved of a duty, but then contends these motives matter only in special cases where a person has other obligations which trump the putative obligation in question. For instance, he says, imagine a case where if a psychiatrist divulges her patients’ medical secrets, this will somehow promote socially just outcomes. Suppose the psychiatrist is strongly motivated not to divulge such secrets. Here, the psychiatrist may have no such obligation to promote social justice by divulging secrets. But her unwillingness is not what justifies her per se, but rather her pre-existing obligations to her clients which enjoy lexical priority over the demands of social justice (Estlund 2020, pp. 103–4).

Estlund also warns us not to confuse the question of personal moral prerogative with questions of motivation. Many theorists ¹⁵ argue that individuals have a wide sphere of personal prerogative to live their lives as they see fit; they are not required to spend all of their time and effort dedicating themselves to promoting utilitarian ends, human welfare, or the ends of justice. Though Estlund officially remains neutral about the content of morality and justice in Utopophobia, he accepts that prerogative might well block what otherwise could have been duties to act in certain ways. For instance, perhaps we the authors would have better served social justice or human welfare by becoming medical researchers, but it was in our prerogative to choose less utility-enhancing jobs as philosophers. Perhaps Bob would somehow promote social justice by marrying Tom, but it is his prerogative to marry Ed instead. And so on. However, the grounds for thinking people have such prerogative is not that they are callous, selfish, or unwilling to do what's right. Rather, theories of prerogative generally hold that people's interests in living a free and autonomous life of their own design grounds such personal prerogative. Once again, “I don’t want to X” is not a general reason to hold “I have no duty to X”.

Estlund claims that contrary motives block obligations only in special cases, i.e., only when the motive is “clinical”. If a person genuinely cannot choose to do something, due to some psychological disability, the obligation is blocked. But the person must be genuinely disabled. We the authors do not want to donate half our incomes, but we certainly could. If a person could choose to do something, but due to a psychological condition, this would impose severe hardship, then the person may be excused. For instance, perhaps a person might normally be obligated to pull a child from a well, but is at least partly excused when she refuses due to her severe phobia of enclosed spaces and the dark (Estlund 2020, pp. 107–111).

Overall, we must be careful not to treat contrary motives as if they disable people from performing various actions. Otherwise, we threaten to cast people as if they lacked the appropriate kind of free will or moral agency. Rich could give to poor, even though he does not want to. People could avoid free riding on public goods, though they are tempted to cheat. People could work for others’ benefit without expecting a selfish return, though they do not want to. Business owners could play fair and avoid shady practices, though they are tempted to behave badly (Estlund 2020, p. 112). Appealing to contrary motives, such as wealth, status, self-interest, bias, leisure, love, or fear does not block a moral obligation, Estlund says, unless the person is genuinely disabled by these motives, or unless there is an independent principle of justice showing that obligations arising from such motives trump other principles of justice.

Though Estlund has written a systematic critique of concessive political philosophy, he is not an anarchist, even at the level of non-concessive theory. This seems strange, as it burdens Estlund with articulating grounds for the state which are not based on or reactions to people's moral failings.

Estlund's only developed argument for the state appears in Democratic Authority, which, among other things, defends a democratic state against an anarchic alternative (Estlund 2008). Estlund first argues that we should choose whichever of three decision-making methods—anarchy, random lotteries, or the democratic nation-state—tends to produce the overall most just outcomes. He then argues that the democratic nation-state will outperform the other two.

Here's what we find puzzling about Estlund's argument for statism: the central case he uses to motivate the need for the state involves distinctly morally bad behavior. Consider this excerpt:

Ms. Powers, who owned one of the community's general stores, was seen by at least a dozen people (so they say) sneaking out the back of Faith Friendship's general store, with which Ms. Powers's store competes for customers, just before Mrs. Friendship's store burned to the ground. This struck many people as less than surprising, Ms. Powers being a ruthless businesswoman when she isn’t busy entertaining one man or another. This was a year ago, and Ms. Powers has since found it impossible to live a decent life in Prejuria, since no one will talk to her, do business with her, or intervene when she is verbally or even physically accosted, which often happens if she goes out in public. She reasonably fears leaving her house now, and lives on the meager provisions she makes herself. It so happens that this roughly corresponds to the punishment that is known, in the public rules, to be associated with the crime of which she is accused: extended imprisonment. Everyone realizes, though, that she is also in danger of being killed by some of the community's rougher elements (Estlund 2008, p. 138).

Yet Estlund's description of anarchic Prejuria is decidedly not utopian. It describes people who are not merely imperfect, but apparently downright nasty, worse than most people we personally know. ¹⁶ They act in morally deplorable ways. Their suspicions of one another make sense only if they have a long history of moral vice. Indeed, it would make little sense to accuse Ms. Powers of arson if people were always morally good—in this case, no one in Prejuria would be motivated to perform an injustice like arson.

Accordingly, if Estlund intends, as he should, to do non-concessive theorizing, he should describe the situation more like this:

Ms. Powers, who owned one of the community's general stores, was seen by at least a dozen people walking out of the back of Faith Friendship's general store, with which Ms. Powers's store competes for customers, just before Mrs. Friendship's store burned to the ground. This struck many people as a total coincidence since Ms. Powers is well-known for being a kind, honorable, and honest businesswoman when she isn’t busy entertaining one man or another. [We don’t see any reason to change her social life in this non-concessive example, since we don’t regard having an active sex life as morally problematic.] This was a year ago, and Ms. Powers has since continued to lead a wonderful life in Prejuria, with everyone continuing to talk to her and do business with her, and with no one ever verbally or physically accosting her or anyone else. Since everyone in Prejuria always acts rightly all the time, the citizens of Prejuria quickly dismissed the possibility that Ms. Powers burned down her competing store. Indeed, subsequent investigations showed that she was indeed invited by her competitor over for a friendly discussion, much as two athletes might be friendly to each other despite working for rival teams. Instead, her rival store had a freak electrical problem. Her rival store owner was of course perfectly conscientious and already had insurance, but if she had somehow faultlessly gone bankrupt, everyone would have chipped in and gotten her back on her feet, all without any need for coercive taxation.

Indeed, even when Estlund goes on to explain why a state would handle criminal trials better than anarchy, his argument works only because he describes the people in anarchy as suffering from various self-serving biases, vengeful attitudes, free-riding behaviors, and so on. But if we are doing non-concessive theory, then we imagine no one would act in these ways because these behaviors are, by Estlund's own hypothesis, bad and wrong. We will not belabor this point, though, because the problem of criminal punishment would not arise in the first place.

Estlund might reply that properly motivated people may nonetheless have honest disagreements, thus justifying the move away from anarchy. In contrast to anarchic Prejuria, Estlund (2008, p. 138) proposes Juristic Prejuria:

A group of citizens has invented a system whereby a panel of six citizens is randomly chosen and asked to hear the case against any accused person and the case in that person's defense, and to make a decision as to whether the named punishment shall be imposed or not.

Estlund (2008, p. 129) argues that the juristic alternative is superior to the anarchic alternative because, “in Prejuria, there is no qualified disagreement with the proposition that the jury system will be more likely to promote substantive justice than the anarchic arrangement, and also better than a random procedure for choosing decisions.”

Estlund, like Rawls, overlooks non-governmental institutional solutions. He presumes that anarchism implies the absence of judicial or decision-making institutions, despite ample empirical work showing that even real-life, morally flawed people can create and maintain non-governmental dispute resolution mechanisms without relying on the state, even as a reserve or final arbitrator. ¹⁷ Powers and Friendship could, for instance, agree to private arbitration whenever they find themselves with an honest disagreement that they cannot resolve themselves.

Estlund thus must provide an example of a dispute morally flawless people would have that they could not resolve without being subject to a monopolistic institution which threatens them with violence, and, further, must show this case is so important that people would want a state. It must therefore be the case that these supposedly ideal people would, in the absence of such violent threats, choose to act in such ways that it would be justifiable to subject them to state violence. It's a tall order. It appears that the best parts of Estlund's argument for the ideal state largely dissolves into or relies on Kavka's argument, which we consider in the next section.

As noted above, private arbitration is successfully used in the real world with flawed people. The case for this sort of nongovernmental solution is even stronger when we make no concessions to people's predictable moral failings. (We could even imagine that private arbitrators undergo intensive debiasing and educational training to correct for epistemic flaws.) Creating a monopolistic institution which threatens people with violence in order to ensure their compliance is presumptively wrong. We would want to refrain from issuing threats of violence unless we absolutely have to. Just and good people, presumably, would recognize this point, and further, would recognize that they are morally obligated to seek peaceful, fair, and non-violent means of dispute resolution rather than to come to blows or to demand a violent Leviathan rescue them.

You may still have your doubts. But as was the case with Rawls, uncertainty about the comparative effectiveness of ideal statism and ideal anarchism weighs in favor of anarchism. Estlund himself (2008, p. 33) argues that “no one has authority or legitimate coercive power over another without a justification that could be accepted by all qualified points of view.” In the case of morally flawless private arbitration, forming a state that exercises coercive power without compelling reason to believe that it will outperform the voluntary alternative is subject to qualified denial.

§4 Kavka

Like Estlund, Kavka argues that the state is needed for dispute resolution. Even in ideal conditions, people might retain reasonable disagreements about what justice requires (Kavka 1995, p. 3). People can be honest and fair-minded and yet still arrive at different conclusions. Kavka is skeptical that they could satisfactorily resolve these disagreements without the state.

Before giving our reply, we’ll once again note a response that we could—but won’t—make to Kavka. In brief, there's a case to be made for the claim that morally perfect people would never have moral conflicts. Kavka seems to equate moral perfection with something like moral blamelessness. Two people could be perfectly motivated to find the truth about morality and act in accordance with it but nevertheless disagree about what the moral truth is simply because they’re fallible beings.

But an alternate conception could specify moral perfection in terms of acting in a morally flawless way. Suppose a severely dehydrated person knocks on your door in desperate need of water. You give them a bottle of clear liquid marked “water” from your refrigerator with the intention of saving their life. It turns out that the manufacturer mistakenly filled this bottle with clear poison instead of water. You had no way of knowing this, but your action nevertheless results in the death of the stranger. Perhaps your action was morally wrong even though you are not to blame for performing it.

Here again, the reason we don’t pursue this line of objection is simply because it would make our job too easy. If morally perfect beings always get things right and never disagree, then Kavka's argument for the state never gets off the ground. We aim to show that even on Kavka's own account of moral perfection, his argument for the state is unsuccessful.

As an initial point, we’re more optimistic about nongovernmental methods of dispute resolution than Kavka seems to be. There are plenty of strategies for resolving disputes that do not require the state's involvement. Suppose your neighbor runs a small woodworking business out of her garage. ¹⁸ All of that sawing is loud, though, and it interferes with your nap in your hammock. How will this conflict be resolved?

Maybe you really value napping in hammocks and your neighbor just barely prefers woodworking over, say, oil painting. In this case, you could offer to pay your neighbor to switch to painting. If you value the nap more than she values woodworking, you’ll offer enough financial compensation to motivate her to switch. If, on the other hand, she values woodworking more than you value the nap, she won’t accept the offer. Either way, we have a nongovernmental resolution of the conflict.

In reply, one might argue that it's unfair that you have to pay your neighbor to stop woodworking to get a nap, while your friend who lives in a quiet neighborhood does not. ¹⁹ However, your neighbor could lodge a similar complaint if, say, the state forces her to stop woodworking—it's unfair that she has to stop whereas her woodworking friend whose neighbor doesn’t mind the noise doesn’t. So it's unclear how a governmental solution will help here. Moreover, if these inequalities are unjust, then presumably ideally just people will rectify them via voluntary redistribution because they are, by hypothesis, fully committed to justice.

Or consider again the possibility of private arbitration. Sometimes baseball players and their teams cannot agree on what would count as a fair salary. This disagreement need not be due to any blameworthy character traits—the parties may just have an honest difference of opinion about the value of the player's contribution. Thus, this is the sort of conflict that may arise among ideal agents of the sort envisioned by Kavka.

Here again, the dispute is resolvable within anarchist conditions. The way that such disagreements are in fact settled in the real world (let alone a world of moral angels) is to allow each party to present their case to a panel of private arbitrators, who then render a salary decision. Thus, the parties could resolve their dispute by submitting to the decision of a mutually agreed upon, nongovernmental arbiter. Indeed, if they are moral angels as Kavka assumes them to be, we should expect them to peacefully accept the decision. Perhaps, in Kavka's words (1995, p. 9), ideal moral agents have a “willingness to settle voluntarily practical disagreements by discussion and compromise.”

In response, Kavka argues that there will still be instances where ideal moral agents could value some other moral principles more highly than they value peaceful compromise, in which case disagreements may become intractable and violent. Furthermore, even if most of the ideal agents do value compromise more than anything, they’ll be vulnerable to invasion by an “uncompromiser”—that is, an agent “who will not settle a first-order dispute except on his own terms” (Kavka 1995, p. 11). Over time, uncompromisers will defeat compromisers in cases of moral conflict and eventually being uncompromising will displace compromising as the guiding strategy for the population. Here Kavka (1995, p. 11) assumes that “if (and to the extent that) a disposition works, more people tend to adopt it. That is, either people tend to convert from less successful to more successful strategies, or new members of the population are more likely to adopt the strategies that have been more successful (or both).” The result, then, is a population of uncompromisers locked in conflict with one another. Kavka (1995, p. 12) laments, “How sad that this decline, from the benign anarchy of compromise to the painful condition of festering disputes, is inevitable.”

Kavka argues that a government-imposed resolution of a dispute will end social conflict among angels. For one, the government is the final authority for enforcing settlements, “ensuring that if one acquiesces in a compromise arrangement, one can reasonably expect others to follow its terms so that the elements of the compromise you approve of, as well as those you do not, will materialize” (Kavka 1995, p. 16). Moreover, even morally flawless humans will be incentivized to relent by threat of legal punishment. Lastly, their knowledge that their opponents will be similarly coerced intro agreement assures them of receiving at least part of their preferred terms (Kavka 1995, p. 16).

Before offering a criticism of Kavka that addresses his argument on its own terms, let us register an initial worry. Kavka's scenario differs from Rawls's and Estlund's along a crucial dimension—the introduction of uncompromisers suggests that we are no longer contemplating ideal conditions that make no concession to moral imperfection. We suspect that truly ideal conditions would exclude those who steamroll everyone else to get their way.

The puzzle here is that in order to motivate the need for a state, Kavka has to describe apparently angelic people who have such intransigent conflicts that they are willing to act in ways so horrific or bad that other angelic people would find no superior mechanism to stop such horrors other than creating a monopolistic institution would threatens everyone with violence. Further, Kavka has to say that the original uncompromising angels would not themselves recoil at this result and take it as strong evidence they are acting in morally bad ways. They would not say, “Wow, we are acting in such a bad way that others have to create a state to stop us. We should simply stop.” This seems like an incredible result. On the contrary, the conflicts Kavka describes seem morally bad and morally good people would want to either avoid those conflicts or resolve them in peaceful ways.

Thus, the challenge for Kakva is to describe a situation in which it is morally permissible for people to steamroll each other or come to blows, and yet also morally permissible for other people to create a state to stop them (rather than using other means). But even if he could show all this is indeed permissible, that would not be enough. He has to also show that morally good people would not find this situation so undesirable—for the very reasons Kavka offers on behalf of a state—that they would voluntarily choose to avoid it in the first place.

But perhaps we’re wrong about this and such uncompromisers would exist among moral angels. The case for statism remains in doubt, as the introduction of uncompromisers comes back to haunt Kavka's own argument for the state. ²⁰ The creation of a state merely shifts the fight between uncompromisers from the private sector to the ballot box. The zealots the state was introduced to tame are the very agents most motivated to take control of the state itself—either to force others to comply with their moral principles or to avoid being forced to comply with others’ principles. Geoffrey Brennan and James Buchanan (2008, 10.4.53) write:

Suppose that a monopoly right is to be auctioned; whom will we predict to be the highest bidder? Surely we can presume that the person who intends to exploit the monopoly power most fully, the one for whom the expected profit is highest, will be among the highest bidders for the franchise. In the same way, positions of political power will tend to attract those persons who place higher values on the possession of such power. These persons will tend to be the highest bidders in the allocation of political offices […] Is there not the overwhelming presumption that offices will be secured by those who value power most highly and who seek to use such power of discretion in the furtherance of their personal projects, be these moral or otherwise?

To illustrate, take a simple case. Suppose that members of the Social Justice Party and the Honored Tradition Party find themselves locked in an intractable moral conflict. An election is held to determine which party will take hold of their society's political institutions and thus get their way in the disagreement. Suppose further that the vote count indicates that the Social Justice Party has won. Would the zealous members of the Honored Tradition Party passively and peacefully accept the results of this democratically-decided conflict resolution procedure? If so, then it's unclear why they wouldn’t passively and peacefully accept the results of a nongovernmental democratic decision.

On the other hand, we might expect that the zealots’ commitment to their own moral agenda will override their commitment to compromising and abiding by the results of a democratic election. After all, they are, by hypothesis, more committed to getting their own way than they are to peaceful compromise. In this case, the ideal statist is not entitled to assume away the possibility that the Honored Tradition Party will, for instance, violently contest the election results that would prevent them from zealously pursuing their moral agenda. The proposed solution to the conflict can itself become an object of conflict, and potentially a violent one at that.

Here again, we believe that a “tie” between statism and anarchism goes to anarchism. Establishing an institution that exercises a monopoly right to coerce is, we’ve argued, presumptively wrong. The burden of justification is on the statist to provide compelling reason to believe that the formation of a coercive state will provide a significant advantage over anarchy. Indeed, we think that the presumptive wrongness of a state is particularly clear in the case of Kavka's uncompromisers, given that it will predictably become a tool used by the more powerful to subordinate the less powerful for the reasons mentioned above.

As noted earlier, though, if we expect the problem of moral conflict to be significantly more severe under anarchism than statism, then the presumption of liberty may be overridden. But there's reason to doubt that the problem of moral conflict will be significantly more severe under ideal anarchism. Indeed, there's reason to believe that the creation of a centralized institution with monopoly right to coerce will worsen moral conflict. Attempting to acquire that monopoly right is the dominant strategy for both/all sides of a moral dispute. By way of example, if uncompromising proponents of secular schooling don’t enter the political fray, uncompromising advocates of intelligent design ought to lobby the state in order to impose the moral agenda to which they are zealously committed. If uncompromising proponents of secular schooling do enter the political arena, uncompromising advocates of intelligent design still ought to lobby the state if only to avoid being imposed upon by the secularists. And vice versa.

Under ideal statism, citizens have an incentive to fight for control of the political apparatus that picks the “winners” and “losers” of moral conflicts to ensure that they’re among the winners (indeed, truly zealous citizens may literally fight over the political apparatus, for instance, in the form of violent contestation of election results). In the absence of a centralized institution with monopoly right to coerce, both sides may be more likely to retreat to their own separate corners and thus minimize the chances of a moral conflict.

Of course, the argument that a state will exacerbate rather than alleviate moral conflict is admittedly speculative. But so too is the argument that a state will alleviate rather than exacerbate moral conflict. Thus, we’re left with speculation on either side. We contend that in the absence of anything close to a decisive victory for other side, the presumption against coercion ought to tilt us toward anarchism.

§5 Conclusion

A fully just society is one in which all contribute to public goods voluntarily, donate generous sums to charity from the goodness of their hearts, and seek peaceful ways to cooperate with one another. It would not rely on coercion or threats of coercion. Whatever one thinks of its feasibility, a fully just society would operate in accordance with anarchist principles rather than statist principles.

Rawls's argument for ideal statism overlooks the ways in which nongovernmental arrangements can solve assurance problems; moreover, he overlooks the possibility that assurance problems wouldn’t arise in the first place in ideal conditions. Estlund's case for the claim that a society without moral failings would not be anarchistic is unsuccessful in part because he unwittingly invokes moral failings when motivating the need for a state. Lastly, Kavka assumes that the state is needed to solve moral conflicts among even morally blameworthy people, but doesn’t recognize that the state itself may worsen rather than improve conflict.

To the extent we have reason to be skeptical of nongovernmental solutions to problems that may arise in ideal conditions, we have equal reason to be skeptical of governmental solutions. But we have argued that the “tie” goes to nongovernmental solutions: establishing a monopolistic institution that threatens people with violence to secure their compliance is presumptively wrong and the burden of justification rests with those who would establish it. We conclude that a theory that concedes no space for our moral failings concedes no space for the state either.