Does public reason liberalism lead to anarchy? Moral disputes and attitudes to risk

Is it ever legitimate for the state coerce you; and, if so, when? That is, when, if ever, is it legitimate for the state to impinge upon your freedom by forcefully preventing you from doing one thing or forcefully ensuring that you do another? According to the public reason liberal, such as the Gerald Gaus of The Order of Public Reason, it is legitimate when: (i) there is some structure in place of which this act of coercion is a particular result, such as a legal system with specific laws and a group of people tasked with enforcing them using coercion if necessary; and (ii) each reasonable person subject to this system has sufficient reason to endorse it.

Of course, there is much to be filled in here. Who counts as reasonable? And when does someone have sufficient reason to endorse such a structure? We’ll return to the first question a little below, but let’s begin with the second since it’s more central to the worry I wish to explore. Two features of the public reason liberal’s account stand out. First, a person can have sufficient reason to endorse a structure even if they haven’t actually done so and even if they wouldn’t endorse it if asked. What is required is rather that some idealised version of the person endorse it in some hypothetical situation. David Enoch has criticised this part of the account at length (here and here), and I have little to add to his treatment. I mention it here only because the objection I wish to raise goes through even if you aren’t moved by Enoch’s arguments.

So let’s talk about the second feature of the public reason liberal’s account of sufficient reasons. To have a sufficient reason to endorse a structure, you needn’t prefer it to everything else; indeed, you can prefer plenty of other things to it, and pretty strongly. All that is required is that you prefer it to anarchy. After all, the public reason liberal is primarily interested in justifying deviations from anarchy; they wish to distinguish between those cases in which your liberty is restricted legitimately and when it is restricted illegitimately. And their defining claim is that it is restricted legitimately when it the restriction is a result of a structure that each person would prefer to the state of anarchy.

It is because they set the bar for sufficient reason so low that public reason liberals think they avoid the so-called anarchy or empty set objection, which says that there are no structures that everyone has sufficient reason to endorse, because people’s preferences are too diverse. Surely, the public reason liberal responds, there are structures that everyone prefers to anarchy, that most benighted situation from which we surely all wish to flee; while our preferences are diverse and while every structure is therefore going to displease at least some members of the public, surely they will accept such displeasure in order to avoid anarchy.

Of course, by lowering the bar so far, they invite a different objection: there are many illiberal structures that everyone has sufficient reason to endorse if all that means is preferring it to anarchy. Even members of an oppressed group and those who care about those members will prefer a moderate degree of oppression or exclusion or poor treatment over the situation in which there are no protections at all. A state that excludes gay men from employment in the public sector but imposes no further restrictions on them is a horrible state. But it wouldn’t be unreasonable for me, as a gay man, to prefer it to anarchy. And indeed it may well be that everyone prefers it to anarchy. But it seems strange for a liberal to conclude that any coercion used to enforce this employment prohibition is legitimate.

Nonetheless, this is not the objection I want to focus on here. Rather, I want to argue that, even with bar for sufficient reason set so low, even with the restriction only to the reasonable, and even if we wish to ask only idealised versions of members of the public in hypothetical situations, there is still no structure that each member of the public has sufficient reason to endorse, at least for many publics we find in our world. That is, setting the bar so low in fact doesn’t avoid the anarchy or empty set objection.

There are two reasons for this. First: there is enough diversity of moral views among the public that, for some things people might do, there will be both members of the public who prefer anarchy to a structure that prohibits them and members of the public who prefer anarchy to a structure that permits them. Second: some members of the public see anarchy as a high-risk, high-yield possibility, where there will be big winners and big losers, and they are sufficiently risk-seeking that they are prepared to risk being one of the losers in order to open the possibility of being one of the winners.

It might help to begin by thinking more abstractly about how the set of structures that everyone has sufficient reason to endorse might end up empty on this proposal. First, there might be some property that a structure either has or it lacks, and there are members of the public that rank anarchy above any structure that has the property and there are members of the public that rank anarchy above any structure that lacks the property. Second, there might be a whole list of properties that structures either have or lack, and for each of these properties, there are members of the public who prefer anarchy to any structure with that property, and there are also members of the public who prefer anarchy to a structure with none of those properties. Third, there are members of the public who prefer anarchy to any alternative structure. The sort of moral diversity we find in actual publics gives rise to the first two ways in which anarchy might follow from public reason liberalism; the sort of risk-seeking I described above gives rise to the third.

1. Moral disputes

Here’s how the first structure might arise. What we need is a sort of action or practice for which there are some who disapprove of it so vigorously that they would prefer anarchy to any legal structure that permits it, and for which there are others who value the freedom to engage in it so strongly that they would prefer anarchy to any legal structure that prohibits it. How can this be?

Let’s begin with a fictional example of extreme disagreement over a religious practice to see how it might arise, and then ask whether there are cases in the real world. It is easy to imagine a religious practice or rite or ritual that has developed in two different ways in two different denominations of the religion—let’s call them version A and version B—where the practitioners of A consider B an abomination in the sight of God, while the practitioners of B consider A just such an abomination; indeed, there are and have been such practices in reality, but we needn’t consider a concrete case. Now, the fact that the practitioners of A consider B an abomination does not necessarily mean they prefer anarchy to any structure that permits B. It depends upon the intensity of their disapproval, but also whether they believe that this is their battle to fight. Some believers might think that God will mete out the punishment to B’s practitioners and they should merely stand out of the way; but others might think it is their duty to wipe out the alternative practice. And this is the root of their preference for anarchy. For something that is often missing from discussions by public reason liberals is the second-order coercion involved in a structure that permits a particular practice. Consider a structure that permits B. What that means is that this structure doesn’t coerce people into abstaining from B, and so it appears on its surface less coercive. And yet, by including this permission, it usually introduces second-order coercion, for it will usually include a further law that prohibits people from forcefully preventing others from practising B; indeed, it will usually include a further law that prohibits people from forcefully preventing others from doing anything that the first-order laws permit. So what is often missing from these discussions is the recognition that permitting a practice is nearly always accompanied by prohibiting people from forcefully preventing others from engaging in that practice. When a law permits something, it doesn’t say only that the state won’t prevent people from doing it; it says also that the state will prevent others from trying to prevent people from doing it. This is what is meant by saying that the state has a monopoly on violence: it is only the state that can forcefully prevent members of the public from acting in particular ways, and the laws of the state lay out the ways of acting that will prompt such forceful prevention; so, if something is permitted, the state won’t prevent it, and since it has a monopoly on such prevention, it will prevent anyone else from trying to prevent it.

Now, the public reason liberal will say that this is how it should be, and I don’t deny that. But it does reveal why the practitioners of A, if sufficiently opposed to B, might prefer anarchy to a structure that permits B. For, in the state of anarchy, they can try to prevent B without the threat of the state’s coercive power. Of course, that’s not to say that they won’t meet with resistance: practitioners of B won’t simply surrender. But the state’s coercive power is awesome. It is backed by the resources of the whole country. And so the practitioners of A might reasonably think they stand a better chance of eradicating B in an anarchic state. Of course, they also thereby run a greater risk of being defeated themselves, for the state won’t be there to defend their practice of A nor to prevent any attempt by the practitioners of B from trying to eradicate them. But if their disapproval of B is intense enough, this may well be a risk they’re prepared to take. Suppose practice B involves actions you consider among the most morally despicable people can commit—and recall that religious practices of the past have involved the sacrifice of humans. Might it not be reasonable to prefer anarchy to a state that throws its full coercive power behind allowing this practice to continue? Gaus notes that, as well as wanting to live in accordance with their own conception of the good, people often also valuable living together in a society. But they do not wish to live together in a society in which morally abhorrent practices are permitted. So sufficiently strong moral disapproval can render anarchy preferable to any society like that.

Perhaps, then, there were religious practices that excited this level of condemnation in the past; but is there anything, religious or not, that plays a similar role today? Do we today have two groups, one who prefers anarchy to prohibiting a practice and one who prefers anarchy to permitting it? It’s an empirical question, of course, and there is often a gap between a person’s professed preferences and those they truly hold, but it seems that at least some current moral and related positions are like this. One source is disagreement over the moral relevance of a particular sort of being. For instance, consider the sort of veganism that is motivated by thinking that all currently farmed animals have substantial moral status that renders it profoundly immoral to keep them in captivity, subject them to suffering, and kill them for food or clothing. 92 billion animals are killed for food every year. For someone who considers them to have moral status, this is murder on an unfathomable scale. And, where such slaughter is permitted by law, this mass murder is protected by the awesome force of the state that permits it. It isn’t hard to imagine that our vegan might prefer a state of anarchy to this. In such a state, they might hope to prevent more animal deaths, since their efforts at prevention would not meet with the full power of the state, though of course it would meet with something else.

Now, the public reason liberal might think that whatever force our vegan revolutionary would meet under anarchy would be worse than the force they’d meet under a liberal state. But this is an empirical question. It is one about which we might be quite uncertain; and, importantly for our purposes here, it is one about which different members of the public might significantly disagree.

This brings us to an important feature of public reason liberalism that raises significant problems for that position. Consider the individual rankings of structures on which the public reason liberal builds their definition of sufficient reason. These are the preferences the members of the public have to which we advert when we want to know which forms of coercion are legitimate. In these preference orderings, the items ranked are what decision theorists call prospects, not what they call outcomes. Suppose I offer you a bet that pays you £10 if it rains tomorrow and nothing if it doesn’t, and I set the price at £3. Then there are three relevant outcomes: (i) you end up with £7 more tomorrow than today; (ii) you end up with £3 less tomorrow than today; (iii) you end up with exactly the same tomorrow as today. The first outcome comes about if you choose to accept my bet and it does rain tomorrow, so that you pay out £3 for the bet and take in £10 because you win it; the second outcome comes about if you choose to accept my bet and it doesn’t rain tomorrow, so that you pay out £3 for the bet and take in nothing because you lose it; and the third outcome comes about if you choose to reject my bet, so that you pay out nothing and take in nothing whether it rains or not. In this situation, there are outcomes: (i), (ii), and (iii). And there are two prospects: Accept the bet; Reject the bet. The things I fundamentally value are the outcomes: they specify how the world is, or will be, in enough detail that I can assign something like a utility to it. The prospects don’t. Rather, I have a probability function that says how likely each of the outcomes is given each of the prospects; I use that to calculate the expected utility of the prospects, which is the value I assign to them; and I rank them by that value. So, for instance, I might assign 70% probability to rain tomorrow, and so assign 70% probability to outcome (i) given that I accept the bet; and so on. In the rankings of structures to which the public reason liberal appeals, the prospects are things like: Anarchy; Libertarian structure with small state, protections of private property and contracts, and little else; Liberal democratic state with such-and-such institutions and so-and-so laws and democratic elections; etc. And the outcomes are the different ways the world might unfold. Different prospects make different outcomes more or less likely. Each individual assigns a probability to each state of the world given each possible structure. How likely is this outcome given anarchy? How likely is it given libertarianism? How likely is it given this particular theocracy? And so on. They then use those, together with the utilities they assign to the outcomes, to rank the structures, probably using expected utilities. However, just as there might be reasonable disagreement about how valuable the different outcomes are, so might there be reasonable disagreement about how likely each of the structures makes each of the outcomes. And it seems plausible that there is a perfectly reasonable way of setting those probabilities such that the vegan revolutionary might rank anarchy above any structure that permits the captivity, suffering, and slaughter of two billions animals because they have judged that on balance it is anarchy that gives them the best shot at minimising these practices.

So hopefully I’ve made it plausible that the morally-motivated vegan might prefer anarchy to structures that permit the exploitation of animals. But I need something more than that. I also need to show that there are also members of the public who prefer anarchy to any structure than prohibits the exploitation of animals. Is that plausible? Are there people who would abandon the rule of law before giving up the possibility of eating meat? I’m not sure. Some people speak as if they do have that preference. But that could be sabre-rattling, and it could simply be that they rank anarchy quite highly anyway, regardless of the alternative; a possibility that we’ll take up below. What we’re looking for at this point is an issue over which two groups are so divided that, while both sides find anarchy abhorrent, they find it less abhorrent than permitting the practice of the other side. And perhaps veganism doesn’t give that. But I suspect there are disputes that do. As I said above, disputes over moral status can often have this form. Perhaps some disputes over abortion have this form: some who sincerely believe foetuses have moral status from very early on in a pregnancy might prefer anarchy to a structure that permits abortion during this stage, since they consider it murder; those who believe foetuses don’t have that moral status, or believe whatever status they have is trumped by the status of the person who is pregnant, might prefer anarchy to a structure that prohibits abortion at all stages, since it constitutes a profound violation of that person’s rights and puts some in the way of significant dangers from the pregnancy. Another sort of dispute that can have this format is a land dispute, particularly when the land disputed has particular cultural or religious significance to both sides: each party might prefer anarchy to a structure that cedes any of the land to members of the other.

Before continuing, I should pause to say that the point I wish to make does not turn on any view of the moral rightness of either side of any of these disputes. All that I need is that there are members of the public with these views, and their rankings are not excluded from consideration by the public reason liberal. Now, of course, the public reason liberal does wish to exclude some people on moral grounds: Nazis and psychopaths are often listed explicitly. But I don’t think they will be able to exclude all such disagreements in that way. If they exclude those who wish to permit animal exploitation, they exclude vast numbers of people currently alive. And it’s hard to see how they would determine whom to exclude in the case of the land disputes. And in any case, as David Enoch has emphasised, once you start to use substantial and moderately controversial moral judgments to exclude members of the public, it becomes less clear why you need public reason liberalism in the first place, since you simply appeal to what you take to be moral truths to determine the legitimacy of coercion: it’s legitimate if it prevents an action that is sufficiently seriously morally wrong; it’s illegitimate if it doesn’t.

So I think there are cases in which one group prefers anarchy to all structures that prohibit a practice while another group prefers anarchy to all structures that permit it. And, if there are, the set of structures that each member of the public has sufficient reason to endorse is empty. But it’s worth noting, before we move on to consider the role of risk, that something weaker gives us the same conclusion. Suppose there are a number of practices, and, for each one, there are members of the public who feel so strongly about it that they prefer anarchy to any structure than permits it. One of them might be the practice of killing animals for food or clothing; another might be non-procreative sex; divorce might be another; as might walking on this sacred site, and worshipping in a particular way on that other site; and so on. Perhaps, for any one of these, taken singly, there’s no member of the public who values the freedom to engage in that practice so strongly that they’d prefer anarchy to any structure that prohibits it, but there certainly are members of the public who prefer anarchy to any structure that prohibits all of them. That is, the members of the public will tolerate some restrictions on their liberties in order to live in a society with those who abhor a certain practice—Gaus is right about that. But they won’t tolerate the extreme restriction on their liberties required to reach agreement with all of those compatriots. In this case, again, there is no structure that each member of the public has sufficient reason to endorse. If the structure permits one of the practices listed, there’s a member who prefers anarchy to it; if it prohibits all the practices, there are other members who prefer anarchy. So, anarchy follows.

Talking of prospects also lets us see why a potential response from the public reason liberal won’t work either. We have been acting as if what the individuals rank are state structures (or the lack of them) together with the laws those states will enforce. But they might say that all we rank are the structures, which provide the procedures by which laws will be chosen, but don’t specify the laws themselves. And, in that case, they might claim, even those who find a particular practice morally abhorrent will prefer a liberal democratic structure that lays out the ways in which it will be determined whether that practice is permitted or prohibited to anarchy, since such a structure opens the possibility that the practice will be prohibited, as they wish. But of course it also opens the possibility that the practice will be permitted, which they find abhorrent. And, providing they think it more likely that anarchy will lead to the eradication of the practice than liberal democracy, they will rank anarchy higher. The point is that drawing attention to the fact that it is prospects and not outcomes that are ranked by members of the public, we see more clearly that their beliefs about the outcomes to which each system is most likely to give rise become as important as the utilities they assign to those outcomes.

2. Risking anarchy

Let me talk a little now about risk. I hope to convince you that it is reasonable not only to prefer anarchy to any structure that permits a practice you find morally abhorrent, but to prefer anarchy to any alternative structure at all. The attitudes, values, and opinions that lead to this preference are not particularly appealing. If I knew someone who had them, I would not rush to be their friend nor hold them up as exemplars to the next generation. But that does not prevent them from being reasonable.

Above, I noted that the items ranked by members of the public are prospects, and I simply assumed that prospects are ranked by their expected utility. But some rational choice theorists deny this. For instance, Lara Buchak thinks they should be ranked by their risk-weighted expected utility, and Chris Bottomley and Timothy Luke Williamson think they should be ranked by their weighted-linear utility. The details don’t matter here. All that needs to be said is that both tell you to rank a prospect by a quantity that is determined not only by the probabilities you assign to the outcomes given that prospect and the utility you assign to the outcome, but also by your attitudes to risk. The expected utility of a prospect is the sum of the probability-weighted utilities of the outcomes. Let’s say that this is the risk-neutral approach. A risk-averse approach would give more weight to the outcomes with the lowest utility (the worst-case outcomes) and less to the outcomes with the highest utility (the best-case ones); and a risk-seeking approach would give more weight to the outcomes with the highest utility and less to the outcomes with the lowest. As a result, a risk-averse person might prefer a less risky option, which is much more likely to give a middling utility, over a more risky option, which has a good chance of giving a high utility but also a good chance of giving a low utility, while a risk-averse person with the same probabilities might prefer the riskier option.

Now, we might imagine a person with the following combination of attitudes to risk, utilities, and probabilities. In their attitudes to risk, they’re risk-seekers. That is, the best-case outcomes loom larger in their decision-making than they do for the risk-neutral who ranks prospects by their expected utility, or the risk-averse person, and the worst-case outcomes loom smaller. As a result, they are more inclined to rank a more risky prospect, which opens the possibility of high utility but also the possibility of low utility, above a less risky one, which opens neither the possibility of high nor the possibility of low but only really the possibility of middling utility.

In their utilities, the person in question cares primarily about their own well-being and that of their family, friends, and community. They might give some weight to the well-being of others, but it is much much lower than the weight they give to their close circle.

In their probabilities, they recognise their safety will be less robustly protected in an anarchic state than in a liberal democracy, but they think it likely they will nonetheless band together as a community to protect one another from outside incursions, which will mitigate this to some extent. And, what’s more, they think the absence of a coercive state will allow them to build up wealth for themselves, their family, friends, and community in ways that a liberal democracy will prevent. After all, they reason that, while protections of private property provide safety for their own belongings and land, they also put the property of others beyond their grasp, and they are willing to risk the loss of their property to have the better opportunity to seize someone else’s and thereby enrich themselves.

Combining these attitudes—the risk-seeking, the limited circle of people whose well-being has substantial weight for them, and the beliefs about how they and their circle will manage in an anarchic structure—we obtain a preference for anarchy above all else. And that, of course, leads to anarchy for the public reason liberal: if there is just one member of the public who ranks anarchy above all else, then there can be no structure that every member of the public has sufficient reason to endorse.

3. What went wrong?

Where has public reason liberalism of this form gone wrong? I think there are a number of lessons from the sort of arguments just given. Perhaps the central one is that they have underestimated the appeal of anarchy in certain situations and for particular people. In part, that is because they have paid less attention to the second-order coercion implicit in any state that permits a practice: that state will then prevent any attempts to prevent that practice taking place. When we combine these facts with the sort of strong disapproval of practices that I was considering above, where the person not only wishes not to engage in the practice themselves, but also thinks that anyone engaging in the practice is a moral abhorrence, we can begin to see the appeal of anarchism. Still, we might think, we wouldn’t rank it above all structures that permit the practice ourselves, since we judge anarchism even less likely to limit the horror of the practice of which we disapprove than some structure that permits it. But the question is not whether we judge this, but what it would be reasonable for any member of the public to judge. And since this seems a case in which there might be reasonable disagreement, it seems quite plausible that someone might reasonably judge that anarchy is their best bet.