I. The Puzzle of Legitimacy

When a state has the right to rule, then it is legitimate, according to the most common understanding of legitimacy. States have the right to rule when they have a certain bundle of rights. At a minimum, this bundle comprises a Hohfeldian liberty-right to enact and enforce (certain kinds of) laws. ^{Footnote 4} Traditionally, the right to rule is conceived as also comprising a Hohfeldian claim-right to be obeyed, with corresponding duties or obligations to obey on the side of the citizens. I will here assume that this claim-right is not necessary for legitimacy. But while the state may not need this claim-right, a mere liberty-right to enact (certain kinds of) laws is not sufficient. ^{Footnote 5} In addition to a moral liberty to enact and enforce (certain kinds of) laws, the state also needs Hohfeldian moral powers to thereby impose moral and legal duties on citizens. In general, a power is a second-order capacity to change people’s first-order status as defined by claim-rights and liberties (and duties and “no-rights” as their respective correlates). A power correlates with a Hohfeldian liability to have one’s first-order status changed. The state needs moral powers to impose duties on citizens because this is what the state does when it enacts laws: it imposes moral and legal duties on citizens to respect these laws. The state’s power to impose duties is acknowledged in most accounts of legitimacy (or “political authority”). ^{Footnote 6} Realists do not seem to disagree with this picture of legitimacy. Thus, John Horton says that “the role of an account of political legitimacy is to explain how it is that a state has the right or, as I shall say, the authority, to govern those who are subject to it.” ^{Footnote 7} And he writes that judgments about state legitimacy imply that the state is “entitled to make claims on me that I as, say, a British citizen, along with other British citizens, take to impose obligations on us.” ^{Footnote 8}

To be sure, sometimes the terms “legitimate” and “legitimacy” are used in a different way. Sometimes “legitimate” is simply used as a synonym for “justifiable,” without any specific connection to the right to rule. Sometimes “legitimacy” is used as a purely descriptive term for states that are acknowledged as states by the international community. But political philosophy and political theory have usually been interested in legitimacy as a normative notion, and as a normative notion that is distinct from the generic “justifiability.”

Some philosophers distinguish between a state’s legitimacy and its political authority. For example, Allen Buchanan regards a state as legitimate when it has Hohfeldian liberty-rights to enact and enforce (certain kinds of) laws, and as having political authority when it has a claim-right to be obeyed in addition. ^{Footnote 9} Since I think that states need more than mere liberty-rights, but less than a claim-right to be obeyed, I use “legitimacy” and “political authority” interchangeably for states that have moral liberty-rights to enact and enforce (certain kinds of) laws and have moral powers to thereby impose moral and legal duties on citizens.

When states are legitimate, they have a certain bundle of rights. But, of course, a state is merely an institutional structure, in which real persons fill certain roles. ^{Footnote 10} Hence in the end, it is particular persons that have the rights that constitute a state’s legitimacy. It is true that these persons have the right to rule only derivatively. Basically, the right to rule resides in institutional roles. But this does not change the fact that real persons must fill these roles in the end, and that real persons, hence, come to have the right to rule that is assigned to a role. One cannot discuss the right to rule of the state without discussing the right to rule that persons have over other persons. It should also be noted that the right to rule is sometimes dispersed over different roles within the institutional structure of the state and, hence, over different persons. Most importantly, though, one part of the institutional structure of a state will be the legislature, which has the right to enact laws and thereby impose duties on citizens. Members of the legislature (whether the legislature is constituted by members of a parliament, or by an absolute monarch) thus command the most important part of the right to rule.

Now the puzzle is how states could actually be legitimate: How could some persons acquire the right to rule over others? ^{Footnote 11} This is a perplexing question because quite plausibly there are no “natural” rulers. Setting aside the case of children and the seriously cognitively incapacitated, all persons can be assumed to be equals. There are no natural relations of authority and subordination among persons. If there are any “natural rights” or “human rights” at all, they are certainly equal rights, held by everyone. No one naturally has a right that other persons lack, and certainly not the right to rule over others. If that is correct, then we have to come up with some account of how states (and particular persons that fill certain roles within states) could come to acquire a right to rule. Yet if persons are conceived as originally having equal rights — as being equals — it seems hard to find such a story. This is the puzzle of legitimacy.

Of course, the Lockean liberal tradition has a straightforward answer to the puzzle of legitimacy: the state’s right to rule is understood as based on the free consent of everyone. If some have the right to rule over others, this right must have been voluntarily transferred to them. If persons originally have the same rights, then only free transfer can lead to an unequal distribution of rights. The obvious problem for this Lockean account of legitimacy is, of course, that such free consent has never been given. ^{Footnote 12} Contemporary Lockeans like A. John Simmons, thus, draw the conclusion that all existing and probably all future states are illegitimate. They lack the right to rule. ^{Footnote 13}

I would like to add four short sidenotes before I come to realist theories of legitimacy that try to avoid the conclusion drawn by consent theorists. The first is that “states” in their modern form have come into existence (more or less) in the nineteenth century. Medieval Russia or ancient Babylon may not properly be called “states” in the modern sense, although they certainly were political communities, and their rulers certainly claimed to have the right to rule. For the sake of simplicity, I will for the most part also speak of “states” when referring to older forms of political community.

Second, solving the puzzle of legitimacy is not identical to explaining political obligations. Political obligations are what citizens are supposed to have. Legitimacy refers to the rights that the state is supposed to have (as well as the persons filling certain roles within the institutional structure of the state). Citizens could have political obligations without the state being legitimate, since political obligations need not be imposed by rulers. For example, a theory of political obligation could claim that citizens have an obligation not to oppose states that do more good than harm, even if these states are not legitimate and did not impose any obligations on citizens. Another theory of political obligation could claim that citizens have an obligation to support certain state institutions when they freely accepted benefits from those institutions, even though this obligation was not imposed on them. So legitimacy and political obligations are separate matters (although of course related). John Horton indeed does not refer to his associative theory of political obligations in his recent article on state legitimacy. ^{Footnote 14}

Third, the puzzle of legitimacy is not dissolved in democracies. While it is true that in a democracy no person is born with a right to rule over others, and everyone can in principle be elected and thus acquire a right to rule over others (i.e., the liberty-right to co-enact laws and the power to thereby impose duties on citizens), this does not change the fact that some have the right to rule over others without this being legitimized by everyone’s consent. Voting does not help for two reasons. First, many people do not vote. Second, we have no vote on whether we want to grant the right to rule to some group of persons. ^{Footnote 15}

Fourth, rights are not scalar or gradual, and so legitimacy must be a binary notion. ^{Footnote 16} Either a state has the right to rule or not, either it is legitimate or not. But a state may come closer or less close to being legitimate, and in that sense legitimacy may be regarded as a “reachable threshold.” ^{Footnote 17} What brings a state closer to being legitimate will depend upon one’s theory of legitimacy.

II. Attitude-Sensitivity

Simmons’s conclusion that all states are illegitimate seems unacceptable to many political theorists, including realists like Horton. ^{Footnote 18} We can call it the Simmons-conclusion. We seem to need some theory of legitimacy that allows us to conclude that at least some states are legitimate. Of course other, non-Lockean strands within the liberal tradition have already tried to provide such a theory and, hence, have denied that actual consent is necessary for legitimacy. Yet at least the realist thinkers I mainly discuss here — John Horton and Bernard Williams — apparently want to retain something from the Lockean consent tradition. ^{Footnote 19} Horton wants to “restore the connection between political legitimacy and the beliefs and attitudes of those subject to it.” ^{Footnote 20} As literal consent cannot be had, the connection to the citizens’ attitudes must be something weaker than consent. So the idea is to avoid the Simmons-conclusion while meeting what I would like to call the desideratum of attitude-sensitivity. I think that attitude-sensitivity would indeed be an attractive feature for a realist theory of legitimacy, mainly because it would be surprising if a solution to the puzzle of legitimacy could ignore people’s actual attitudes toward being ruled. But different realist theorists may be attracted to an attitude-sensitive theory of legitimacy for different reasons, and I do not want to present any specific argument for an attitude-sensitive theory. I take it as given that realists want to avoid the Simmons-conclusion and meet the desideratum of attitude-sensitivity at the same time.

In Horton’s account, state legitimacy requires acceptance by a great part of the citizenry. The cultural-historical context determines what this means more specifically. ^{Footnote 21} Acceptance can be seen as a more realist substitute for consent, one that is still sensitive to the attitudes of real citizens. But, Horton says, acceptance does not “ground” or “justify” legitimacy:

What actually grounds legitimacy can vary:

Something other than acceptance grounds legitimacy, but it does so only because citizens accept the validity of that grounding. Hence citizens’ acceptance is necessary for the state’s legitimacy. Since there is often sufficient acceptance, this account allows the state to realistically acquire legitimacy. And because acceptance is a necessary condition for legitimacy, the Simmons-conclusion is avoided without violating the desideratum of attitude-sensitivity.

I should mention that Horton sometimes suggests that legitimacy does not require acceptance, but acceptability to most. More precisely, he invokes the criterion of acceptability not for the legitimacy of states, but for the legitimacy of modus vivendi arrangements. One may wonder how modus vivendi arrangements could have the right to rule and hence be “legitimate,” given that they are not the kind of thing that could have rights. But presumably, modus vivendi arrangements are here conceived as involving state institutions. (I will get to this point later.) Horton says that for “something to count as . . . ‘a modus vivendi,’ it has to possess some quality of legitimacy for those subject to it.” ^{Footnote 24} He sets constraints on both the content of a modus vivendi and its form. The content is “primarily the avoidance of serious civil disruption and the maintenance of a level of social order that is at least sufficient to enable the parties subject to it to live minimally worthwhile lives.” ^{Footnote 25} And its form must be “broadly consensual,” “broadly “acceptable,” or “agreeable” to those who are party to it. ^{Footnote 26} Yet when explaining these notions, he seems to assume acceptance again, not acceptability. He emphasizes that acceptance can be “to varying degrees reluctant, grudging and qualified,” ^{Footnote 27} and that acceptance renders the arrangements legitimate, even when it takes place against the background of power inequalities, ^{Footnote 28} at least as long as the acceptance “is not the product of clear, wilful, systematic and comprehensive deception by those with political power.” ^{Footnote 29} Thus, in the end it seems to be acceptance, not acceptability that is necessary for legitimacy, according to Horton. An advantage of the focus on acceptance is that it is more attitude-sensitive, compared to the acceptability-criterion. Something can be acceptable without tracking any of the actual thoughts and feelings of real people. To be sure, it should be somehow related to what people want and need, if it is to be acceptable, but it need not mirror their actual attitudes.

I now come to the other prominent realist account of legitimacy, that of Bernard Williams. Williams identifies the “first political question” as the “securing of order, protection, safety, trust, and the conditions of cooperation.” ^{Footnote 30} To solve the first political question is the primary task of the state. But not all states that solve the first political question are legitimate, according to Williams. States also have to provide an “acceptable” solution to the first political question. ^{Footnote 31} This is what his so-called “basic legitimation demand” requires. The basic legitimation demand asks for a justification of state power “to each subject.” ^{Footnote 32} When it is justifiable to each subject, then it is acceptable to each subject.

Acceptability is not the same as actual acceptance, as explained above. Something can be acceptable to some group of persons without any of them actually acknowledging that it is acceptable to them. That is why acceptability as a criterion of legitimacy does not fully meet the desideratum of attitude-sensitivity. Williams sometimes suggests that legitimacy requires actual acceptance, not acceptability. He advances a “critical theory principle” which says that “the acceptance of a justification does not count if the acceptance itself is produced by the coercive power which is supposedly being justified.” ^{Footnote 33} This principle is similar to Horton’s claim that acceptance is not to be the product of clear, willful, systematic, and comprehensive deception by those with political power. I here leave open what exactly is to count as coercion and deception, and how exactly we should understand that acceptance is the “product” of coercion and deception. ^{Footnote 34} Yet although the critical theory principle suggests that actual acceptance is necessary for legitimacy, Williams elsewhere clearly states that it is not acceptance, but acceptability: “[W]hen it is said that government must have ‘something to say’ to each person or group over whom it claims authority . . . it cannot be implied that this is something that this person or group will necessarily accept.” ^{Footnote 35} But because acceptance is clearly more attitude-sensitive than acceptability, realists should prefer Horton’s theory over Williams’s, if they stick to the desideratum of attitude-sensitivity.

III. Context-Sensitivity

There is a second desideratum that realist theories of legitimacy aim to meet. It is to be a realist theory as opposed to a “moralist” theory. One aspect of avoiding moralism is to avoid a theory of legitimacy that sets an ahistorical universal standard of legitimacy and thus neglects any relevance of historical and cultural context. Realists think that it would be odd to advance a theory of legitimacy that implies that almost all states before, say, the eighteenth century were illegitimate. One should not anachronistically judge Sparta as illegitimate just because Sparta was no modern liberal democracy, for example. Another realist desideratum, thus, is to develop a context-sensitive theory of legitimacy. I call it the desideratum of context-sensitivity.

Horton writes: “By contrast, what I am suggesting is that one of the things we should look [for] is an account that can genuinely accommodate the possibility that political legitimacy may be based on different criteria in different times and places.” ^{Footnote 36} And Horton’s theory of legitimacy quite clearly satisfies the desideratum of context-sensitivity, because at different times and places people will accept different grounding stories for the state’s legitimacy. Acceptance is required at all times and places, but the additional legitimation story can vary. So Spartans will regard their state as legitimate for different reasons than contemporary British citizens regard their state as legitimate, which is again different from why Tibetans regard the Dalai Lama as legitimate. ^{Footnote 37}

Some realists may argue that Horton’s theory does not fully meet the desideratum of context-sensitivity. Because acceptance is a necessary condition for legitimacy at all times and places, they may argue that the theory asks too much of states in different historical circumstances. The theory would imply that all states with institutionalized slavery were illegitimate, because they were (presumably) not accepted by the slaves. Horton offers two potential replies to this worry: either there is a threshold of sufficient support (only wide acceptance being necessary), or legitimacy is a relative matter and states can be legitimate relative to some, but at the same time not legitimate relative to others. ^{Footnote 38} I do not think the second answer is promising. If a state could be said to be legitimate relative to some, but illegitimate to others, it would remain an interesting open question whether it is legitimate overall, and we would need some criterion for that. So the first answer is the better answer: legitimacy merely requires wide acceptance, not acceptance by everyone subject to the state or regime. Hence if Babylonian slaves did not accept the legitimacy of the Babylonian empire, this does not necessarily undermine its legitimacy. Likewise, if a few British anarchists do not accept the legitimacy of the British state, this does not undermine the legitimacy of the British state. The critic may insist that we need a more precise criterion for how wide state acceptance must be. How many anarchists would undermine the legitimacy of the British state? I cannot elaborate an answer to this question here; I guess that some vagueness is unavoidable and unproblematic.

I now come to Williams’s account of legitimacy and the desideratum of context-sensitivity. To meet this desideratum is quite clearly one of the main motives behind his account of legitimacy. And the desideratum is met because the historical context determines how the basic legitimation demand can be met. A legitimation must “make sense” in the relevant context, as Williams says. ^{Footnote 39} For us, suggests Williams, only liberal states are acceptable: “Now and around here the [basic legitimation demand] together with the historical conditions permit only a liberal solution: other forms of answer are unacceptable.” ^{Footnote 40} But in other contexts, say in medieval Russia or in Sparta, other solutions to the first political question are acceptable.

As explained in Section II, Williams mostly treats acceptability, rather than actual acceptance, as necessary for legitimacy. When we treat acceptability instead of actual acceptance as necessary for legitimacy, then we get a theory of legitimacy that is highly permissive about legitimacy: it allows regimes that are in fact widely rejected to count as legitimate, as long as they can at least be said to be acceptable. Yet a realist critic may still not find Williams’s theory permissive enough. Take states that institutionalize slavery again, like Babylon. Surely such states were not only not accepted by the slaves, they were also not acceptable to the slaves. Again we seem to be forced to judge such states as illegitimate. Some realists find this judgment anachronistic and context-blind. Edward Hall thinks that it would be “fantastically unhistorical” to condemn all slave-holding societies as illegitimate. ^{Footnote 41} Matt Sleat thus suggests a modification to Williams’s theory that allows it to be more generous in granting legitimacy to states. The basic legitimation demand, according to Sleat, should not require justifiability to all subjects; rather, the constituency for the basic legitimation demand should be determined by the historical background. ^{Footnote 42} Thus, sometimes acceptability to all subjects is required for legitimacy, sometimes acceptability to all citizens (but not to slaves) is required, and maybe sometimes acceptability to the ruling class is sufficient. This would obviously make it easier for slave-holding states to be granted legitimacy. This is a point, though, where some will become skeptic about the desideratum of context-relativity. How far should we go down the road to pure relativism about legitimacy? Is it not cynical to say that Babylonian slavery was perfectly legitimate? I think that plausibly the desideratum of context-relativity has to find its limits at the desideratum of attitude-sensitivity. When a regime is not widely accepted, then it cannot be called legitimate, even if we talk about a regime in a historical context where wide acceptance did not count for much. Realists should thus resist Sleat’s modification of Williams’s theory of legitimacy and stick to Horton’s “wide acceptance” criterion, if they want to have both an attitude- and context-sensitive theory of legitimacy.

Some realists may argue that the desideratum of context-sensitivity should be met in a different way. They may point out that it is misguided even to apply the notion of legitimacy to societies from distant periods in which the issue of legitimacy was never raised. A good realist reply is that if the issue of legitimacy indeed was never raised, then this means that the state was widely accepted, that therefore the state should not count as illegitimate, and that there is no reason not to call it “legitimate.” If, on the other hand, the issue of legitimacy was raised, and many persons did not accept the state — like, for example, Babylonian slaves — then we should consider that state illegitimate. This is what a realist theory of legitimacy should say, if it is to be attitude-sensitive and context-sensitive at the same time.

IV. Realist Legitimacy and the Puzzle of Legitimacy

I see the attraction of theories of state legitimacy that avoid the Simmons-conclusion while meeting the desideratum of attitude-sensitivity and the desideratum of context-sensitivity. But the worry is that Horton’s and Williams’s theories do not solve the puzzle of legitimacy. Lockean consent theory solves the puzzle of legitimacy because it explains how unequal rights can come into being — how some persons can come to have the right to rule over others. Consent can generate new rights, and so it can generate the right to rule. How can consent do this? Alan Wertheimer distinguishes “subjective” and “performative” views of consent. ^{Footnote 43} According to the subjective view, consent is a mental state or a mental act. ^{Footnote 44} According to the performative view, consent is a certain kind of public act, observable and understandable by others. ^{Footnote 45} Wertheimer argues that consent can be morally transformative on the performative view only: only if understood as performative could it create new moral duties and rights. This is especially clear in the context of consent to sexual relations. Mere mental acts cannot be morally transformative; they cannot give some other person the liberty-right to have sexual intercourse with some other person. To be able to generate such a right, consent must be understandable as such; and this is only possible on the performative view. A rapist cannot try to argue that the victim said “yes” in her mind.

Acceptance may look similar to consent at first sight, as both reflect people’s attitudes. But it is now easy to see that acceptance is very different from consent. Acceptance is a mental state, not a performance or public act. And as such it cannot give rise to new duties or rights. But a theory of legitimacy must explain the rise of new rights, since the state’s legitimacy consists in a bundle of rights that the governed lack, while originally all persons have equal rights. Thus, acceptance cannot explain the state’s legitimacy because it cannot solve the puzzle of legitimacy. Note that acceptability is even worse on that count. It is not even a candidate for something similar to giving consent. It is neither something performed publicly nor a mental act or mental state, and thus it cannot be a proper substitute for consent. There is no hope that it could solve the puzzle of legitimacy and explain the rise of new and unequal rights. I conclude that realist theories of legitimacy cannot solve the puzzle of legitimacy. This is worrisome, of course, because a theory of legitimacy should solve the puzzle of legitimacy. I now discuss three objections to my argument that realist accounts of legitimacy cannot solve the puzzle of legitimacy. I will discuss arguments against acknowledging the puzzle of legitimacy as a puzzle in the next section.

One objection is that of course acceptance is sometimes morally transformative: we call such acceptance “tacit consent.” I agree that one can give tacit consent, but I deny that it is the mental state of acceptance that is morally transformative in tacit consent. Let me explain. The difference between explicit consent and tacit consent is not that one utters words (such as, “Yes, I agree!”) in the former type of consent. One can give explicit consent by nodding one’s head. The difference between explicit and tacit consent is that in the former type of consent one performs an action that has no other purpose than to give consent. ^{Footnote 46} In tacit consent, one gives consent without performing an action whose sole purpose is to give consent. Both explicit and tacit consent are consent and as such are morally transformative. But tacit consent is trickier, because it seems both unclear under what conditions tacit consent is given and what exactly one consents to when tacit consent is given. Conventions are crucial here. For example, one gives tacit consent to pay the bill, once one takes a seat and orders a meal in a restaurant. One need not explicitly say that one will later pay the bill. But what makes this an instance of tacit consent is not the mental state or mental act of accepting that one will pay the bill. It is the public act of taking a seat and ordering a meal. It is a probably universal convention to regard this as consent to pay the bill. Could one build the case for state legitimacy on tacit consent instead acceptance? No, because quite clearly there are no conventions that allow us to count a person’s nonemigration — or anything else — as tacit consent to the state’s authority (as David Hume famously pointed out against John Locke’s attempt to ground political authority on tacit consent). ^{Footnote 47}

A second objection says that I mischaracterized acceptance. ^{Footnote 48} In fact acceptance is not only a mental state or mental act, but also behaviorally manifest. One can observe that others accept the state, for example, in how they talk about the state, or in how they behave toward state officials. I do not deny that. But I deny that it helps to show that acceptance can be morally transformative. Either behaviorally manifest acceptance amounts to tacit consent, in which case we are back at the first objection, or behaviorally manifest acceptance does not amount to tacit consent, in which case we still lack an explanation of how this acceptance could give rise to new rights for some persons.

A third objection to my argument is that Horton’s and Williams’s theories should not be seen as a substitute for Lockean consent, but as substitutes for hypothetical consent. The normative point of hypothetical consent is very different from that of actual consent. The point of hypothetical consent is not to track the will of real persons, but to track the reasons that real persons have to accept some arrangement (even though they may in fact dissent). ^{Footnote 49} It is important to see that “[a]ppeals to hypothetical choice, acceptability, or reasonable nonrejectability have a very different moral basis and force than do appeals to actual choice.” ^{Footnote 50} Hypothetical consent signifies that everyone has sufficient reason to accept something, even though some may in fact dissent. So maybe Horton’s and Williams’s theories are better understood as realist counterparts for hypothetical consent theories. They focus on people’s reasons, not their will.

In reply, one should first point out that an attitude-sensitive theory of legitimacy cannot plausibly be seen as a substitute for hypothetical consent. Acceptance and actual consent reflect the actual attitudes of persons, not the reasons they have. Acceptability and hypothetical consent reflect the reasons persons have, not their actual attitudes. Hence if we stick to the desideratum of attitude-sensitivity, we should take Horton’s criterion of acceptance, not acceptability, as necessary for legitimacy, and so we cannot claim to have a realist substitute for hypothetical consent. Second, if we were to drop or weaken the desideratum of attitude-sensitivity and use Williams’s criterion of acceptability as necessary for legitimacy, then we may claim to have a realist substitute for hypothetical consent. But the problem, then, is to make a hypothetical consent theory of legitimacy work. ^{Footnote 51} Hypothetical consent to the state’s right to rule indicates that everyone has sufficient reason to accept the state’s right to rule. Yet why should the bare fact that people have sufficient reason to accept the state’s right to rule make it the case that the state actually has the right to rule? The fact that it would be good if some person had moral powers over others cannot plausibly make it the case that she has moral powers over others. ^{Footnote 52} Hypothetical consent seems to be unable to do what actual, real consent can do: transfer rights. Of course, hypothetical consent theories would merit a fuller discussion than I can deliver here. But the prospects are not very promising.

V. Is it a Puzzle?

I argued that a realist theory of legitimacy is attractive insofar as it is attitude-sensitive and context-sensitive. Yet acceptance cannot work as a more realistic substitute for (explicit or tacit) consent, because it is a mere mental act or mental state and as such is not morally transformative. For that reason, a realist theory of legitimacy cannot explain how the state could acquire the right to rule. It cannot solve the puzzle of legitimacy. That is why it is ultimately unconvincing. The most obvious reply to my argument is that realists like Horton and Williams do not even want to solve the puzzle of legitimacy. They do not see it as a puzzle. To show why one should not see it as a puzzle, four points can be made.

First, Williams may contend that it is a puzzle only from a moralist point of view. Williams takes “mainstream” political philosophers as “moralists” because they treat political philosophy as applied ethics, advancing an “ethics first view.” ^{Footnote 53} Realists, in contrast, see politics as a distinct realm of normative inquiry. They want to think about politics “from within.” At the same time, they do not want to give up all normative reasoning. Williams contends that the basic legitimation demand may well be a moral principle, but that it is not a moral principle that is “prior to politics.” Instead, Williams says, it “is a claim that is inherent in there being such a thing as politics.” ^{Footnote 54} “Politics,” properly understood, is not based on brute force: “The situation of one lot of people terrorizing another lot of people is not per se a political situation: it is, rather, the situation which the existence of the political is in the first place supposed to alleviate (replace).” ^{Footnote 55} Williams wants to treat “might does not imply right” as an axiom, maybe even as analytically true about politics. ^{Footnote 56} Thus, the basic legitimation demand is built on the very notion of politics. As such, it is normative without relying on moral principles outside the political realm.

I cannot fully engage the debate about moralism and the rejection of ethics-first views here. But I would like to at least briefly express my skepticism about the possibility of “nonmoralism.” ^{Footnote 57} Williams’s trick seems to be to build moral principles into the very definition of politics. There are two worries about that. First, we usually do not talk about politics like that. Is Kim Jong-un really not engaged in politics? Williams seems to think so. But I do not think most people do. It is ironic that realists, usually prepared to theorize the dirty business of politics, come up with a moralized definition of what politics essentially is. ^{Footnote 58} I find it more natural to accept that deception, coercion, usurpation, and so on are political phenomena. Second, once we acknowledge that making normative claims about politics means making moral claims, the distinction between moral claims “from within” and “without” politics seems to lack a point. We should simply use sound and relevant moral ideas in our thinking about politics, and this is something that both Williams and so-called moralists try to do. They just disagree about which moral claims are sound and relevant for the moral evaluation of politics. For that reason, a realist cannot reject the puzzle of legitimacy with a generic verdict against “moralism.” A realist must point out why the specific moral claims involved in the puzzle of legitimacy are unsound or irrelevant for political thinking. This brings me to the second point.

Second, realists can argue that the specific moral claims involved in the puzzle of legitimacy are unsound or irrelevant for political thinking. The moral claim involved in the puzzle of legitimacy is that all persons originally have equal rights. None originally have the right to rule over others. The best argument for resisting that moral claim is to point to the “naturalness” of living in states. Horton writes:

Instead, Horton thinks, we should just take the existence of states — or “political communities” more broadly — as “natural.”

If we start with seeing persons as naturally embedded in political communities, and if political communities always contain relations of authority and subordination, then there is not much sense in ascribing equal rights to all persons or in seeing persons as equal. That some persons have the right to rule over others is a basic feature of politics, which itself is a “natural” phenomenon. Accordingly, there is no need for a morally transformative act that would explain how the right to rule could come into existence. There is no puzzle of legitimacy.

In reply, one should concede that political communities are natural in Horton’s sense. But one can insist that the puzzle of legitimacy nonetheless remains a puzzle. The claim that persons originally have equal moral rights is not shown to be unsound or irrelevant to politics by pointing out that we naturally live in political communities and hence in relations of authority and subordination. We might naturally live in illegitimate political communities, where rulers rule without having the right to rule. Indeed the claim that persons are equals and have equal original rights is widely accepted (without awareness of how this makes state legitimacy a puzzle, of course). And it is an extremely plausible moral claim. One can surely ignore the puzzle of legitimacy or find it uninteresting — but this does not make it go away.

Third, one could argue that the puzzle of legitimacy is no puzzle because real people do not see it as a puzzle. The argument very much depends on what we see as the proper task of political philosophy. Horton sees the purpose of political philosophy in “understanding” or “interpreting” politics. ^{Footnote 61} He writes:

Since ordinary people do not worry about the puzzle of legitimacy, solving it is no job for a political philosophy that explains and interprets politics. To the contrary, since most British citizens accept the legitimacy of Britain, for example, this is a phenomenon political philosophy should try to explain, understand, and interpret.

There are two available replies to this. First, one can try to defend a different methodological outlook. One need not dispute that the tasks Horton identifies are important and worth engaging, of course, but one can nonetheless point out that political philosophy should also be able take a radically critical stance toward political realities, including the very existence of the state, and even in light of wide acceptance of the state. Yet this is not the place to elaborate methodological claims, and so I have to leave it at that. Second, one can argue that the puzzle of legitimacy is, after all, not so foreign to Horton’s own methodological approach in political theory. The puzzle comes up when we take seriously that citizens usually regard their states as legitimate, while on the other hand, we also take seriously that they believe that persons are equals (or have equal rights). Most people are not aware that these two things are in tension. Thus, political theorists should try to make sense of this, and explain how we could have both: equality (or equal rights) and legitimate states. The puzzle of legitimacy is not a puzzle that comes up against the background of controversial abstract and complicated moral claims that only philosophers understand.

Fourth, a realist could argue that I miss the very point of legitimacy and thus wrongly read a puzzle into state legitimacy. I ignore what is sometimes called the “circumstances of politics,” or “circumstances of legitimacy.” Basically, these circumstances are the need for stability and order in the face of deep disagreement including disagreement on justice. ^{Footnote 63} When the circumstances of politics (or legitimacy) apply, then we need political procedures that are available under such circumstances, and this is the context where the notion of legitimacy comes in. It denotes the procedures that are appropriate under the circumstances of politics (or legitimacy), procedures that are acceptable to all despite deep disagreement on substantial issues. For that reason, we should use a standard of legitimacy that is less demanding and less controversial than the standard of justice, not a standard that is impossible to attain, like the standard of universal consent. ^{Footnote 64}

In reply, I agree that legitimacy is distinct from justice. I also agree that we need certain procedures — or modus vivendi arrangements more generally — that allow us to live in peace under circumstances of disagreement on justice. But nothing in our need to have modus vivendi arrangements helps to solve the puzzle of legitimacy. For that reason, one should not call all (acceptable) modus vivendi arrangements “legitimate,” just because we need them — at least if one associates legitimacy with the right to rule. ^{Footnote 65} Moreover, strictly speaking modus vivendi arrangements are not the kind of thing that could be legitimate, because they are not the kind of thing that could have rights (including the right to rule). While it is true that modus vivendi arrangements can involve state institutions, they need not involve state institutions – they can take many different forms. Moral norms or conventions can be working modus vivendi arrangements. And when modus vivendi arrangements do involve state institutions, then we should call these state institutions “legitimate” or “illegitimate,” not the modus vivendi arrangements that involve them. State institutions should be called legitimate when they have the right to rule. To explain how state institutions could have the right to rule requires solving the puzzle of legitimacy. This is what a theory of legitimacy should aim to do.

VI. Conclusion

I can agree with Horton and Williams that whether a state is widely accepted (or acceptable) is highly important for its moral evaluation. And there is nothing wrong, in principle, with calling widely accepted (or acceptable) states “legitimate.” (One should not fight over words.) But if a theory of legitimacy is to solve the puzzle of legitimacy — and I argued that it should — then the task is to explain how some persons can acquire the right to rule over other persons. Horton’s and Williams’s realist theories do not help with this task.