On the Value of Political Legitimacy

Article

On the value of politicallegitimacy

Mathew CoakleyNew York University, USA

AbstractTheories of political legitimacy normally stipulate certain conditions of legitimacy: thefeatures a state must possess in order to be legitimate. Yet there is obviously a secondquestion as to the value of legitimacy: the normative features a state has by virtue of itbeing legitimate (such as it being owed obedience, having a right to use coercion, orenjoying a general justification in the use of force). I argue that it is difficult to demon-strate that affording these to legitimate states is morally desirable, and that obviousalternative conceptions of the value of legitimacy (notably epistemic and instrumental)are not without problems of their own. The intuitive triviality of establishing the value ofnormative legitimacy may mask a serious problem.

Keywordslegitimacy, obedience, political authority, democratic legitimacy

On the value of political legitimacy

Theories of political legitimacy normally focus upon and defend certain conditions oflegitimacy, that is, they specify the features a state or polity must possess for it to be con-sidered legitimate, and try to show how and when these may be met.1 Yet there is obvi-ously a second question of the value of legitimacy, that is, of the normative features alegitimate state has by virtue of it being legitimate (such as it being owed obedience, hav-ing a right to use coercion, or enjoying a general justification in the use of force). I amgoing to discuss in this article a surprising, but troubling problem concerning the latterquestion, namely, that legitimacy, if it has any such value, seems only to change how weshould act or judge overall in those situations in which, prima facie, we should not wantit to. As such, it seems to have apparent moral disvalue.

Corresponding author:Mathew Coakley, New York University, Department of Politics, 19 W4th St, New York, NY 10012, USAEmail: [email protected]

Politics, Philosophy & Economics10(4) 345–369

ª The Author(s) 2010Reprints and permissions:

sagepub.co.uk/journalsPermissions.navDOI: 10.1177/1470594X10387272

ppe.sagepub.com

ioanna kazakou

ioanna kazakou

Now, as a conclusion this seems absurd. Even some anarchists could reject it.2 But, asI hope to show, it is much harder to avoid than it might initially appear. The basic prob-lem arises from what we might call the ‘counterfactual import’ of a duty or justification.That is, we can ask: Which are the situations in which a duty or justification alters overallhow we should act or judge? These are the situations in which it ‘makes a difference’.Legitimacy-based duties or justifications appear only to have a counterfactual importin situations in which they justify, or require, the infliction of a harm which cannot bejustified by reference to anyone affected.

For example, an obligation to obey the law of a legitimate state is redundant when wealready have sufficient reason to obey (such as ‘do not murder’), makes no differencewhen disobedience is owed no matter what state passes such a law (‘be complicit in tor-ture’), and thus only seems to matter for laws that should, on their merits, because of theharm they promote, be disobeyed, but should nonetheless be obeyed if commanded by alegitimate state. As such, an obligation to obey the law of a legitimate state appears unde-sirable, not because it conflicts with moral autonomy, natural rights, or has not been con-sented to, but because its only counterfactual effect is to require us to obey laws that, ontheir merits, should be disobeyed. If we have (somehow) consented to such an obligation,then we may have made a morally bad choice in so doing. Similarly, the counterfactualeffect of a state enjoying a general justification in the use of coercion because it is legit-imate appears to be to permit, excuse, or justify acts that are, based on their nature,morally undesirable. These are deeply counterintuitive implications, and in the first partI try to set them out as cleanly as possible.

In the second part, I survey seven seemingly promising responses: (1) that a legitimatestate’s right to rule is necessary and not merely sufficient for justified coercion; (2) that alegitimate state’s right to rule is sufficient, but not necessary, for justified coercion; (3)that a legitimate state enjoys the right to enforce pre-existing rights; (4) that legitimacy iscrucial in those situations in which there is reasonable disagreement; (5) that obligationsbased on legitimacy only apply to acts that, on their merits, are morally optional; (6) thatgovernment is legitimate if it is a reliable epistemic authority; and (7) that the concept oflegitimacy may promote institutional improvement, such as in spreading democraticreform. Despite their apparent appeal, I will try to show why all of these, at a minimum,face serious difficulties.

The mere existence of the problem is not necessarily dispiriting; there are so manyimpressive competing theories of why and when political institutions are legitimate thata compelling account of the value of legitimacy that solves the counterfactual problemwould presumably help us to choose between them. I thus try in what follows to showmerely how overcoming the problem is at least not an obviously easy task. Perhaps thereis some way of establishing the value of legitimacy in something like the form it has typi-cally been taken to exist. The argument of this article is simply that the obvious ways ofdoing so either fail or introduce new problems: there really is a difficulty here.

The potential value of legitimacy

What is the value of political legitimacy? That is, what normatively desirable propertiesdoes a legitimate state have by virtue of it being legitimate? Simmons characterizes a

346 Politics, Philosophy & Economics 10(4)

standard moral conception of a state’s legitimacy (and one underlying the Lockeanaccount) as ‘the complex moral right it possesses to be the exclusive imposer of bindingduties on its subjects, to have its subjects comply with these duties, and to use coercion toenforce these duties’ (Simmons, 1999: 746). Estlund states that

By authority I will mean the power of one agent (emphasizing especially the state) to

morally require or forbid actions by others through commands. (To forbid x is to require

not-x, and so I will usually simply speak of the moral power to require). By legitimacy I

will mean the moral permissibility of the state’s issuing and enforcing its commands owing

to the process by which they were produced. (Estlund, 2008: 2)

Despite differences in usage, there are two basic ideas here, and we might accept one, theother, or both: that legitimate government has a right or general justification in the use orthreat of force, or that one has a duty to obey the commands (commonly assumed to bethe laws) of a government that is the legitimate authority or where the commands stemfrom a legitimate procedure. I am going to refer to legitimacy as having normative valueif any one of these normative concepts (a duty to obey, a general justification of coercion,or a right to employ it) can be shown to be desirable when held by a government if legit-imate. In doing so, I mean initially to exclude purely instrumentalist accounts of legiti-macy whereby laws and institutions are legitimate when, and to the extent that, theybring about desirable outcomes, but it is the desirability of the outcomes, not how orby whom they were produced, that lends them legitimacy. This, I think, is the best wayto understand the sense in which Hobbes (1994) views order-securing government aslegitimate: it is that a government secures order, not who rules or how laws are arrivedat, that ultimately confers legitimacy.3 It is also true of those, such as Van Parijs (1996),who hold a procedure or institution to be justifiable to the extent it secures justice in itsoutcomes, not in the means by which it produces them. Legitimacy can also be used toapply to certain actions irrespective of what type of institution undertakes them or pro-cedure authorizes them. Thus Nozick (1974) offers, ultimately, a set of actions that nostate, no matter how instituted, could legitimately do (the uncompensated violation ofone’s rights).4 It is also possible to apply a Rawlsian concept of legitimacy not to statecoercion in general or to those who undertake coercion, but, rather, to specific acts ofcoercion.5 This article focuses on normative non-instrumental legitimacy as applied topolitical institutions, that is, on theories that hold one has a duty to obey, or that thereis a general justification of force or coercion or right thereof, by virtue of a government’sor a procedure’s legitimacy.6 If so, what is the problem?

An initial outline of the counterfactual problem:force, coercion, and duties

If the use of force or coercion is partly or wholly justified because it is done by alegitimate state, then this changes the overall assessment of such acts only in those situa-tions in which the use of force or coercion would be unjustified on its own terms. Forexample, consider perhaps the most common and prominent use of the state’s coercive

Coakley 347

apparatus: to punish, or threaten to punish, some particular individual. The following arethree possibilities if one holds a desert theory of punishment.

1. The individual deserves the punishment (or to be coerced) and any state would bejustified in punishing him or her.

2. The individual does not deserve the punishment (or to be coerced) and no statewould be justified in punishing him or her.

3. The individual does not deserve the punishment (or to be coerced), but a legitimatestate is still justified in punishing him or her.

The following are the same three possibilities if, instead, one holds some sort ofconsequence-based theory of punishment.

1. Punishing (or coercing) the individual produces relevantly good consequences, andany state would be justified in punishing him or her.7

2. Punishing (or coercing) the individual produces relevantly bad consequences, andno state would be justified in punishing him or her.

3. Punishing (or coercing) the individual produces relevantly bad consequences, but alegitimate state is still justified in punishing him or her.

One possible intuitive example of item 1 in both formulations would be someone whocommitted murder or rape – even an illegitimate state would seem to many to be justifiedin using its organized force to lock up or to try to deter coercively such an individual. Anintuitive example of item 2 would be someone who was criminalized for being Jewish –no state would be justified in punishing him or her simply for being Jewish. Thus, a the-ory of legitimacy only matters for punishment or coercion if, and to the extent that, itentails that there are many and important cases in which individuals do not deserve tobe punished or doing so brings about relevantly bad consequences, but a legitimate stateis still justified in punishing them. The puzzle is why we should want this, namely, thatcategory 3 both has cases that fall within it and that these are significant.

Here is the same type of worry for a duty to obey a state or polity by virtue of its legiti-macy. Consider a particular law that is interpreted as requiring us to ‘do P’. A crude divi-sion of cases might be as follows.

1. Doing P is required by reference to those affected by my doing P, and thus I shoulddo P whether commanded by a legitimate or illegitimate state.

2. Not doing P is required by reference to those affected by my doing P, and thus Ishould not do P whether commanded by a legitimate or illegitimate state.

3. Not doing P is required by reference to those affected, and I should thus not do P,unless commanded by a legitimate state.8

An intuitive example of item 1 for many might be refraining from murder: whether itis illegal under a legitimate or illegitimate state, one should not murder, this being jus-tified by reference to the person affected, the would-be victim. An example of item 2might be torture: even if a legitimate or illegitimate state passed a law requiring citizens


to capture and torture Muslims if the opportunity arises, this is not what citizens shoulddo. Thus, the puzzle is why we should want a duty whose sole counterfactual import (thedifference it makes) is that there are many cases that fall into category 3. Why would wewant there to be acts of obedience that, considered on their merits, one ought not to do,but because they are commanded by a legitimate state one should do them?

This concern is distinct from two other general skeptical worries. The first argues thata duty to obey conflicts with certain individual rights or moral powers, thus Wolff(1970), for example, claims that obeying an authority because it is an authority is incom-patible with moral autonomy. The second casts doubt on either the cogency or applic-ability of particular arguments in favor of legitimating a political structure. Thus, onecan argue against voluntarism that, even if consent to a legitimate polity could createobligations to obey the law, no one (or almost no one) has so consented, and that tacit,coerced, or hypothetical variants do not bind.9 The problems raised above are distinct inthat they do not require a strong commitment to a particular notion of moral autonomyand hold across legitimacy-grounded theories independent of their specific argumentsfor why we are obligated or why force is justified.

One thing necessary for the above problems to arise, however, is for it to be possibleto assess the normative desirability of obedience to various laws, or of the committing ofparticular acts of coercion, independently of by whom they are undertaken or how theyare passed. This ‘independence assumption’ assumes that the content of the law or thecontent of an act of coercion can be normatively evaluated ‘on its merits’. Here are someexamples of things a state might do: coerce murderers, outlaw consensual adult homo-sexual relationships, enforce a speed limit, racially segregate schools, provide tax-funded healthcare for the poor, ban the hajib, protect property rights, require restaurantworkers to wash their hands after using the bathroom, and outlaw torture. Throughoutthis article I will assume that (in some cases and in principle) ‘based on their merits’we can positively or negatively evaluate normatively the desirability of a state undertak-ing these or, when relevant, of people obeying the implied commands underpinningthem, where this reflects considerations related to their content – what racial segregationinvolves, the benefits of a speed limit, the health effects of washing one’s hands, and soon. This does not entail that the overall-things-considered evaluation of such acts will notinclude consideration of who undertakes them or how or by whom the decision to do so ismade – the desirability of this is after all what is initially at stake. I merely assume thathowever we wish lexically to describe them (as just, as desirable, as a (pro tanto) duty, asjustified) they can be partly normatively evaluated based on their merits. The counter-factual problem is about how, if so, we should relate their overall normative evaluationto this merit-based evaluation if we include the evaluation of by whom or by what pro-cedure they are authorized, produced, or undertaken.

Illustrating the problem with democratic legitimacy

The counterfactual problem relies on comparing obedience or coercion in a legitimatestate with such actions in an illegitimate one. But one notable danger of any such com-parison is of intuitive normative overdetermination – in other words, that when compar-ing these two state forms our background awareness of the range of ways a ‘typical’

Coakley 349

illegitimate state may be bad may hinder us in cleanly assessing the role legitimacyshould and should not play. Dictatorships, for instance, may be inescapably dominatedin the minds of many with the expectation of egregiously bad laws and bad actions andtherefore the specific impact of their illegitimacy is hard to discern. Thus we requireexamples of states that, under particular prominent legitimacy theories, would be illegi-timate, but which might in expectation produce many or at least some laws similar to alegitimate state. For the sake of argument, I am therefore going to assume simply thatlegitimate government is somehow democratic, and by this refer either to those theoriesthat emphasize democracy’s fair or equal procedural features or, alternatively, to thosethat emphasize the role of consent via voting.10 What might we use to illustrate an ille-gitimate state? For such proceduralists (those who focus upon the equal concern or fairtreatment a legitimate democratic state would afford all citizens) here would be an exam-ple of an illegitimate state:

Fridocracy: only citizens born on Fridays are allowed to vote and the political authority is

otherwise constituted as a regular democracy.11

Alternatively, for a democratic consent theorist, it might still be the case that in a‘Fridocracy’ people born on a Friday have political obligations, as they are electorallyenfranchised.12 Thus, for these theories, let us use for our illustrative illegitimate statea ‘Lottocracy’ (which might, notably for a proceduralist, be legitimate):

Lottocracy: the upper and lower houses of the national legislature, and all state or regional

assemblies, are chosen by a lottery of all adult citizens. Executive offices are chosen by

these houses.

To return to the problem, the following is the key question about coercion: Whenwould the general justification of force or coercion enjoyed by a legitimate state makea difference to our overall assessment of an act of force or coercion such that a Fridoc-racy or Lottocracy would not be justified in doing so (because it was undeserved, causedunwarranted harm, and so on), but it would be normatively acceptable for a democracy todo so? For if there are no such cases, then the general justification of coercion enjoyed bylegitimate states must have no or negligible normative weight: it is always overridden byconsidering what is done. Thus, there must be some cases, considered on their merits, inwhich most governments would have been unjustified in so acting, but the general jus-tification of the use of force or coercion enjoyed by a legitimate democratic state made(for them) the causing of such harm acceptable. Why, though, would we want it to bemorally acceptable for democracies to harm citizens when we freely admit it is unaccep-table for governments in general, such as Fridocracies, to do so?

Alternatively, with regard to duties to obey legitimate states, consider someone decid-ing whether to provide homegrown marijuana for her grandmother who has arthritis. Forempirical simplicity, let us assume that she is justified in believing that it will ease hergrandmother’s pain. Doing so is illegal. She considers what is best for all those affected(including potentially her risk of punishment), and on these grounds concludes that sheshould grow and provide the drug. She now factors in, however, that the law was passed


by a legitimate democratic government. Should she change her assessment of what todo? If a theory of legitimacy that grounds an obligation to obey the law hascounterfactual import, then the answer must sometimes be ‘Yes’ (her grandmother’s painor the danger of punishment can be altered if the specific empirics are not quite right).These, after all, are exactly the cases in which a duty to obey is not redundant: we do nothave sufficient independent reason to obey (such as when we do not murder) nor is obey-ing heinous and thus we should not do so even with such a duty to obey (such as whencommanded to torture). A duty to obey matters counterfactually if, and only if, consid-eration of the interests of those affected means one should disobey, but the obligation toobey outweighs this. Yet such a duty cannot be founded on the interests of those affected,because these do not vary depending on the providence of the law and are already part ofour deliberation. The consequence is that, by living in a democracy, she is morallyrequired to produce a harm (to her grandmother) that she would not be required to pro-duce living in an illegitimate state, such as perhaps a Fridocracy or Lottocracy. Theredoes seem something morally disquieting here.

All of the above problems used the concept of either an illegitimate or a legitimatestate. However, legitimacy may be held to be often a matter of degree: states can be moreor less legitimate. This does not remove any of the concerns though, indeed it helps illus-trate the difficulties with the view that force is justified, or obedience owed, due to factsabout the type of state that undertakes such coercion or authors such laws, rather thanabout the content of the coercive act or law. If the coercion of a state becomes more jus-tified as a state becomes more legitimate, then this will make a difference in those casesin which such coercion is not justified anyhow and not universally unjustifiable. Thus, asa state becomes more legitimate, we would expect there to be more potential cases inwhich it could justifiably coerce those who did not deserve to be coerced or in which theresults of so doing were relevantly bad. As a state becomes more legitimate, we risk anincrease in something seemingly normatively undesirable: the justified infliction of pun-ishment or the threat thereof on those who do not deserve it or the bringing about of badconsequences. So too with duties. If, for example, the strength of the duty one has to obeythe law depends on the degree of political legitimacy, then as a state becomes more legit-imate there will be more and more cases in which one should obey the law when disobey-ing would be justified solely considering all those affected. As the state becomes morelegitimate, one should expect to be morally required to bring about more harm. Thisseems perverse.

The overall structural worry is as follows. Consider coercive, or use-of-force backed,action X that is undertaken by a state: it may be imprisonment, the threat of going to jail,a range of other sanctions, or whatever. The overall challenge for a theory of legitimacyis to show why a legitimate state being normatively justified in doing X is desirable if itis undesirable that states in general should do X. In other words, why is it good for ademocracy to be able justifiably to do X when it is bad for all other states (such as aFridocracy or Lottocracy) to do X because of the type of action X is (doing X is in therelevant sense bad for those affected)? Similarly, a theory of legitimacy-grounded dutiesneeds to establish why it is desirable for moral agents to obey a law to do Y if doing Y issomething that moral agents in general should not do and would not do in an illegitimatestate because, on its merits, doing Y is morally bad. Which laws should we (morally)

Coakley 351

disobey in a Fridocracy or Lottocracy, but obey in a democracy? Which bad, unjust, orunwise laws does an obligation to obey a legitimate state render deserving of obedience?For we already normally have reason to obey those that are good, just, or wise, and thisholds in any state. If there are such cases, if legitimacy makes a difference to an agent’sduties or to the normative evaluation of a state’s force or coercion, then we will havesucceeded in demonstrating the value of legitimacy, but also and only, it seems, of jus-tifying the infliction of some otherwise unjustified bad. This, at least, is the problem.

Possible responses

Are there ways to address or circumvent the problem? Here I consider seven that initiallyseem promising: (1) that a legitimate state’s right to rule is necessary and not merely suf-ficient for justified coercion; (2) that a legitimate state’s right to rule is sufficient, but notnecessary, for justified coercion; (3) that a legitimate state enjoys the right to enforcepre-existing rights; (4) that legitimacy is crucial in those situations in which there is rea-sonable disagreement; (5) that obligations based on legitimacy only apply to acts that, ontheir merits, are morally optional; (6) that government is legitimate if it is a reliable epis-temic authority; and (7) that the concept of legitimacy may promote institutionalimprovement, such as in spreading democratic reform.

(1) A right to rule is a necessary requirement for justified coercion

In setting out the problem and dividing coercive acts into the three cases I included situa-tions in which an illegitimate state could justifiably coerce, and this helped produce thecounterfactual problem as the only difference legitimacy made was to the third category(where such coercion is undeserved or undesirable on its merits, but a legitimate statecould still justifiably undertake it). But what if the right to rule enjoyed uniquely by thelegitimate state is necessary not merely sufficient for justified coercion? The value oflegitimacy, under this counterargument, is that it provides a government with therequired permission to undertake a range of coercive acts, and this, overall, is potentiallygoing to be very desirable indeed (in overcoming collective action problems and so on).

The central difficulty with this account of the value of legitimacy, however, is that ithas a double impact: it provides a legitimate state with the right to coerce, but it denies itto an illegitimate state. While we would not want an illegitimate state to have a generalright to coerce, if legitimacy’s impact is to deny it and its agents the potential ever tocoerce justifiably, then a range of troubling problems are going to accrue immediately.

More concretely, the claim that a right to rule is both only enjoyed by legitimate statesand is necessary to coerce justifiably is going to have the consequence that anyone in anillegitimate state with influence over state coercion will be morally required not tocoerce. Police chiefs should not enforce the law; legislators should always vote againstan action backed by force; prison wardens should leave doors unlocked; judges andjuries should not convict; tax officials should never threaten those who do not pay, andso on. It may also be that one should not work in the public sector as the revenues to payone’s wages will likely be collected with state coercion, thus teachers, doctors, andnurses should maybe resign. This list of activities to avoid doing is not meant to be


facetious, but rather to highlight that while there are often few individuals with directinfluence over whether an illegitimate state becomes legitimate, there are always goingto be a great many involved in its coercion, and in many of these cases, if the lack of statelegitimacy entails that all coercion is normatively unjustified, then moral agents shouldas a result be complicit in causing a great amount of harm: not coercing would-bemurderers and rapists will have predictable, and horrible, consequences that cannot bejustified by reference to anyone’s interests, since it is not the acts of coercion themselves,but facts about their origin that make them morally unacceptable for this concept oflegitimacy. If only a legitimate state can ever justifiably coerce, then those in the coer-cive apparatus of illegitimate states (were they to take this fact seriously) would poten-tially move their citizenry much closer to a Hobbesian state of nature, with all thesuffering that may entail. Of course, if they could make the state legitimate that wouldbe best, but for jurors, judges, police officers, and a great many more the option ofacquitting or not arresting murderers is the more realistic way to avoid violating the mur-derers’ right to not be coerced by an illegitimate state. I doubt few, on reflection, wouldwant to maintain this stance, but if they did, the inhabitants of such a state would seemsignificantly worse off as a result, and legitimacy would, thus, have apparent normativedisvalue (if the concept was normatively meaningless and illegitimate states could jus-tifiably coerce potential murderers, a great many people would plausibly be better off).The view that an illegitimate state can never be justified in using any coercion (as it hasno right to do so) and thus should cease doing so, seems to have deeply unattractiveimplications.

(2) Holding that a right to use coercion is sufficient, but not necessaryto coerce justifiably

If a general right to use coercion is necessary to coerce justifiably and an illegitimatestate lacks such a right, then it should not coerce would-be murderers, among many oth-ers, and deeply unattractive consequences will ensue. But what if such a general rightwas sufficient to coerce justifiably (within certain bounds), but not strictly necessary?While only a legitimate state has a general right to use force or coercion, an illegitimatestate ought sometimes to do so (such as when not doing so causes great harm, such as notcoercing would-be murderers). This overcomes the problems that accrued to counter-response (1). Yet it does so at the cost of reintroducing the original worry.

If an illegitimate state can be justified in using coercion when it is deserved or pro-duces relevantly good consequences, then the only counterfactual work a general rightto coerce does is to give a legitimate state a right to coerce those who do not deserveit or when the consequences of so doing are relevantly bad. Such a state already has aright to coerce those who do deserve it because such coercion is justified by referenceto the victims, potential victims, or agent’s behavior. Thus again we encounter the prob-lem that the justification of force and coercion, whether generally or via the notion of aright to use it, provides a legitimate state with a capacity to be sometimes justified whenacting wrongly, yet is superfluous in other contexts. A democracy, Fridocracy, andLottocracy can be justified in punishing those who deserve it, or when such punishmentproduces relevantly good consequences, but a democracy (by virtue of its legitimacy)

Coakley 353

enjoys the additional capacity to be sometimes normatively justified in punishing thosewho do not deserve it or when the consequences of so doing are relevantly bad. If anystate has a right to act justly, then all legitimacy does is give the legitimate state a rightto act unjustly. The citizens of such a state thus appear to be potentially made worse offby the fact of its legitimacy.

One important potential ‘disanalogy’ would be with individuals, where it might bethat someone can have a right to do what is morally undesirable. Jeremy Waldron, forinstance, has argued that ‘if we take the idea of moral rights seriously, then we haveto countenance the possibility that an individual may have a moral right to do somethingthat is, from the moral point of view, wrong’ (1981: 21). Even if true, there is somethingadditionally especially difficult about justifying the desirability of affording legitimategovernment a right to do wrong while holding that government in general is unjustifiedin so acting, presumably because ‘doing wrong’ harms people or is in some respect badfor them. Rather than claiming that individuals can have a right to do wrong, it would belike claiming that individuals in general have no right to do wrong, except for upstandingindividuals who do enjoy such a right. Note that intuitions about desert would not helphere: it would seem perverse to claim that legitimate government has a right to harm itscitizens because it deserves such a right in virtue of it managing to be legitimate. Even ifall individuals have moral rights that extend to wrongdoing, whether based on autonomyor the general nature of such rights, this both need not apply to institutions and is notanalogous to holding that only some institutions have a right to act wrongly. We cannotjust simply transfer anthropocentric intuitions to governments, and thus need an argu-ment for why it is desirable for some governments to enjoy a right to act wrongly by vir-tue of their legitimacy when this is something undesirable in government in general.

(3) Holding that legitimate states are entitled to enforce existing rights

If we make a legitimate state’s unique right to rule necessary for justified coercion, werob the illegitimate state of the capacity to use coercion when it is desperately needed toavoid heinous outcomes. If we make such a right merely sufficient, it becomes redundantwhen coercion is, on its merits, justified and undesirable when such coercion is, on itsmerits, unjustified. Yet, what if the value of legitimacy is that we confer to a legitimatestate the right to enforce certain rights that exist independent of the presence of the state,where this right is also a right against other claimants to be in such a position?13 Underthis recognizably Lockean stance it is not that legitimacy creates new rights and duties,but rather that it identifies the potential authority which could legitimately enforce them,presumably overcoming what Locke termed the ‘inconveniences’ of the state of nature.In an illegitimate state, these rights remain decentralized and can be enforced either byno one or, alternatively, by everyone.

Yet the difficulty with this conception of the value of legitimacy is that it eithermorally forbids a range of actions that are, on their merits, deeply morally desirableor it merely confers on the legitimate state the capacity to undertake the enforcementof certain rights in situations in which it is, on its merits, not justified in so doing. In otherwords, if in an illegitimate state no one has the collective right to enforce our rights, thenthe problems of counter-response (1) are going to accrue. If everyone has such a right,


then the problems of counter-response (2) are going to arise. The problem of theenforcement of pre-existing rights is thus going to end up with derivative difficulties forthe direct creation of a legitimate state’s right to coerce.

That is, it is the capacity to overcome various collective action problems that identi-fies the value of a state enforcing certain rights, yet this is often going to be a structuralfeature of state power (as Hobbes perceptively highlighted) rather than necessarily a fea-ture of being legitimate, unless we embrace the Hobbesian route and make a state witheffective enforcement by definition legitimate. For example, consider a right to punishattempted murderers stemming from our natural right as owners of our own bodies. Alegitimate state enjoys such a right as we (so the argument goes) have transferred thatright in order to overcome a range of practical difficulties with effectively using sucha right ourselves. But consider those cases in which an illegitimate state could effectivelyenforce such a right – should it do so? If legitimacy is necessary for such action, then weforbid what is, on the enforcement theory, eminently desirable: that we have our rightsrespected. Illegitimate states often have the capacity to coerce would-be murderers, andin so doing, solve a range of problems present in the Hobbesian and Lockean states ofnature, notably our physical vulnerability as individuals. But if the concept of legitimacyentails that illegitimate states lack a right to coerce and, as a result, should not evercoerce or that they are not able justifiably to enforce those rights that we already possess,even when doing so is of immense benefit, then this will leave the inhabitants of illegi-timate states decisively and maybe horrifically worse off.

If, alternatively, the illegitimate state would be justified in so acting, as all such statescan justifiably enforce our basic rights, then in transferring to a legitimate state thecapacity to enforce such rights generally, we either transfer nothing (if such a right onlyextends to the capacity to do what is just) or we provide it with an extra capacity to bejustified when acting wrongly (and this is all we do). Again, legitimacy seems to haveeither a redundant or undesirable impact.

(4) Reasonable disagreement

A fourth response might be to claim that legitimacy’s justificatory role or deserved obe-dience is crucial in precisely those cases in which it is uncertain or debatable whether thestate action is justified or obedience due. A legitimate democratic state is justified inusing coercion or owed obedience because there is a lack of consensus, some contro-versy, or reasonable disagreement. Given the pluralism inherent in modern states, thiswould presumably be very often indeed, if not always.

There are two obvious ways of viewing the importance of reasonable disagreement.The first is as a challenge to stability in that such disagreement, by virtue of its reason-ableness, should not be suppressed or ignored, and thus we should want policies andinstitutions that all can reasonably accept or not reasonably reject. This is, I think, thebest way to understand Rawls’s motivating problem of Political Liberalism, namely,‘How is it possible for there to exist over time a just and stable society of free and equalcitizens, who remain profoundly divided by reasonable religious, philosophical andmoral doctrines?’ (1993: 4). Under this conception it is notably not mere stability thatis desirable, but rather stability achieved in a certain way (by respecting people as free

Coakley 355

and equal). Such a project is, however, minimally about the justification of particularlaws and institutional choices based on their content: Is, for instance, a law banning med-ical marijuana (or taxing inherited income or setting up an independent central bank oran unelected supreme court) something all could reasonably agree to? Such a question,of itself, does not require us to establish any independent normative value to legitimacyoutside a purely merit-based appraisal of the law or institution in question, and whatRawls gives us is the hope that there will be sufficient overlap between reasonable peo-ple for this to be possible.

But we can in principle go beyond this and hold that the presence of reasonable dis-agreement helps establish the desirability of a general obligation to obey the law of alegitimate state, or of the legitimate state having a general justification in the use of forceand coercion or use right thereof. The very fact that we disagree indicates why, on thisview, such an obligation, justification, or right has moral value. If correct, it wouldanswer the counterfactual question and arguably be of relevance to all modern states,where, even if acting in good faith and concerned for others, it seems undeniable thatwe will often disagree and are very likely to continue to do so.

Reasonable disagreement seems inevitable. But its presence does not obviouslyaddress the specific worry. To see why, for the sake of argument let us assume the fol-lowing are true: (1) there are multiple individuals in the society with conflicting moraldoctrines, all of which, however, are reasonable; (2) that the means by which the statearrives at decisions are both just and legitimate; and (3) that all the individuals shouldand do accept that the means by which decisions are arrived at are both just and legiti-mate. These may be very demanding conditions (notably, the combination of the first andthird), but they are the assumptions most sympathetic to the disagreement theorist, so letus grant them. Given these facts, should the individuals accept the desirability of a gen-eral duty to obey such decisions or of the state having a general justification in the use offorce and coercion?

Take, for instance, a particular law, ‘Do Q,’ about the merits of which the individualshave differing opinions based on their comprehensive moral doctrines. Here again arethe three salient possibilities.

1. Doing Q is required by reference to those affected by doing Q, and thus one shoulddo Q, whether commanded by a legitimate or an illegitimate state.

2. Not doing Q is required by reference to those affected by doing Q, and thus oneshould not do Q, whether commanded by a legitimate or an illegitimate state.

3. Not doing Q is required by reference to those affected by doing Q, and one shouldthus not do Q, unless commanded by a legitimate state.

Now assume that there are some individuals in the society who think that, for this par-ticular law, case 1 obtains, some other individuals who believe that case 2 obtains, andstill other individuals who believe that case 3 obtains. All are reasonable, but they dis-agree. However, what they should be able to agree upon is that if case 1 or 2 obtains, thenlegitimacy, via a duty to obey, has a redundant impact. If case 3 obtains, it has an unde-sirable impact. In other words, they (reasonably) disagree as to whether, in this particularcase, a duty to obey is redundant or undesirable. But the counterfactual problem is that a


duty whose impact is either redundant or undesirable seems to have potential disvalue.Exactly the same problem will accrue to the evaluation of coercive acts. Thus, the under-lying worry is that not only does the presence of reasonable disagreement not establishthe value of a duty to obey or right to use coercion, but the fact that that such a duty or aright are prima facie undesirable should potentially be something recognized as part ofthe overlapping consensus between reasonable people. Here, after all, are two types ofquestions we might ask.

A. Is an obligation to obey the law of a legitimate state desirable?B. Is this particular law, X, judged on its specific merits, something we should act in

conformity with?

Here are alternative questions.

A. Is a legitimate state enjoying a general justification of the use of force and coercionor having a right to so act desirable?

B. Is this particular act of coercion, Y, judged on its specific merits, justified?

Reasonable people will often disagree about the B questions. But the presence of suchdisagreement does not represent an argument that answers the A questions. If a state isjustified in using coercion because it is legitimate, then the undesirability of the counter-factual import of this justification is a conceptual consequence of the nature of the jus-tification. We may lack agreement, for instance, on which of the three salientpossibilities (1, 2, or 3) outlined previously obtain in a particular case. Yet whicheverone obtains, legitimacy has a counterfactual import that is either redundant or norma-tively unattractive.

That is, whether we disagree if a particular law is bad, unjust, or unwise does notimpact on whether it is desirable that a law that is bad, unjust, or unwise should beobeyed if a legitimate state commands us to, but disobeyed when all other states so com-mand. Similarly, that we disagree if a particular act of coercion is, on its merits, morallyunjustified does not show why it would be desirable for a legitimate state justifiably todo, or have a right to do, those acts which are, on their merits, morally unjustified. Theproblem arises from the very notion of legitimating acts based on their providence ratherthan on normative facts about the acts, and holds even if, in particular cases, the norma-tive facts about the particular acts are disputed or unclear. (In subsection (6) below, Irespond to the very important claim that a legitimate government is a more reliableauthority about the normative facts, and thus is prima facie justified in its use of forceor owed our obedience to the law.)

Thus, we can develop the Rawlsian project of Political Liberalism as it relates tolegitimacy in perhaps two broad ways.14 The first is as providing a content-based methodof evaluating laws and institutions (that they could be justified based on an overlappingconsensus), whereby laws and acts of coercion are in principle solely evaluated on theirmerits, not on how they are produced or by whom. Legitimacy here is ultimately a way ofdescribing the acts’ desirability in securing stability and acceptance while respectingpeople as free and equal. As such, the counterfactual problem does not arise because

Coakley 357

legitimacy has no separate normative value aside from that based on the content of thelaw or coercion (the previously discussed case 3 cannot arise) as legitimacy simply hasno counterfactual import. That is, under this conception, if a non-democracy were toadopt coercive taxation policies based around the difference principle, it would act justlyand legitimately. Were a democracy to adopt taxation policies that could not be justifiedby the overlapping consensus, it would act illegitimately, even if the decision to do sowas passed by fully just and legitimate institutional arrangements. Alternatively, however,we might say that if institutions are legitimate (such that they could be justified based onthe overlapping consensus and therefore accepted by reasonable people), then their deci-sions are, as a result, legitimate and we have a duty to obey them and coercion undertakenby them is generally justified.15 If so, as indicated above, the fact of reasonable disagree-ment cannot establish the desirability of such a duty or justification, and the problemsremain. The underlying problem of this article, though framed as a clash between thelegitimacy of institutions and the justness or desirability of their decisions, is ultimatelya problem of how we relate the evaluation of what institutions do to the evaluation of theinstitutions. Even if, as with Rawls, we make justice and legitimacy potentially cohere (inthat the first entails the second), the question remains as to how to relate the evaluation ofwhat is done to the evaluation of by whom or how the decision is arrived at.

Thus, the Rawlsian project of Political Liberalism is not necessarily vulnerable to thecounterfactual problem unless we assert a duty to obey or general justification of the useof coercion or right thereof, but if we do, then it does not, via the fact of reasonable dis-agreement, have special resources to overcome it, despite its other immense merits. Ulti-mately, we require a clear and compelling response to the counterfactual problem;reasonable disagreement does not provide it.

(5) Distinguishing moral requirements and permissions

Previously, in setting out the problem as it relates to obligations, all acts of obediencewere parsed into those that, based on their merits, either one morally should or shouldnot do. But what if legitimacy makes a difference in precisely those situations in whichobedience is, on its merits, morally optional? That is, if both a legitimate and an illegi-timate state command one to obey a law X, then the difference legitimacy makes is that ifobeying such a law is, on its merits, morally optional, then you should do so if com-manded by a legitimate state, but not necessarily if by an illegitimate state.

There are three possible cases here.

1. Obedience to law X would harm the relevant human interests at stake, but not suffi-ciently to be normally morally forbidden.

2. Obedience to law X would promote the relevant human interests at stake, but notsufficiently to be normally morally required.

3. Obedience to law X would neither promote nor harm the relevant human interests atstake.

Now, O1 does not circumvent the earlier worry, for legitimacy requires one to pro-duce a harm that morally one would otherwise not be required to produce: it may be a


marginal harm, but it is a harm nonetheless, and legitimacy has apparent moral disvalueand the counterfactual problem accrues. In cases of O2, however, obedience owed to alegitimate state requires one, in such a state, to do what is generally morally desirable,but not normally required: it may, for instance, involve providing some benefit to othersin the society or some deserving group that goes beyond one’s normal moral duties, suchas supporting museums via taxes, volunteering to help the needy, or obeying the bylawsof a park. Under this option, it is not that legitimacy requires one to do something that ismorally bad, as it does with a general duty to obey, but merely to go beyond one’s basicrequirements.16 This avoids the previous worries, as what the legitimate state commandsis something that is advantageous overall, to some others, to civil society, or to someco-citizens, and so on. However, the following is the difficulty with this.

Double punishment: if we owe duties under a legitimate state and they are to the benefit of

all or some deserving group, then we have a reason to fulfill these duties whether the state is

legitimate or not; fulfilling them simply depends on considerations of efficacy and practic-

ability. Thus, if we have less duties under an illegitimate state, then we potentially cause its

subjects to suffer from double punishment: they suffer from a worse government and from

the lower level of concern represented by less duties owed to them.

For example, if we have additional duties to pay taxes to support various cultural publicgoods under a legitimate state, and this, for instance, benefits the very poor, then if wewere to suffer some bad fate (say, a coup), the poor would end up being doubly punished:they would have the bad luck to be ruled by an inferior government and the wealthywould, additionally, now have less of a duty to them. This seems unattractive. (Now,it might be that some of the duties of the wealthy under an illegitimate regime wouldbe less easy to discharge practically as the government cannot be trusted to spend taxmoney wisely or justly, but this is an argument for efficacy, not for less duties: we justcannot reliably fulfill our duties to help those worse off, but if we could, we shoulddo so.)

The problem is that legitimacy-based duties have the potential to harm people in twoways: they can require someone in a legitimate state to do something that is, on its merits,morally bad or they can permit those living in an illegitimate state to avoid helping otherseven though this might be, given the suboptimal government, even more desirable. Per-haps the perversity of this can be reduced if we restrict the scope of legitimacy-groundedduties to things sometimes considered low impact, so that being morally required to doso if commanded by a legitimate, but not by an illegitimate state leaves those generallysubject to illegitimate rule only marginally harmed by our account of the value of legiti-macy. But even if so, the fact that we may need to restrict duties of obedience to thosethings normally merely morally optional and of low impact in order to minimize theharm legitimacy does implicitly confirms the substantive point: legitimacy has norma-tive disvalue.

But what about in cases of O3, in which obedience would neither promote nor harmthe relevant overall human interests at stake? Here it seems that the subjects of a legit-imate state and of an illegitimate state are morally unaffected by the counterfactualimport of legitimacy: neither group is harmed by what is required, unlike with options

Coakley 359

O1 and O2. We may have had to restrict the impact of legitimacy to a very small set ofcases, but it seems here at least not to have any moral disvalue. This, however, is tooquick, for imposing duties on agents is plausibly a burden on them at least. We normallytake this burden to be greatly outweighed by the importance of the duty, but in cases ofO3 this is precisely what cannot be the case. It is true that being required to obey a law oredict whereby obedience neither harms nor helps those affected is likely to be a smallburden, but if this is all legitimacy does, then we have both restricted the concept’simport to a tiny subset of laws and still imposed (in the imposition of a moral duty) acost not justified by reference to any human interests at all. The disvalue may be small,but it exists, and even if only applying to what is neither morally required nor forbiddenlegitimacy fails to have any positive normative impact.

(6) Appealing to epistemic authority

A much more promising response is to argue that, properly constituted, legitimate gov-ernment is a more reliable authority about what we ought to do, and about when force isjustified, than illegitimate government. Now, to overcome the counterfactual problem,such an epistemic approach is going to need to hold that legitimate government does notcreate duties per se, and does not have a general justification for the use of coercion nor ageneral right to coerce. Thus, although it contains a series of very interesting epistemicclaims, the approach of someone such as Estlund is still going to be subject to the coun-terfactual worry, as his theory combines a thesis about the epistemic reliability of democ-racy with a substantive theory of the value of legitimacy – specifically, that in ademocracy the ‘authority and legitimacy of its laws often extends even to unjust laws,though there must be limits on this’ (2008: 8). As such, the concerns raised previouslyaccrue to the theory, since he rejects a pure epistemic route (what he usefully calls a ‘cor-rectness theory’), seeking instead ‘to show how democracy yields moral reasons to obeythe law and a moral permission to enforce it’ (Estlund, 2008: 7). The counterfactualproblem is that such reasons and a permission seem to be normatively undesirable, anda theory that assumes they are desirable will be subject to the worry. But a pure correct-ness theory (most famously, perhaps, that of Raz’s service conception) would potentiallyovercome the counterfactual problem because it quite explicitly does not hold that legit-imate government creates additional new reasons to obey or to judge force justified,rather that it might sometimes be a better judge of when someone should obey or whenforce is justified based on the reasons that already apply.

The central difficulty with a purely epistemic account of obedience and deference tothe assessment of force of legitimate authoritative institutions, however, is that it may notbe able to establish any very clear relationship with legitimate institutions that, in expec-tation, produce better laws than illegitimate ones. To see why, consider the underlyinglogic of epistemic authority.

Raz, applying his ‘Normal Justification Thesis’ to states, writes that ‘The main argu-ment for the legitimacy of any authority is that in subjecting himself to it a person is morelikely to act successfully for the reasons that apply to him than if he does not subject him-self to that authority’ (1986: 70–1). Such an argument is not vulnerable to the type ofcounterfactual worries outlined previously as it does not claim that the fact of a


government being legitimate creates new additional reasons to obey, rather that you willhave a ‘pre-emptive’ reason to obey if the government is a better judge of what you oughtto do than you are. As he observes, sometimes this may be the case, sometimes not.

Raz gives us a coherent and cogent account of authority. To link it to the claim thatone has a pre-emptive reason to judge force to be generally justified for or obedience dueto a legitimate state, it seems like we need the following, not utterly implausible, claim.

L1: a legitimate government’s commands are likely to be a better guide to what a person has

reason to do (or to the true assessment of justified coercion) than that person’s own

judgment.

At first blush, this gives a person a general reason to obey the commands and accept theclaims of justified force made by a legitimate state. Democracies would be legitimate onthis account if and when they normally judge better when coercion is justified and whenone ought to obey than citizens themselves. This is a contestable claim. But even if true,it is not enough. For there is an ambiguity between being an authority relative to theagent and an authority relative to other potential authorities. If two ‘experts’ offer some-one advice on what to do, even if both are better judges of what the person has reason todo, if one expert is a better judge than the other, then the person has a reason to take theadvice of her alone. Similarly, if a state commands one to do X or claims that its coercionis justified, one should accept this not only if one thinks the state is a better judge ofwhether one should do X or of whether its coercion is justified, but if it is the best judgeavailable. There is, after all, no such thing as an epistemic authority simpliciter – the con-cept is only meaningful in comparison. That is, ‘Belief-holder J is an authority about Q’is underspecified, and only potentially correct as a claim that ‘Belief-holder J is anauthority about Q compared to belief-holder K.’ Thus, for a government to be legitimateas an authority what we require is for the following to be true.

L2: a government’s commands are likely to be not only a better guide to what a person has

reason to do (or to the true assessment of justified coercion) than the person’s own judg-

ment, but the best guide of any of the potential authorities available.

I personally doubt that this is very often the case, but we each must decide. However,what such considerations highlight is that, rather ironically, it may be that subjects oftenhave more pre-emptive reasons to accept the commands of, and claims over force madeby, oppressive states than democratic ones, even if the latter more reliably produce goodlaws and justified coercion. If democratic states are often accompanied by relativelyopen societies and political processes, such that those who disagree with the governmentfrom outside and those within the governing system who oppose official policy can com-municate their views and have them disseminated, then we will have a great many poten-tial authorities to contrast with the state’s claims. If legitimate states are democratic, forinstance, we may thus face a certain irony of authority and legitimate states.

The irony of authority and legitimate states: if obligations to obey, or justifications of coer-

cion, are epistemically justified, then the more open the political process and public sphere,

the less likely it is that the state will be the pre-eminent authority for a given agent on a

Coakley 361

given issue. If legitimate states result in open societies, then legitimate states may be less

likely to be the pre-eminent authority for subjects on whether one should obey the law or

accept that state coercion is justified than oppressive illegitimate states, even though the lat-

ter may be expected to produce more bad laws and use unjustified force more frequently.

For example, if an oppressive illegitimate state passes a decree that one must observe acurfew, not eat fish, or temporarily evacuate one’s town due to an imminent threat, then(if, say, the media is controlled and dissenting views are hard to find) one may be leftwith deciding whether the state is likely to be a better judge of what one has reason todo than oneself. In those cases in which it is (and government does have structural advan-tages over private citizens in judging national threats in many situations), one seeminglyhas an epistemic and pre-emptive reason to obey. Under a legitimate state, if one hasaccess to a wide variety of potential authorities and dissenting views, the state isextremely unlikely to be the pre-eminent normative authority for an agent. Whether astate’s commands or claims are authoritative depends not merely on the extent to whichit is a better judge than the agent, but also on whether it is a better judge than other poten-tial authorities. This crucial second clause highlights why even the view that legitimatestates do not create duties or are not per se justified in the use of force, but merely aremore reliable at identifying what one should do and when coercion is justified does notentail any general presumption in favor of obedience or acceptance of their coercion asjustified. If legitimacy depends upon, or accompanies, open political processes and opensocieties, then any general presumption is likely to be undercut by the state’s loss of pre-eminent authority. Even the very contingent and limited epistemic duties and justifica-tions potentially grounded upon Raz’s Normal Justification Thesis fail to underpin anynecessary connection between legitimacy, good governance, one’s duties, and the gen-eral justification of force.

With this in mind, it is possible to see why the claim that legitimate government is anormative authority is, depending on how this is spelled out, going to suffer from theepistemic concerns above, the initial substantive counterfactual worry, or indeed both.Consider, for instance, Joshua Cohen’s account of epistemic populism (1986), wherebyappropriately constituted and motivated, the decisions of majorities provide good evi-dence about what is the general will, that is, the best way of furthering the shared con-ception of the common good. Cohen’s primary goal in setting out epistemic populism isto show how it could be coherent as a collective decision procedure (in that it overcomesworries about majorities lacking coherent transitive preferences), not necessarily toargue that this is the best foundation of political legitimacy. But it does helpfully providea clear potential specification of the claim that majorities can be expected reliably toidentify a general will which we should then obey.

It is important to note, though, when analyzing how epistemic populism relates to theearlier problems concerning the value of legitimacy, that there are two important ideaspotentially commingling here: (1) that majorities are reliable epistemic authorities aboutnormative duties and justifications and (2) that the decisions of majorities establish nor-mative duties and justifications. The first does too little; the second too much. For thefirst fails to show that majorities are necessarily the best epistemic authority of all thoseavailable, and indeed in an open democratic society it would seem bizarre to think they


were. The second claim suggests that when majorities are wrong and pass laws thatnormally we should disobey, that because it was a majority that was wrong we shouldobey, with the accompanying infliction of harm (if majorities cannot be wrong, thenwe have traded a plausible claim about their potential reliable correctness in expectationwith a seemingly implausible one about their infallibility). In other words, if a majoritypasses a law that, on its merits, we should obey, then that a majority passed it is super-fluous in justifying obedience. If a majority passes a law that, on its merits, weshould disobey, then that a majority passed it has an undesirable impact if it entailswe should obey what we normally morally should not obey. The fact that, possibly, inexpectation, majorities may do more of the first than the second simply means that theywill provide more superfluous rather than harmful authorization, but this fails to estab-lish why it is desirable they do any harmful authorization at all, why we should obey gov-ernment is because it is legitimate. The problem remains.

The initial challenge for epistemic accounts is that they need to establish not only thatlegitimate government is a better authority than illegitimate government and that it is abetter authority than the citizen, but that it is also the best authority of all those available.This I described above as ‘bizarre’ to think true of the decisions of majorities in opensocieties, in which there are a very large number of alternative potential authorities. Butthere is one notable theory, Condorcet’s jury theorem, that if it applied to voters couldmake such epistemic superiority nearly inevitable (Baker, 1976: 33–70). Under thefamous theorem, if each voter has a probability of being right of greater than a half,17

voters are voting truthfully on the same binary question,18 and votes are statisticallyindependent, then the probability that the majority is right approaches one as the groupsize approaches infinity. More relevantly, it quickly approaches one if the voters are alittle bit better than a coin flip and fairly numerous, or at least as numerous as in all mod-ern states. Does this establish the value of legitimacy? It seems on the surface that itmight, as if true, the theorem would establish the near infallibility of majority decisions,and thus that as fallible individuals we should defer to the majority’s judgment.

Now, clearly one problem is the empirical validity of the theory, in that the basic Con-dorcet jury theorem (CJT) model makes a series of predictions (including that majoritiesin two large jurisdictions will almost never disagree and that voting choices are not sig-nificantly correlated with income, race, age, religion, and so on19) that are testable and, atleast for all existing democracies, demonstrably false.20 Thus, the theory obviously has ahurdle to clear in that it appears to be directly refuted by most of the main psephologicalfindings of the past 100 years. But even if someone could show that it somehow didapply, and thus that majorities are in fact nearly infallible, perhaps the more fundamentalpoint is that it is not the state’s legitimacy (as we commonly conceive it) that does anywork here: it is the expectation of majorities almost never getting any decisions wrong.That is, the concept of legitimacy does not obviously add anything: it has no value in andof itself.

Still, if someone did believe that majorities were close to infallible and as such asuperior judge to all others in society, then they would have a reason to defer to theirjudgments and it might look like such deference represented the establishing of the legiti-macy of such governance, even if on examination it is their infallibility that is key. Thisis what an account of an epistemically grounded legitimate authority requires: that the

Coakley 363

legitimate institutions be the very best judge of when obedience is due and force justifiedof any of the agents and groups that can provide such judgments. This seems to meimplausibly demanding of any realistic modern institutional form, but we do have a the-ory (the CJT) that if only it were true of human voters would potentially establish this,and it remains important for this reason. Such a theory does not, however, really establishthe independent value of legitimacy, only the usefulness of deferring to a (supposed)nearly infallible decision-maker.

(7) Legitimacy and institutional improvement

Finally, even if legitimacy-based duties and justifications are normatively undesir-able in principle, it still might be that belief in them (and by implication a beliefin the importance of legitimacy) is desirable in practice, most notably perhaps byencouraging the spread of democracy. On this psycho-political view, the value oflegitimacy is in labeling one set of institutional features as qualitatively better thanany other, and thus encouraging people to strive for those features. This is anempirical claim about how our normative language motivates agents with influenceover institutional change. I am somewhat skeptical about it, but for the sake of argu-ment let us assume it is plausible. Does this demonstrate that legitimacy is norma-tively desirable? It does not. What it highlights is the potential instrumental value ofsome people believing it is, not that it is. The belief in a vengeful God, for instance,may make some people behave morally better and therefore promote certain goodconsequences, but that does not demonstrate that God is vengeful, or exists. Thus,even if we have some expected benefits from the widespread belief that legitimacyis normatively desirable, and an equating of this with certain democratic state struc-tures (such as an electoral executive and legislative selection), this does not showthat legitimacy is in and of itself normatively valuable, only that popular belief tothat effect in certain situations may be, and in others may not. It is a very contingentand weak account of the value of legitimacy.

It is also one with risks. For while a widespread belief in the unique legitimacy ofdemocracy may plausibly encourage transitions to this state form, it may also provideexcessive authorization of subsequent actions: the fact that an unjust, unwise, or uncon-scionable act is undertaken through a democratic procedure may lend it an apparentlegitimacy that distracts from its injustice, folly, or moral egregiousness. It may thereforeserve to give a normative veneer to those undesirable practices that seem entrenched fea-tures of democratic nation-states, notably the low or negligible level of concern system-atically afforded to those outside the state’s territory or to future generations influencedby what it does. Not only does the psycho-political defense of democratic legitimacyidentify not the correctness of a belief in the value of legitimacy, but rather merely thepotential instrumental utility of such a belief, there is also a real possibility of it fosteringa sort of normative confusion whereby we conflate popularity with justice. To misleadpeople normatively has dangers.

In addition, it may be that the very nature of the notion of legitimacy as currently con-strued will unjustifiably prejudice us toward certain types of institutions and not others ifwe make it primary in institutional evaluation. While a Fridocracy enjoys no obvious


advantages over the representative democratic republic, a Lottocracy or some variantthereof might, at least in that it is less likely to show bias toward the wealthy, well born,and well known. Yet the popular identification of legitimacy with consent, and confla-tion of that with voting, may hinder a clear-headed and widespread assessment of therelative merits of a range of state forms. This is not a hypothetical worry. Manin haschronicled the really striking historical transformation from lotteries being viewed asinherently democratic and elections as being aristocratic to the equating of democracywith voting, and the latter with consent. ‘There is every reason to believe’, he writes,‘that it is this view of the foundation of political legitimacy and obligation that led to theeclipse of lot and the triumph of election’ (Manin, 1997: 86). The implication of this arti-cle is that, potentially, this enormous change in political thought and institutional evalua-tion was wrongly premised.

Conclusions

The core problems highlighted previously have the same basic structure. If the factthat a state is legitimate creates duties or justifies coercion, then we can ask the fol-lowing counterfactual question: When does this change what we should do, or whendoes it change our overall assessment of an act of force or coercion? In both cases,it does so when the balance of more general reasons we have (based on people’sinterests or on what they deserve) is outweighed by the new duty or justification.That is, it only makes a difference when it ‘tips the scales’ against people’s inter-ests, all things considered, or justifies what is, on its own terms, unjustifiable. Bothimplications seem deeply unattractive. Yet they merely flow from the fact that theinstitutional source of laws and coercion does not necessarily remake people’s inter-ests nor alter whether they deserve punishment or if coercion produces relevantlygood consequences.

The overall problem of political legitimacy is commonly treated as a dispute aboutwhat would need to be true of a state or polity so that it would have a right to use coercionand be owed obedience. This assumes that such rights and obligations are normativelydesirable when held by a legitimate state, but not by an illegitimate one. Yet this requiresjustification and, as I have tried to show, that is not straightforward at all.

Given that a compelling account of the value of legitimacy would also plausiblyhelp us settle the question of the conditions of legitimacy, a theory of politicallegitimacy that fully addresses the counterfactual problem will, prima facie, besuperior to its competitors. Recognizing the problem, however, is a prerequisite forsuch progress.

Notes

I am very grateful to Barbara Buckinx, Ronald Dworkin, Russell Hardin, Dimitri Landa, Peter

Northrup, Ryan Pevnick, Michael Rosen, Lucas Stanczyk, two anonymous referees at Politics,

Philosophy and Economics, participants at the Northeastern Political Science Association 2008

Annual Meeting, the Harvard Graduate Conference in Political Theory 2009, and the NYU Polit-

ical Theory Group for comments, and in particular to Michael Kates and Bernard Manin for very

helpful discussions about the overall problem. A draft of this article was awarded the Northeastern

Coakley 365

Political Science Association Wilson Carey McWilliams Prize for the Best Political Theory paper

of 2008. I am grateful to the Northeastern Political Science Association for their support and

encouragement.

1. Skepticism about such theories often worries that the stipulated conditions are in fact never

met, or met in ways that invalidate the overall claim. Perhaps the most famous early example

of such a critique is David Hume (1985).

2. A coherent anarchist position is that legitimate government is not necessarily undesirable,

but rather, in modern states, simply unachievable in that it could only ever be grounded on

actual, explicit, freely given consent. This seems compatible with the central arguments of

both Simmons and Green in their skepticism about whether we have ever (and plausibly,

for most people, will ever) meaningfully consented to a state in a way that would correctly

obligate one. See Simmons (2001) and Green (1989). Thus for someone who accepts Sim-

mons’ broad position (that a political authority may be morally justified in using coercion

without necessarily being legitimate – the latter being required to create a duty to obey),

the counterfactual problem speaks to the potential undesirability of legitimate government,

not that of having a political authority. Simmons, and much traditional skepticism, worries

that we have not meaningfully consented to a duty to obey. The counterfactual worry, by

contrast, implies that we should not so consent, as a duty to obey appears morally

undesirable.

3. A fuller account of the case for this understanding of Hobbes is to be found in Russell Hardin

(1991).

4. Though Nozick seems to portray a range of state forms as illegitimate, it is, strictly speaking,

only the undertaking of certain actions that is illegitimate under his theory. A democratic state

could, in principle, decide only to coerce in the same manner as a minimalist one and, thus,

always act legitimately, though in practice it seems deeply implausible that it would do so.

His theory is, thus, despite appearances, best thought of I think as a theory of right or legiti-

mate action, not of the legitimacy of any institutional forms per se, though this is an atypical

reframing. See Robert Nozick (1974).

5. For a very interesting proposal that moves in this direction, see Corey Brettschneider (2007).

6. A very clear taxonomy can be found in Fabienne Peter (2007).

7. Note that these consequences could be due to the particular instance of the punishment or more

generally due to the set of instances falling under the punishment rule as an institutional pol-

icy. See John Rawls (1955).

8. ‘Those affected’ may include giving extra consideration to the agent’s own prudential con-

cerns: the argument is not meant to assume a strong impartiality where the interests of the

agent are only given the same weight as those of all others in assessing what should be done.

It is thus intended to be perfectly compatible with what Samuel Scheffler has called an ‘agent-

centered prerogative’. See Scheffler (1994).

9. An extremely cogent modern account is Simmons (1979). Such skepticism seems, however, to

be faced with repeated challenges: witness, for example, consent re-emerging in Margaret

Gilbert’s novel and intriguing account of obligation as joint commitment (2006).

10. This, I think it is fair to say, is both the most widespread current assumption about legitimate

government, and true of arguably the most impressive recent accounts in the literature. For an

illuminating comparison of two of these, see Thomas Christiano (2009) and David Estlund

(2009).


11. A ‘Fridocracy’ would seem to be illegitimate for Thomas Christiano (2008), for instance, as

such a state explicitly does not publicly treat citizens equally, affirming the unequal political

status of its citizens who were not born on a Friday. A Fridocracy might also be illegitimate for

Rawls (1993), that is, subjects not born on a Friday seem to be able to reasonably reject a

constitutional order affording them neither voting rights nor political influence.

12. Though arguably eclipsed in the recent literature by proceduralists, legitimacy as consent has

both a long and extremely influential lineage capturing two very powerful ideas. The first,

drawing on Locke (1988), is that we have some pre-political rights that we must give consent

for government to violate (which government typically will). The second, drawing on Rous-

seau (1988), holds that legitimate governance requires a certain type of participation, as this is

the only way to enjoy the value of freedom as self-rule. Consent additionally has great popular

appeal as the foundation of political legitimacy, and thus any broad discussion of the topic will

need to engage with such a view.

13. I am very grateful to the first anonymous reviewer for suggesting that this response may be at

the back of many readers’ minds.

14. Rawls’s primary focus is on the ideal theory case, thus there is much disagreement about how

best to infer guidance on nonideal cases. For a recent discussion, see Simmons (2010). On the

one hand, Rawls’s ‘liberal principle of legitimacy’ clearly applies to constitutional essentials.

See his statement of this in Political Liberalism (Rawls, 1993: 137). On the other hand, the

notion of public reason and justification seems to apply this principle to the decisions of states.

Thus, for the sake of this article, I subsequently assume the reasonableness of interpretations

based on either understanding – that legitimacy is primarily an institutional evaluation or that

it also applies to all decisions.

15. Rawls, plausibly, also leaves these options partly open, as the concept of a ‘basic structure’

clearly does not extend to all policies, but at the same time, clearly does extend to some. See

his discussion in Rawls (1993: 257–88).

16. I think this is the central thrust of Thomas Senor’s response to Simmons when he worries that

political duties may be more extensive than basic moral duties, although his account is not

explicit on whether one would only have these duties under legitimate government. See Senor

(1987). If they hold under all governments, then they have no implications for legitimacy per

se. Thus, for the sake of argument, let us assume some duties that go beyond our basic moral

ones do indeed require legitimate government.

17. There are a range of ways of relaxing these conditions, for instance, by requiring the average

probability of being right to be greater than a half. See, for instance, Bernard Grofman et al.

(1983: 261–78).

18. Regarding relaxation of the binary-question assumption, see Christian List and Robert Goodin

(2001: 277–306).

19. If, as a jurisdiction approaches a million voters, the probability a majority will be correct

approaches one, then the probability that two jurisdictions each of well in excess of a million

voters will disagree should be close to zero. To illustrate the obvious potential problem with

this, Texas and California are large jurisdictions, but they frequently disagree (over choice of

president and a range of policies). If the CJT were true and established the very high prob-

ability of majorities being correct, then their disagreement should be spectacularly rare. In

fact, it is common.

20. For an excellent survey of the recent main US data, see, for instance, Andrew Gelman (2008).

Coakley 367

References

Baker K (1976) Condorcet: Selected Writings. Indianapolis: Bobbs-Merill.

Brettschneider C (2007) The rights of the guilty: Punishment and political legitimacy. Political

Theory 35: 175–99.

Christiano T (2008) The Constitution of Equality: Democratic Authority and its Limits. Oxford:

Oxford University Press.

Christiano T (2009) Debate: Estlund on democratic authority. Journal of Political Philosophy 17:

228–40.

Cohen J (1986) An epistemic conception of democracy. Ethics 97: 26–38.

Estlund D (2008) Democratic Authority: A Philosophical Framework. Princeton, NJ: Princeton

University Press.

Estlund D (2009) Debate: On Christiano’s The Constitution of Equality. Journal of Political

Philosophy 17: 241–52.

Gelman A (2008) Red State, Blue State, Rich State, Poor State: Why Americans Vote the Way They

Do. Princeton, NJ: Princeton University Press.

Gilbert M (2006) A Theory of Political Obligation: Membership, Commitment, and the Bonds of

Society. Oxford: Oxford University Press.

Green L (1989) The Authority of the State. Oxford: Oxford University Press.

Grofman B, Owen G and Feld S (1983) Thirteen theorems in search of the truth. Theory and

Decision 15: 261–78.

Hardin R (1991) Hobbesian political order. Political Theory 19: 156–80.

Hobbes T (1994) Leviathan, Curley E (ed.). Indianapolis: Hackett.

Hume D (1985) Of the original contract. In: Miller EF (ed.) David Hume: Essays Moral, Political

and Literary. Indianapolis: Liberty Classics.

List C and Goodin R (2001) Epistemic democracy: Generalizing the Condorcet jury theorem.

Journal of Political Philosophy 9(3): 277–306.

Locke J (1988) Two Treatise of Government, Laslett P (ed.). Cambridge: Cambridge University Press.

Manin B (1997) The Principles of Representative Government. Cambridge: Cambridge University

Press.

Nozick R (1974) Anarchy, State, and Utopia. New York: Basic Books.

Peter F (2007) Democratic legitimacy and proceduralist social epistemology. Politics, Philosophy

and Economics 6: 329–53.

Rawls J (1955) Two concepts of rules. The Philosophical Review 64: 3–32.

Rawls J (1993) Political Liberalism. New York: Columbia University Press.

Raz J (1986) The Morality of Freedom. Oxford: Oxford University Press.

Rousseau J-J (1988) On The Social Contract, Cress D (ed.). Indianapolis: Hackett.

Scheffler S (1994) The Rejection of Consequentialism. Oxford: Oxford University Press.

Senor T (1987) What if there are no political obligations? A reply to A.J. Simmons. Philosophy

and Public Affairs 16: 260–8.

Simmons AJ (1979) Moral Principles and Political Obligations. Princeton, NJ: Princeton

University Press.

Simmons AJ (1999) Justification and legitimacy. Ethics 109: 739–71.

Simmons AJ (2001) Justification and Legitimacy: Essays on Rights and Obligations. Cambridge:

Cambridge University Press.


Simmons AJ (2010) Ideal and non-ideal theory. Philosophy and Public Affairs 38: 5–36.

Van Parijs P (1996) Is democracy compatible with justice? Journal of Political Philosophy 4:

101–17.

Waldron J (1981) A right to do wrong. Ethics 92: 21–39.

Wolff RP (1970) In Defense of Anarchism. New York: Harper and Row.

About the author

Mathew Coakley is a student at New York University. His PhD explores the basic optionsregarding how to relate the evaluation of agents, actions, and institutions, and what this entails forsocial and political life.

Coakley 369

Documents

On the Value of Political Legitimacy