This is an electronic version of an article published in Political Theory 38.1 (2010): 121-130. Copyright 2010 Arash Abizadeh. The definitive version is available as a PDF from the publisher here.
Democratic Legitimacy and State Coercion: A Reply to David Miller
I have argued that, to be democratically legitimate, a state’s regime of border control must result from political processes in which those subject to it—including foreigners—have a right of democratic participation. David Miller resists that conclusion by challenging my assumptions about the nature of, and relationship between, democratic legitimacy, state coercion, and autonomy. In particular, Miller rejects propositions (1.2) through (3.2) below:
(1.1) Principle of Rule of Law: The state’s deployment of physical force and its threats of punitive harms against persons are legitimate only if carried out according to public, general, impartially applied, standing laws.
(1.2) Principle of Democratic Legitimacy: The state’s laws are democratically legitimate only if those subject to them have a right of participation (on terms consistent with their equality and freedom) in the political processes from which those laws result.
(2.1) Coercion Claim: The state’s laws subject persons to coercion by virtue of credibly authorizing the deployment of physical force and threatening punitive harms against them (backed up by force).
(2.2) Coercion Principle of Democratic Legitimacy: The Principle of Democratic Legitimacy is true (in part) because of the truth of the Coercion Claim, that is, those subject to the state’s laws have a right of democratic participation because such laws subject them to coercion.
(3.1) Autonomy Claim: The state’s credible authorization of the deployment of force and threat of punitive harms against persons  (presumptively) invades their autonomy (by virtue of subjecting them to coercion).
(3.2) Autonomy Principle of Democratic Legitimacy: The Principle of Democratic Legitimacy is true (in part) because of the truth of the Autonomy Claim, that is, those subject to the state’s laws have a right of democratic participation because such laws (presumptively) invade their autonomy (by virtue of subjecting them to coercion).
Against (1.2), Miller claims that only laws imposing positive obligations to undertake particular actions, but not laws imposing negative obligations to forbear, require democratic justification; against (2), that one is subjected to coercion only by conditional threats of punitive harm aimed at compelling (rather than preventing) particular actions; and against (3), that one’s autonomy is compromised only by compulsion-threats (rather than prevention-threats). Although I shall begin my defence of (1.2) by defending (2) and (3), it is important to see that, in fact, one may subscribe to (1.2) without accepting the specific grounds articulated in (2) and/or (3) (and may subscribe to (2) without accepting (3)). This is because there are different ways to motivate the Principle of Democratic Legitimacy. My argument is that anyone who accepts this democratic Principle—even should they reject the Coercion Principle or Autonomy Principle—is thereby committed to my thesis about borders. To deny that thesis, it seems to me that Miller is forced to exempt a vast array (perhaps even the majority) of the state’s laws from the requirement of democratic justification.
1. The Nature of State Coercion
Section 25 of the Criminal Code of Canada declares that persons “authorized by law” to enforce the state’s laws are, if they act “on reasonable grounds, justified . . . in using as much force as is necessary for that purpose.” Officers are authorized, for example, to use “force that is intended or is likely to cause death or grievous bodily harm to a person” where such force is required to prevent (arrest, arrêter) the person from causing “death or grievous bodily harm” to others. Section 235, moreover, declares that “Every one who commits first degree murder or second degree murder is guilty of an indictable offence and shall be sentenced to imprisonment for life.” These laws are paradigmatic of two distinct types of legal coercion: the first type of law authorizes the use of physical force and even violence; the second threatens to inflict punitive harm against persons should they violate the law. The former authorize coercive acts; the latter issue coercive threats. 
Of course laws themselves do not physically force anyone to do or not do anything; nor can laws impose the sanctions they threaten. The actual enforcement of the law relies in part on another (and perhaps the most obvious) form of state coercion: coercive acts themselves. These occur when state agents directly deploy physical force against an individual: for example, to prevent her from violating (or continuing to violate) the law, or in the course of administering sanctions for such violations. This direct form of state coercion is not what I here call legal coercion: although the rule of law requires the state’s coercive apparatuses, such as police and military, to act only within the framework of a legal system that authorizes (and constrains) their use of force, these apparatuses are not themselves laws.
The modern Rechtstaat, then, typically subjects individuals to its coercive power in at least three ways: its agents sometimes subject individuals to noncommunicative direct coercion (coercive acts); its laws subject individuals to noncommunicative legal coercion (in authorizing coercive acts); and its laws subject individuals to communicative legal coercion (in threatening punitive harms). Direct state coercion can often be fairly limited in scope: the persons subject to it are those whom agents of the state directly and forcibly restrain or compel (by imprisoning, beating back, pushing, pulling, etc.) Therefore, even though the credibility of the state’s legal authorizations or threats depends on its (typically overwhelming) capacity for direct coercion, it is nonetheless important to see that the coerciveness of (credible) law inheres in the authorization or threat itself, and not in the actual deployment of force or infliction of harm. To take the latter case first: it is true that if the threatened harm comprises a deployment of physical force (such as beating back or incarceration), then carrying out the threat would be an additional instance of (direct, noncommunicative) coercion. But the threat itself is a distinct and prior instance of coercion, one that, unlike its noncommunicative counterpart, operates by bringing pressure to bear on the coercee’s will. Similarly, the authorization of force, when credibly institutionalized, involves standing coercive apparatuses whose agents surveil or “invigilate” people, that is, they stand ready and disposed to deploy physical force against people should they be violating the law. Those who are (legally) subject to such invigilation, like those subject to the law’s threat of punitive harms, are subject to legal coercion.
In his provocative essay, David Miller joins Friedrich Hayek in making the counterintuitive claim that laws such as sections 25 and 235 are not in fact coercive, that is, do not normally subject people to coercion. In The Constitution of Liberty, Hayek began by distinguishing, as I have done, between the direct deployment of physical force and the threat to inflict harm, but proposed to reserve the term “coercion” exclusively for the latter. A person is subject to coercion, Hayek initially suggested, when he receives a conditional  threat of harm intended to bring about a certain conduct in him. But then, in a move that infuriated many of his libertarian colleagues, Hayek proceeded to argue that when such threats take the form of (a) general, (b) impartially applied laws whose violation is (c) avoidable and whose threatened sanctions are known and so roughly (d) predictable, they do not in fact subject persons to coercion: “At least insofar as the rules providing for coercion are not aimed at me personally,” and “[p]rovided that I know beforehand that if I placed myself in a particular position, I shall be coerced,” and “provided that I can avoid putting myself in such a position,” then “I need never be coerced.”
All four of Hayek’s criteria have been subjected to withering criticism by his libertarian colleagues, but my concern here is with the one picked up by Miller: avoidability. Hayek’s point here turns on a distinction, also deployed by Miller, between threats aimed at preventing versus those aimed at compelling a fairly specific action. When “the state uses coercion to make us perform particular actions,” Hayek acknowledged, the coercion is “unavoidable.” (His paradigmatic example, of course, is paying taxes.) But when the threat is preventative—as in a law against murder—the coercion is said to be avoidable. Thus Hayek ended up distinguishing between the mere “threat of coercion” that is “avoidable,” and “actual and unavoidable coercion.” The former (supposedly avoidable) type of coercion is precisely what Miller calls “hypothetical coercion.” Like Hayek, he concludes that laws that are simply preventative, and so involve supposedly avoidable, hypothetical coercion, do not subject people to coercion.
The problem with this Hayekian argument is that it rests on a conceptual error. It conflates (a) the coerciveness of the threat itself (communicative coercion) with (b) the direct deployment of force that the state may use in carrying out its threat (noncommunicative coercion). Having initially refused to call the direct deployment of force “coercion,” Hayek was subsequently forced to call it precisely that, to argue for his avoidability thesis: the “threat of coercion” simply means the threat of physical force (and not the “threat of a threat.”) Miller’s argument suffers from the same equivocation: after having, like Hayek, restricted his discussion of coercion to coercive threats, he then argues that “hypothetical coercion” does not subject a person to coercion because, after all, one can avoid the direct deployment of force. (He says that “prevention can sometimes take the form of threatening coercion,” and glosses the coercion as the police arriving and “remov[ing]” my neighbour “from my premises.”)
But the whole point of a coercive threat is that it subjects a person to communicative coercion even if, having succumbed to it, she avoids the threatened deployment of force. All conditional threats are “hypothetical.”  To avoid the (direct, noncommunicative) coercion involved in carrying out the threat is not to avoid the (communicative) coercion involved in being threatened. Indeed, this was the point of Miller’s own 1989 critique of Hayek, which Miller tellingly illustrated with a preventative law: as he rightly put it then, the Hayekian argument
. . . confuses [communicative] coercion with the imposition of sanctions for noncompliance. A rule of law, considered in its coercive aspect, typically takes the form: ‘Don’t do X; if you do, Y will happen to you.’ . . . If I know about the rule, I can of course avoid having the sanction applied to me . . . But this does not mean that I have not been coerced: I have been coerced, for which the most straightforward evidence is that my behaviour has been altered by the threat.
Preventative laws are coercive laws.
Miller might respond that it is my own analysis of coercive threats—and in particular how I link coercion to autonomy—that commits me to the Hayekian thesis that prevention-threats are not coercive. After all, our linguistic intuitions about coercion’s extension are ambivalent and murky; as Miller rightly points out, how we define the concept for philosophical purposes partly depends on what normative work we expect it to do. And as I use the term, the normative significance of state coercion is that it (presumptively) invades autonomy and hence triggers a demand for democratic justification to the very people subject to it. So even if Miller were to concede that legal prevention-threats are coercive in some sense, he would still argue that only threats designed to compel fairly specific actions, or prevention-threats that fail to leave intact an adequate range of valuable options, are coercive in the autonomy-invading sense requiring democratic justification. His argument implies, in other words, that not just sections 25 and 235, but a vast array of laws that impose negative rather than positive legal obligations, are exempt from the demand of democratic justification. This strikes me as a straightforward abandonment of democratic theory, so it is important to see precisely where the argument has gone astray.
Miller and I are agreed that prevention-threats invade autonomy if they fail to leave intact an adequate range of valuable options (Raz’s second condition of autonomy). To be more precise: a conditional threat of harm geared to preventing the coercee from doing what the coercer proscribes (i.e., a  threat meeting the four necessary conditions 1-4 specified on p. 58 of my original text) invades autonomy (and so is coercive in my autonomy-invading sense) if it thereby leaves the coercee with an inadequate range of valuable options (i.e., if it meets condition S2 specified in footnote 56). Our disagreement lies in my claim that a prevention-threat (meeting conditions 1-4) also (presumptively) invades autonomy (and so is coercive) if it threatens the deployment of physical force against the coercee (condition S specified on page 59). In other words, some prevention-threats that leave intact an adequate range of valuable options nonetheless invade autonomy, because they invade independence (Raz’s third condition). Miller denies this (by effectively collapsing Raz’s third condition into the second). The example he provides to motivate the denial is that of Jane, who threatens Peter that she will not accompany him if he goes to a particular Thai restaurant (but will if he goes to any other restaurant). As Miller rightly observes, Jane’s threat neither invades Peter’s autonomy nor subjects him to coercion: one’s independence is not compromised simply “whenever a course of action” is “ruled out.” But the trouble with Peter-and-Jane examples like this is that they tend to abstract away completely from why we care specifically about state coercion, which involves, for example, the threat of overwhelming physical force. Miller’s example fails to speak to our disagreement, since it meets neither condition S2 (inadequate options) nor S (threat of force). To help adjudicate our disagreement, we need examples that meet condition S but not S2. So imagine instead that Jane credibly threatens to kill or incarcerate Peter afterwards, should he go to the Thai restaurant. On Miller’s (counter-intuitive) account, this would not be a coercive, autonomy-invading threat; on my account, it most certainly would be.
Credible threats of physical force compromise the coercee’s independence because they threaten to use her body for purposes that are not her own. The point is illustrated by another example meeting condition S but not S2: the happy servant, whose enlightened master has left her an adequate range of valuable options, whose protection she does not wish to leave, but who is nonetheless under threat of corporal punishment should she try. The threat subjects her to coercion and invades her independence.
Miller’s own example meeting S, but not S2, is that of a Scottish landowner who threatens to have shot anyone trying to set foot on his land. What distinguishes this from my killer-Jane example is that some of the people threatened by the landowner have no interest in doing what he proscribes. So perhaps Miller’s point is that, contrary to what my account implies, a prevention-threat cannot invade the independence of, and subject to coercion, people who have no actual interest in the proscribed action. But that point would be mistaken,  as the happy servant, who has no current interest in escape, demonstrates. (As Isaiah Berlin famously pointed out, an account according to which one’s freedom is compromised by obstacles to only what one actually prefers perversely implies that one’s freedom could be increased simply by adapting one’s preferences. So too with independence.) Or perhaps Miller’s point, in emphasizing that the landowner’s threat concerns a remote island, is that when the possibility of the threat actually being carried out is so remote—because it is unlikely that one would, in the normal unmanipulated course of events, ever develop an interest in the proscribed action—then independence is not compromised. (Remoteness distinguishes Miller’s example from the happy servant, whom we can certainly imagine at some point developing an interest in escape.) But (developing an interest in) landing on the Scottish island is no more remote a possibility than me (developing an interest in) murdering; and yet, as I have already argued, section 235’s threat against my body invades my independence (presumptively), subjects me to coercion, and requires democratic justification. So too with the landowner’s threat, which is why uttering threats of violence is properly a matter for (democratically justified) legal regulation.
3. Democratic Legitimacy
Miller invokes a third example to resist the conclusion that persons subject to laws authorizing or threatening the deployment of force are owed a democratic justification. He asks us to consider an obnoxious neighbor who persists in trying to enter my house, but is forcibly prevented from doing so by me and, eventually, the police. It is true, of course, that I (or the police) do not at that moment need to “democratically justify” to my neighbour (whatever that would mean) my use of force to bar him entry. My warrant lies in sections 40 and 41 of the Criminal Code, which authorize persons to use “as much force as is necessary to prevent any person from forcibly” entering their “dwelling-house,” and to use force both “to prevent any person from trespassing on” the “real property” in their peaceable possession and “to remove a trespasser therefrom.” But that is not the register on which democratic justification operates: democratic legitimacy does not normally require that individuals “democratically justify” to others their application of laws in particular instances; it requires that the formulation of laws result from democratic processes of justification involving those subject to the laws. If Miller really means to exempt not just sections 40 and 41, but the whole range of property laws that impose negative obligations, from the demand of democratic justification, then my worry is that, instead of repelling my  thesis about what democracy requires for borders, he has simply evacuated the terrain of democratic theory.
I am grateful to my former teacher for the opportunity, thanks to his engaging critique, to clarify my argument, and to Evan Fox-Decent and Noëlle Sorbara for valuable comments on a previous draft.
 For the significance of the “presumptively” qualification, see note 16.
 Miller subjects all three claims to the qualification that, if imposing negative obligations or issuing prevention-threats fails to leave intact an adequate range of valuable options, then they too require democratic legitimation, subject persons to coercion, and invade autonomy.
 I define physical force as physically acting on a person’s body or restricting the physical space in which her body can move, and follow Scott A. Anderson in defining violence as a subset of force geared to injuring, incapacitating, or inflicting pain on a person’s body. “Coercion as Enforcement,” November 3, 2008, http://ssrn.com/abstract=1294669.
 Grant Lamond, “The Coerciveness of Law,” Oxford Journal of Legal Studies 20, no. 1 (2000): 39-62.
 Noncommunicative coercion thus includes the direct deployment of physical force and the authorization of such deployment. It may seem that laws authorizing the use of force are coercive in the communicative sense, perhaps implicitly articulating a standing coercive threat. But the coercion involved here does not intrinsically depend, as threats do, on communication: a secret directive authorizing the use of force against someone by a standing coercive apparatus subjects that person to coercive invigilation just as much as a published law. But a threat, unless successfully communicated, itself exerts no pressure on anyone. Note also that, although here I focus on physical force, I acknowledge that noncommunicative coercion is not restricted to the deployment or authorization of physical force: other preemptive techniques may also qualify as coercive.
 My use of “invigilation” follows Philip Pettit, “Freedom and Probability,” Philosophy & Public Affairs 36, no. 2 (2008): 206-20.
 Friedrich A. Hayek, The Constitution of Liberty (Chicago: University of Chicago Press, 1960), 133-34 (cf. 137-38) and 142.
 Ronald Hamowy, “Freedom and Rule of Law in F.A. Hayek,” Il Politico 36, no. 2 (1971): 349-76; Murray N. Rothbard, The Ethics of Liberty (New York: New York University Press, 1998), chapter 28.
 Hayek, Constitution of Liberty, 142-43.
 Which leads to highly counterintuitive results. As Hamowy points out, it implies that the threat to kill someone if he buys anything from so-and-so, when he has other shopping options, is not coercive (“Freedom and Rule of Law,” p. 358). Miller himself noted Hayek’s conceptual error in earlier works (Market, State and Community [Oxford: Clarendon Press, 1989], chap. 1, and “Introduction,” in Liberty [Oxford: Oxford University Press, 1991], 14).
 Miller, Market, State and Community, 28.
 References are to Arash Abizadeh, “Democratic Theory and Border Coercion,” Political Theory 35, no. 1 (2008): 37-65. Miller is quite right to say that my claim that a threat is coercive only if it invades autonomy is a substantive claim. My original formulation was misleading in this respect: I should have said that conditions 1-4 are necessary for a proposal to count as a coercive threat, and that specifying the further condition S (or S2, or S3, etc.), which jointly with 1-4 is sufficient for the proposal to be coercive, requires showing that 1-4 plus S (etc.) (presumptively) invades autonomy.
 Since condition S is where our disagreement lies, and since this is also where my account of coercive threats departs from the Nozick–Raz account, it is not wholly accurate for Miller’s critical purposes to assimilate my account to theirs. Scott A. Anderson argues that Nozick-type accounts are “coercee-focused” and lose track of the techniques of coercion (such as the deployment of physical force) that largely animate normative concerns about state coercion. “Of Theories of Coercion, Two Axes, and the Importance of the Coercer,” Journal of Moral Philosophy 5, no. 3 (2008): 394-422.
 The independence condition is closely mirrored by the nondomination condition championed by neo-republicans like Philip Pettit, A Theory of Freedom (Oxford: Oxford University Press, 2001).
 Isaiah Berlin, Four Essays on Liberty (Oxford: Oxford University Press, 1969), xxxviii. The second Razian condition can be met solely by adapting preferences: if a person’s values change, preexisting options may thereby become valuable. This means that autonomy is susceptible to adaptive preferences. I believe that this is as it should be, but not so for independence—but there is no room to argue that here.
 I say “presumptively” because I wish to leave the following possibility open for the specific case of a threat (or authorization) that leaves intact an adequate range of valuable options, but is nonetheless coercive by virtue of threatening (or authorizing) physical force. If such a threat consists in a law, then one might argue that, although the law subjects persons to coercion, the presumptive invasion of their independence is overcome if the law has been democratically justified by and to the very persons subject to it. This interpretation of the relation between independence and democratic justification (which mirrors the neo-republican view [see Pettit, Theory of Freedom] about the relation between nondomination and democratically forcing the power-holder to track the subject’s common avowable interests) of course leaves intact the demand for democratic justification. Raz himself equivocates on this: he says that “all coercive interventions invade autonomy,” but that in a liberal state with “guaranteed adequate rights of political participation . . . its coercive interventions do not express an insult to the autonomy of individuals.” Joseph Raz, The Morality of Freedom (Oxford: Clarendon Press, 1986), 156-57.
 I hesitate because I have trouble believing that Miller, a staunch democratic theorist, really wants to say this. But he does say that “the person who is being prevented” from entering my house “does not necessarily have a right to be included in the institution or practice from which the prevention emerges.”