How Not to Justify the State

How Not to Justify the State: Robert Nozick’s Attempt

1. My main purpose here will be a fairly modest one. I will be criticizing a single attempt to justify the state as a morally legitimate institution. In my defense I would point out that this particular attempt is an important and persuasive one: namely, the one that Robert Nozick presents in Part I of Anarchy, State, and Utopia. I will argue that the justification he gives is actually inconsistent with certain a certain principle for which he make a plausible case elsewhere in the book. I will present an alternative principle, one that is also I think plausible, and show that it too is inconsistent with the sort of justification of the state that he attempts. I will conclude with some rather negative observations on the possibility of justifying the state.

I am limiting my subject to this one attempt to justify the state because it is an example of a certain sort, a certain family of justifications of the state: namely ones that are individualist and deontological. It is individualist in that it attempts to show that states can function as states by exercising rights that individuals have independently of state decree. It is deontological in that these rights are conceived as side-constraints. They are not simply goals to be maximized.

I see at least one good reason for taking this sort of justification of the state most seriously. I will very briefly try to explain what that reason is, simply in order to motivate the rest of my discussion. My explanation will be so brief that it is apt merely to annoy those who are not part of the relevant moral and philosophical tradition. I hope not, but there is little I can do to avoid it.

First, consider that the main reason why there is a problem about justifying the state in the first place is, after all, one that arises from thinking about the behavior of the state in individualist and deontological terms. It is this. States typically claim the right to do things that violate deontic norms that, intuitively, apply to individuals. They claim the right to force you to pay for their services, even if you don’t want them (taxation). They claim the right to coercively prevent people from crossing an imaginary line which is not the boundary of their property and in fact has no very close relationship to property rights at all (immigration restrictions). They also sometimes claim to force people, admittedly innocent of any wrongdoing, to do extremely dangerous work, even coercively exposing them to a high probability of violent death (military conscription). These are things that individuals such as you and I may not do. Yet the agents of the state, who allegedly may do them, are themselves individuals. How can this be right?

There are broadly three different strategies for answering this question. (I note in passing that you can also reject the question, but that is to raise issues I will not deal with here.) First, you might deny that these deontic constraints apply to individuals across the board. One family of strategies of this sort is comprised of ones based on utilitarianism. Second, you can deny that the being that these agents represent – the state – is an individual or a collection of individuals, but a profoundly different sort of entity. These strategies would presumably all involve holistic (and so non-individualistic) thinking of one sort or another. These two strategies can be very intimately connected, so much so that it can be difficult to distinguish between them. After all, one thing that is so unique about the holistic entity the state, is precisely the fact that certain deontic norms do not apply to it, that it may do with impunity what you and I may not do. Both these sorts of strategies encounter certain sorts of problems. One of the most obvious is what I suppose you could call the problem of excessive explanatory power. The supposition that deontic norms do not apply to agents of the state, or that their actions cannot be evaluated as the actions of ordinary people can explain why states may do things like collecting taxes, but they rather obviously expose one to the hazard of “explaining” why they may do things that are clearly and seriously wrong. In the case of utilitarianism, this problem has surfaced in the literature on the paradoxical solutions it seems to yield to problems like that of whether to execute one innocent person in order to quell a crime wave for which no real perpetrators can be found.¹ Such doctrines have an often-noticed tendency to justify the unjustifiable. Rather obviously, we face the same sort of hazard if we suppose that the state is a superpersonal entity that grants its agents corresponding moral superpowers. This fact tends to undermine one’s confidence in the ability of these strategies to legitimately justify the things that, if taken uncritically, they would seem to justify.

The source of this problem is of course that one has dropped one of the two assumptions that I have identified as creating the problem of justifying the state in the first place. If one keeps them, one is supposing that in order to justify the state, one is justifying the actions of individual human beings, and that one needs to show that the things that they do – at least, enough of those things to constitute them as a state – are consistent with moral constraints that apply to individual human beings. In doing so, one reduces the hazard of excessive explanatory power. Of course, for the same reason, one increases the hazard of insufficient explanatory power – insufficient for the purpose of justifying the state. The constraints built into the method make its results, if there are any, more impressive, but they also decrease the likelihood that there will be any. Exploring this tradeoff, between impressiveness and chances of success, is my ultimate purpose in this paper. I will offer some reasons for thinking that the individualistic and deontological approach in fact cannot justify the state. As should be obvious, from what I have said already, it does not follow that one of the others approaches is better. As I have suggested, they have problems of their one, ones that may be worse.

2. What sorts of state activities need to be justified, in order to justify the state itself? Obviously, they would have to be activities that are essential to something’s being a state. It is not clear that any of the three state activities I have cited so far are essential to a state’s being a state. An organization can be a state though it does not stop people from wandering across its boundaries. Nor need it force people to fight for it. Taxation, one might think, is essential, simply because it seems to be something that actual states always actually do. But it is not clear that an organization that did not collect taxes would thereby fail to qualify as a state. States need to collect fees to pay their expenses, but these are not necessarily taxes. They will not be taxes if, like conventional prices on goods and services, they are only collected from those who choose to accept those services. If one has the choice of refusing the state’s protection, if one may opt out of the system and thereby avoid paying for protection, then the payments, when made, are not taxes. In Nozick’s system, one does have such a choice. Payments to his system are accordingly not taxes, and yet he thinks that the system is a state.² It is at any rate not clear that we need to justify taxation in order to justify the state.

There is another characteristic of the state that need not be justified in attempts to justify the state, though for a different reason. States are coercive systems. Their agents are often armed and they both threaten and on occasion inflict violence on people. This feature of the state need not be justified in this context for the simple reason that, so far, its agents might merely be exercising rights that everyone has. We have a right of self-defense, which at times requires one to both threaten and inflict violence. And if we have a right of self-defense, we must also have a right of other-defense, of defending people other than ourselves from attack. Of course, states who punish those who violate the rights of others seem to be doing something rather different from defense in the sense of preventing or deflecting an attack. But both Nozick and his predecessor John Locke assume that, if there were no civil authorities to do so, we would have the right to punish people after the fact who violate the rights of others. If we follow them in making this assumption, that everyone would in the absence of the state have a right to punish, then we need not justify the fact that it does it. If we suppose that these rights are transferrable from principal to agent, we will have no trouble imagining somewhat state-like entities emerging in a state of nature. People would have a right to form protective associations, with a certain division of labor within the association. Some members could rightfully exercise the right to punish violators of rights on behalf of the others. Indeed, people in such a situation probably ought to make such arrangements, if only they have the resources to do so. If they do, there will then be organizations with some obviously state-like functions. We can even imagine them formulating their rules of operation, both substantive and procedural, and publicizing them in some way or other.

But I am speaking of these protective associations in the plural, and this of course contrasts with the nature of the state. It does not merely carry out its typical functions but in some way or other monopolizes some of them. As Max Weber famously said, “the state is a human community that claims the monopoly of the legitimate use of physical force in a given geographical area.”³ It claims that it may punish and we may not. Can this feature of the state be justified as an exercise of rights that individuals have, or would have if there were no civil authorities? It is to this question that I will now turn. As I have said, I will focus on Nozick’s attempted solution to this problem.

3. Nozick’s attempted solution has three parts. The first part consists in expanding the definition of the state. In effect, he relaxes the requirement for what is to count as a state. He does not claim that the organization that he justifies – the Dominant Protective Association, or DPA – has the right to prohibit all others from punishing and exacting compensation for violations of rights. Rather, what it does is to decide which of the other protective agencies will be permitted to operate within its territory. It examines the rights-enforcing methods of other agencies to see whether they are sound ones or not. In particular, it passes judgment on the decision-making procedures by which they settle questions of justice, deciding whether they fall within the range of permissible procedures. He believes that possessing and exercising this right is sufficient to make this entity a state. I suppose that the intuitive idea here must be that, though this is not exactly a monopoly of force, it is a monopoly of a certain force-related right: this particular protective association is the only agent to decide who will be allowed to use force and who will not. He thinks this is sufficient to make it this organization a state but, to acknowledge the fact that this claim relaxes the traditional Weberian definition of the state, he sometimes refers to the organization he has justified as “a statelike entity.”⁴

On his view, this organization is unique only in that it is in a physically dominant position that makes it, and no other agency in the area, able to decide who is a permissible wielder of force and who is not. Yet though it alone is in a position to exercise this right, everyone possesses it. Exactly what is this right, that everyone possesses? This brings us to the other two “parts,” as I have just called them, of his justification of the state. He gives two different answers to this question. In Chapter 5, he says that the dominant protective association may punish anyone who applies to one of its own clients a procedure that it deems unreliable or unfair.⁵ This is because we all have, in addition to various substantive rights against having various wrongful things done to us, a procedural right to the effect that decisions about what will be done to us be made in the right way. Because of the peculiar role this right plays in his justification of the state, it clearly must be a natural right. In addition, according to a line of reasoning that centers mainly in Chapter 4, Nozick holds that the DPA may prohibit agents (apparently, either individuals or organizations) from applying procedures to its own members that expose them to excessive levels of risk that they will be punished even though they are innocent.⁶ This is the third part of his justification of the state. Of these two lines of reasoning – what might be called the argument from risk the argument from procedural rights – I will briefly comment on the last one, and then I will discuss the first at length.

4. Actually, there is another thread in Nozick’s justification of the state that I should also probably say something about before moving to these two arguments. There is an extended discussion in Chapter 4 of the fact that people who use risky methods thereby inspire fear in others. Some of Nozick’s subsequent references to this discussion (eg., on p. 89) might give the impression that it constitutes an argument from fear, so to speak, that plays just the sort of role in his justification of the state as the arguments from risk and procedural rights. I do not think that this is how he means the discussion of fear to be taken. The other two arguments are meant to show that the unreliable or unfair methods used by some agents are, in and of themselves, what Nozick calls “boundary crossings.” In other words, they are violations of rights. Rather the argument from fear is meant to show that, on the assumption that these things are boundary crossings, the proper response to them is not to permit them provided that their victims are compensated, but to simply prohibit them.

The idea is that it is impossible to compensate people for the fear that they would experience as a result of being exposed to the possibility of undeserved harm. He asks us to imagine the following two scenarios.⁷ X hears that Y, having had an accident and broken his arm, was later compensated with $2,000. X thinks that this is adequate compensation: it covers the injury. Then Z tells X that, some time in the next month, Z will break X’s arm and then pay him $2,000. X spends the whole month a nervous wreck. He is not indifferent to the combination of broken-arm-plus-$2,000. In some way, the $2,000 fails to cover the injury. Is he being inconsistent? Of course not. There is something in the second scenario that was not in the first: the fear engendered by knowing of the injury in advance. Fear, Nozick points out, is not a “global” emotion: it fastens on parts of packages rather than basing itself on on-balance assessments.⁸

There is a possible response to this, which Nozick puts in the form of a questions: “Why couldn’t someone who commits an assault compensate his victim not merely for the assault and its effects, but also for the fear the victim felt in awaiting some assault or other.” He replies:

This might be plainly untrue, depending on how one interprets it. How one should interpret it is not obvious. For “procedure of justice” cannot mean a formal, legal procedure because the sort of procedure that we are dealing with here may be very informal indeed. Perhaps it could mean something like “any procedure that results in behavior that can be substantively just or unjust.” Read in this way, the principle is excessively broad and onerous, and would create a vast array of hitherto unknown rights and obligations. Depending on one’s conception of justice, many – perhaps all – of the things that anyone does that have effects on others can be just or unjust. Yet surely I do not generally or typically have an enforceable right to information about how others make decisions that effect me. In many cases, this would mean having rights regarding the thought-processes of others, a situation that would clash with very basic intuitions we have about privacy.

Probably, Nozick means to limit “procedure of justice” to procedures in which someone is deciding whether to punish someone or exact reparations for some alleged violation of someone’s rights. This interpretation makes the principle much more plausible. But notice that what it says, on this interpretation, is that everyone may resist, by force presumably, the application of the decision procedure itself, before it has resulted in any substantive injustice. Why would this be true, if it is? It seems that the only possible answer, either on this interpretation of the principle or on the other, less plausible one, would have to be that an unreliable procedure would be more likely to result in a violation of rights than a more reliable one, that it (in other words) exposes one to higher levels of risk. That would mean that the procedural rights argument, insofar as it is plausible, collapses into the risk argument.

6. What then, of the other argument, the focus of Chapter 4 of Nozick’s book? Here the story is I think much more complex. It is also of much more general interest. Nozick says that actions “that risk crossing another’s boundaries pose serious problems for a natural rights position.” I would add that the reason he gives for this claim actually makes risk a problem for any moral position, because the reason is not moral but logical: “Imposing how slight a probability of a harm that violates someone’s rights also violates his rights?”¹¹ This might be called the line-drawing problem. Risk is the probability that some harm (in this case, a rights violation) will occur. But probability is a continuum, containing infinitely many degrees, ranging from 0 to 1. Which degree of probability that a rights violation will occur, constitutes a rights violation in itself? Surely, we cannot say that all degrees above 0 are violations. In that case, on what basis can we draw a line between levels that are acceptable and those that are not? This is certainly an interesting problem, and it is a problem that should be solved no matter what your moral theory might be. What should our solution be? Nozick sensibly says that perhaps “the cutoff probability” should be lower for more serious harms, but that does not answer this question.¹²

He does not try to solve this problem, at least not in Chapter 4, but he does make an interesting comment on the nature of the possible solutions in a footnote. There, he points out that two “types of theories could be developed.” First, a “theory could specify where a line is to be drawn without this position’s seeming arbitrary, because though the line comes at a place which is not special along the probability dimension, it is distinguished along the different dimensions considered by the theory.” That would be one possible type of theory. Alternatively, he says, “a theory could provide criteria for deciding about the risky actions that do not involve drawing a line along the probability (or expected value or some similar) dimension.” He adds, provocatively: “Unfortunately, no satisfactory specific alternative theory of either type has yet been produced.”¹³

It seems there must be some reason to hope that this problem can be solved if we are to place any confidence in the risk argument. In addition, I suggest that there is a second problem that stands in the way. I will call it the actual-potential problem. In the Lockean natural rights tradition that Nozick represents, coercion is justified by the fact that it is a response to a boundary crossing. A boundary crossing is something one actually does. Risk is the likelihood that certain things might actually happen. But this seems to be a fundamentally different sort of thing from a boundary crossing. This problem can be put in terms that would seem to apply to any moral theory. Intuitively, if we are going to punish you it should be for something you have done, and the same goes for exacting reparation. Punishing you for what you might do is not right. But if I am exposing you to a risk of a broken arm, that only means that I might break your arm. Precisely because it is something I might do, it is not something I have done. But why doesn’t that mean that I can’t be punished, or otherwise coerced, on account of it?

As with the line-drawing problem, there seem to be two sorts of possible solutions. We might give some reason why things that have not been done can have the same moral significance as if the agent in question has done them. Or we may find something that the risk-creator is actually doing (where creating risk is not arbitrarily described as “actually doing something”), something that can serve as a basis for justifying the use of force.

It seems to me that the argument from risk requires some hope of a solution to both of these problems, the line-drawing problem and the actual-potential problem. In the next section I will argue that, surprisingly enough, Nozick lays out a doctrine in a later chapter which fairly directly implies a solution to both of these problems – including the one that he has pronounced to be so far unsolved. But I will also argue that it solves them in a way that actually prevents the risk argument from serving the function of justifying the state.

7. I have said that one reason the risk argument is interesting is that it raises an issue with a much broader application than playing a role in an attempt to justify the state: namely, the ethics of coercive responses to risky behavior. The issues it raise might be potentially even wider than that. This even wider issue might be called the ethics of basing coercion on future events. It includes not merely the ethics of responses to risky behavior but, in addition, the ethics of preemptive attack, including of course preemptive warfare. In a way, forcibly stopping a risky activity is a sort of preemptive attack. If all I am doing is to drive down University Avenue in Madison Wisconsin at 100 miles per hour while roaring drunk, I have not hurt anybody – yet. The police have a right to lead me away in handcuffs – and anyone in a Lockean state of nature would have the same right – because we don’t have to wait until I have hurt someone. Similarly, a country does not necessarily have to wait until its enemy is actually, presently dropping bombs on it before it uses force in self-defense against it.

Nozick discusses the issue of preemptive attack in Chapter 6, and because of the strong connection between the two sorts of issues, what he says there can be applied, with interesting results, to what he says about risk in the risk argument.

The connection between the two sorts of issue is actually more specific than I have so far said. Consider the way Nozick formulates the preemptive attack problem. The “usual doctrine,” he says, holds that “under some circumstances a country X may launch a preemptive attack, or a preventive war, upon another country Y.” For instance, X may attack Y if “Y is itself about to launch an immediate attack upon X,” or, again, if “Y has announced that it will do so upon reaching a certain level of military readiness, which it expects to do some time soon.” On the other hand, X may not “launch a war against Y because Y is getting stronger, and (such is the behavior of nations) might well attack X when it gets stronger still.” Why does self-defense plausibly justify force in the first cases but not in the last?¹⁴

Notice that it is possible to formulate a core issue involving risk in exactly the same way. Under certain circumstances, you have a right to preemptively stop my bad driving practices before they have hurt anyone. You may stop me if I am driving 100 miles per hour on a city street while drunk. You may also stop me if I am driving drunk but not speeding. But on the other hand, you may not forcibly stop me from drinking in a bar earlier in the evening, on the grounds that I might go out and injure someone because of my diminished mental capacities. Self-defense, or something very much like it, justifies force in the first cases, though not in the last. Why?

With the obvious parallels between these two sorts of issue, it seems natural to try to formulate an answer to the “why” question for one sort of issue and then try to apply that answer, perhaps suitably adapted, to the other. Given that the common structure I have attributed to the two sorts of issue looks very much like the original line-drawing problem, an obvious strategy would be to try to assimilate the preemptive attack problem to the risk problem. If the difference between the two preemptive attack cases that justify force and the one that does not is the degree of probability that Y will attack X – the probability that Y will attack X given that it is getting stronger falls below some threshold – then the solution to the question about risk might well be the key to the answer to the question about preemptive attack.

Nozick gives a simple and, I think, telling argument against finding the difference between the preemptive attack cases in these sorts of probability considerations. Take the probability that Y, having grown stronger, will attack X. Now imagine instead the following scenario: “Y is now about to wield a super-device ... that, with that degree of probability, will conquer X.” Y is committed to trigger the device at the end of one week, and the countdown has begun. Clearly, X may use force in self-defense to prevent this from being done. But if probability were the crucial consideration, this case would have to be treated the same as the third preemptive attack case.¹⁵ So it looks like the preemptive attack problem cannot be assimilated to the risk problem.

This is actually good news, because the risk problem does not look like the sort of problem that can be solved on its own. After all, both of Nozick’s suggestions for possible sorts of solutions to the line-drawing problem involved basing the line on non-probablistic considerations. Either we can find a level of probability that is special because it is linked to something other than probability, or we define unacceptable risky behavior in terms of something other than probability altogether. Interestingly, when Nozick proposes an answer to the question about preemptive attack, it is a version of the second sort of possible solution to the line-drawing problem about risk. The difference, he says, between the third preemptive attack case and the other two is not about the probability of future boundary-crossings at all. It is about the relation between those possible future events and what the agent has already done. The principle that, according to him, distinguishes between them can be adequately paraphrased as follows:

This is both clearer and more focused on the issues at hand than the original statement of the principle.

The fact that this principle applies directly to the problem of prohibiting risky activities, and is as plausible as it is when so applied, means that it so far enhances the risk argument for the legitimacy of the state. It will, if we accept it, solve the problem of how we can draw a non-arbitrary line between risks that may be prohibited and ones that may not. However, it solves this problem only by creating a more serious one. After all, if the principle does give us reason to think that a non-arbitrary line can be drawn, it does so by drawing that line, and it is important for the argument from risk that the risky decision-procedures of competing protective associations fall on the “wrong” side of this line.

I think they do not, and for a fairly simple reason. Suppose that some such agency is using an unreliable procedure to decide whether to punish me for something I am supposed to have done. Suppose that, because it is unreliable, the decision they would come up with, were they to execute it, would be unjust. Their use of this procedure is harmless unless they make a further decision to do wrong: namely, to execute their unjust decision. This is true of any unreliable decision-procedure. The unjust punishment, and the decision process that selects it, are distinct and separable. To say that it may be forcibly resisted in self defense is analogous to saying that country X may preemptively attack country Y at a certain point upstream from the moment when Y’s attack on X becomes imminent, and even before Y has committed itself to attacking X: namely, when Y is using unreliable rules of evidence in the cabinet meeting in which it is deciding whether to do these things. Obviously, this is not wrong in a way that gives others a right to violently attack them.

There is an interesting feature of Nozick’s discussion of preemptive attack that, in an odd way, actually reenforces the point that I am making here. He actually does notice that what he is saying there is logically relevant to the risk argument, but the only connection he notes is positive, that it strengthens one aspect of the earlier argument. This was his attempt to resist the possibility that the risk argument “proves too much,” that it would not only justify the dominant protective agency in passing judgment on the methods used by other agencies and prohibiting those who use procedures that are unreliable or unfair, but would also do something he says it does not do: namely, justify it in prohibiting people from forming or joining other protective associations at all. He says that his treatment of the preemptive attack principle shows that his “argument provides no rationale for [such] actions and cannot be used to defend them.”¹⁷ His point seems to be that merely joining another protective association is an act that is not itself a boundary crossing considered in itself and is harmless in the absence of a subsequent decision on their part to do wrong. My point is that the same is true of agencies that do use unreliable decision procedures, insofar as they are only using the procedures and not coming to unjust decisions or acting on them.

9. Is there an alternative principle, one that also avoids or solves the line drawing and actual-potential problems? I can think of one, but it undermines Nozick’s risk argument in the same way that his preemptive attack principle does.

There is a reason why someone who is partial to Nozick’s position would want to investigate the possibility of an alternative, even aside from the fact that, if I am right, his own principle conflicts with his justification of the state. His principle could be used to defend censorship of pornography. Suppose, for purposes of illustration, that a familiar feminist view of pornography is correct. Pornographic works have a certain effect on the attitudes, values, and institutions that are part of our culture. This effect, generally, is to undermine respect for the rights of women (this can be made part of the definition of “pornography” without affecting my point). This has the further effect of changing the way men in general treat women in general, a change that is very much for the worse. If there is pornography being produced and consumed, there will be more violent attacks on women than there would otherwise be. Now take two young women. A is an actor in a pornographic video. B is a rape and murder victim. B’s attacker was influenced by the culture to which A is contributing. If it were not for the activities of A and others like her, B’s the influences on the attacker’s behavior would not have been quite vicious enough to result in his attack. As it was, they were. Nozick’s principle would seem to imply that women, or agents acting for them, would have the right to coercively interfere with A’s activities. After all, according to the causal theory we are assuming, what A does is not harmless, and will do its harm in the absence of any subsequent decisions to do wrong on her part. She has made her contribution to the bad effect.

Someone might say at this point, “Why is this such a bad result for someone who is sympathetic to Nozick’s position? Wouldn’t this feminist theory, if true, just show that the true Nozickean position, contrary to what we had thought, would favor coercive measures against pornographers, just as it should favor coercive measures against people who dump toxic waste into the water supply?” My reply would be that it only implies that in the presence of Nozick’s preemptive attack principle. There is an alternative principle that would not require this revision. The alternative rests rather heavily on two concepts. One is the concept of an act. I will have to assume that we are very good, without any fancy theories to help us and with only the relevant facts, at determining whether an event that occurs is some particular agent’s doing. The other is the idea of a risk’s eventuating in harm. Every risk is a certain probability of a particular (type of) harm occurring. If that (type of) harm actually does occur, and occurs because of the those same risky conditions, then the risk has eventuated in harm. Now my alternative principle would be this:

(The last qualification is necessary because not all harms are violations of rights.) One way to grasp the intuitive idea behind this principle would be to think of it as an attempt to solve the actual-potential problem in the context of a natural rights position. Why would you have a right to stop me from speeding down a city street while drunk when I have not actually violated anyone’s rights? One answer to this question would be this. If my act results in someone else’s death, the harmful event would in that case be being killed. I will have killed them. And the act by which I will have done so would be – just what I am doing now, speeding down the street while intoxicated. The intuitive idea is this: You do not have to wait until this lethal act of mine, which is actually in progress, is competed before you take coercive action against me. The same would be true if I am driving down the street drunk but not speeding. My present act can become lethal, and you needn’t wait until it does. The theory is here is closely related to the one that enables us to punish someone for murder though they only gave orders, or payed someone, to kill, and did not pull the trigger themselves. What they did was to perform an early portion of the act, where the entire act – the killing – is one that they are carrying out in concert with others.

On the other hand, this principle does not include, in the category of acts liable to resistance by force, the case of the mere drinker. If all I am doing is quietly stunning my brain with alcohol, then I may well be enhancing the probability that I will harm someone later on, but if I do so, I will not be harming them by quietly drinking in a bar. The same is true – and here the alternative principle departs from Nozick’s – of the actor in the pornographic video. In that case, the harmful event was not the (by hypothesis) risky act that the actor performed. She did not rape or kill B. Although it is obviously true that killing someone involves causing or contributing to causing someone’s death, this principle, you might say, is based on the idea that not every sort of contribution counts.

Again, as I have suggested already, the same thing is true of the protective agency or individual who uses unreliable or unfair decision procedures. If those procedures result in my being punished unjustly, the harmful event is the punishing, which begins with the formation of the intention to inflict it, and not with the decision-procedures that lead up to it.

10. Nozick’s attempt to justify the a state-like institution raises an interesting and very real issue of much wider application: why may we sometimes coerce someone on account of an infraction that they have not yet committed? He is surely right in contending that we may sometimes do so. We have looked at two plausible principles that would, if accepted, authorize us to do such things. But – though one of them comes from Nozick’s own text – neither one reaches far enough to authorize the DPA to abolish competitors simply for using excessively risky decision procedures. His strongest argument seems to have failed.

1. Perhaps the locus classicus here is an argument to this effect by Carritt, quoted in Rawls’ “Two Concepts of Rules.” John Rawls, Collected Papers, ed. by Samuel Freeman (Cambridge, Massachusetts: Harvard University Press, 1999, orig. pub. 1955), p. 26.

2. As I will discuss below, one has the option of contracting with another protective agency (ie., one not prohibited by the dominant one). Robert Nozick Anarchy, State, and Utopia [hereafter ASU] (New York: Basic Books, 1974), Ch. 4 and 5, passim.

3. Max Weber, “Politics as a Vocation,” in From Max Weber: Essays in Sociology, trans. and ed. by H. H. Gerth and C. Wright Mills (New York: Oxford University Press, 1946), p. 78

4. ASU, pp. 117-8.

5. ASU, p. 101.

6. ASU, p. 88. Although, as I have put it, this argument “centers on” Chapter 4, it is stated most clearly here, on the first page of Chapter 5.

7. ASU, p. 66.

8. ASU, p. 70. I would argue that all emotions are non-global in this sense.

12. Ibid.

15. Ibid.

16. ASU, p. 127. I have altered Nozick’s wording in several ways to eliminate features that would distract from the present discussion, but not, I think, in ways that significantly alter his meaning. I will discuss one such change in the text below. There are several others. What he actually says on p. 127 this: “Perhaps the principle is something like this: an act is not wrong and so cannot be prohibited if it is harmless without further major decisions to commit wrong....” I have changed “cannot be prohibited” to “cannot be resisted with force” to make it more immediately clear how it is related to our discussion, which is about using force. It is clear from Nozick’s discussion that the action that is ruled out by this principle is the use of force, that this is what “prohibiting” means here. Also, the principle as he states it here is ambiguous as to whether the further decision to do wrong referred to in the principle is a decision on the part of the agent who may (not) be resisted with force, or whether it is decision on the part of anybody. I have disambiguated the principle in line with an explicit statement he makes on page 129: “This principle does not claim that no one may be held responsible or be punished for attempting to get others to do wrong because to succeed the attempt requires the decision of others to do wrong. For the principle focuses on whether the thrust toward wrong already has been made and is now out of that person’s hands.” Thus, what he means does seem to be what I have him say in the text, above.

17. ASU, p. 129.