Many libertarians appear to believe that libertarianism can be stated as a simple and convincing moral principle from which everything else follows. Popular candidates are "It is always wrong to initiate coercion" and "Everyone has the absolute right to control his own property, provided that he does not use it to violate the corresponding rights of others." If they are right, then the obvious way to defend libertarian proposals is by showing that they follow from the initial principle. One might even argue that to defend libertarian proposals on the grounds that they have desirable consequences, as I have done throughout this book, is not only a waste of time but a dangerous waste of time, since it suggests that one must abandon the libertarian position if it turns out that some coercive alternative works better.

One problem with deducing libertarian conclusions from simple libertarian principles is that simple statements of libertarian principles are not all that compelling. Lots of people are in favor of initiating coercion, or at least doing things that libertarians regard as initiating coercion. Despite occasional claims to the contrary, libertarians have not yet produced any proof that our moral position is correct.

A second problem is that simple statements of libertarian principle taken literally can be used to prove conclusions that nobody, libertarian or otherwise, is willing to accept. If the principle is softened enough to avoid such conclusions, its implications become far less clear. It is only by being careful to restrict the application of our principles to easy cases that we can make them seem at the same time simple and true.

The easiest way to demonstrate this point is with a few examples. In order to define coercion, we need a concept of property, as I pointed out at the beginning of this book--some way of saying what is mine and what is yours. The usual libertarian solution includes property rights in land. I have the absolute right to do what I want on my land, provided that I refrain from interfering with your similar right on your land.

But what counts as interfering? If I fire a thousand megawatt laser beam at your front door I am surely violating your property rights, just as much as if I used a machine gun. But what if I reduce the intensity of the beam--say to the brightness of a flashlight? If you have an absolute right to control your land, then the intensity of the laser beam should not matter. Nobody has a right to use your property without your permission, so it is up to you to decide whether you will or will not put up with any particular invasion.

So far many will find the argument convincing. The next step is to observe that whenever I turn on a light in my house, or even strike a match, the result is to violate the property rights of my neighbors. Anyone who can see the light from his own property, whether with the naked eye or a powerful telescope, demonstrates by doing so that at least some of the photons I produced have trespassed onto his property. If everyone has an absolute right to the protection of his own property then anyone within line of sight of me can enjoin me from doing anything at all which produces light. Under those circumstances, my "ownership" of my property is not worth very much.

A similar problem arises with pollution. Libertarians sometimes claim that since polluting the air over anyone else's property is a violation of his property rights, pollution can be forbidden in a libertarian society except when the pollutor has the consent of the owners of all affected land. This argument is used to attack schemes such as effluent fees (discussed in Chapter 26), which are designed to limit pollution to its economically efficient level--the point at which further reductions cost more than they are worth--but not to eliminate it.

Here again, the problem is that an absolute right to control one's property proves too much. Carbon dioxide is a pollutant. It is also an end product of human metabolism. If I have no right to impose a single molecule of pollution on anyone else's property, then I must get the permission of all my neighbors to breath. Unless I promise not to exhale.

The obvious response is that only significant violations of my property rights count. But who decides what is significant? If I have an absolute property right, then I am the one who decides what violations of my property matter. If someone is allowed to violate my property with impunity as long as he does no significant damage, we are back to judging legal rules by their consequences.

A similar problem arises if we consider effects that are small not in size but in probability. Suppose I decide to play Russian roulette, with one small innovation; after putting one cartridge in my revolver and spinning the cylinder, I point it at your head instead of at mine before pulling the trigger. Most people, libertarian or otherwise, would agree that you have every right to knock the gun out of my hand before I pull the trigger. If doing something to someone (in this case shooting him) is coercive, then so is an action that has some probability of doing that something to him.

But what if the revolver has not six chambers but a thousand or a million? The right not to be coerced, stated as an absolute moral principle, should still apply. If libertarianism simply consists of working out the implications of that right, then it seems to imply that I may never do anything which results in some probability of injuring another person without his consent.

I take off from an airport in a private plane with a cruising radius of a thousand miles. There is some (small) probability that my instruments will fail, or I will fall asleep, or for some other reason I will go wildly off course. There is some probability that the plane, having gone off course, will crash. There are things I can do which will reduce these probabilities, but not to zero. It follows that by taking off I impose some (small) probability of death and destruction on everyone through whose roof I might crash. It seems to follow from libertarian principles that before taking off I must get permission from everyone living within a thousand miles of my starting point.

I am not claiming that libertarians who argue from rights rather than from consequences believe that you cannot light a match on your own property, or fly an airplane, or breathe out; obviously they do not. My point is that simple statements of libertarian rights taken literally lead to problems of this sort.

One can avoid such results by qualifying the statements: saying that they apply only to "significant" violations of my rights, or violations that "really injure" me, or that by breathing and turning on lights and doing other things that impose tiny costs on others I am implicitly giving them permission to do the same to me. But once one starts playing this game one can no longer use rights arguments to draw clear conclusions about what should or should not happen. People who believe in taxes can argue just as plausibly that taxes do not really injure you, since the benefits they produce more than make up for the cost, or that everyone implicitly consents to taxes by using government services.

The longer I have thought about these issues, the more convinced I have become that arguments about fundamental moral principles do not provide answers to enough important questions. In particular, they provide no answer, and no way of getting an answer, to a whole range of questions about where to draw lines. It seems obvious that we want property rules that prohibit trespass by thousand megawatt laser beams and machine-gun bullets but not by flashlights and individual carbon dioxide molecules. But how, in principle, do you decide where along that continuum the rights of the property owner stop? We want rules that prohibit me from demonstrating my marksmanship by shooting a rifle at flies hovering around your head but do not prohibit all airplane flights. We want rules that prohibit trespass by elephants but not by satellites orbiting three thousand miles over my roof.

One tempting approach to such issues is to try to go back to the origin of property in land. If we knew how I acquired ownership of land, we might also know what that ownership consists of. Unfortunately, we do not know how I acquired ownership to land. John Locke, several centuries ago, suggested that we acquire land by mixing our labor with it, but he did not explain how, when I clear a piece of forest, I acquire not only the increased value due to my efforts but complete ownership over the land. How, in particular, do I acquire the right to forbid you from walking across the land--something you could have done even if I had never cleared it? Later libertarian theorists have suggested other grounds for establishing ownership in land, such as claiming it or marking its boundaries. But no one, so far as I know, has presented any convincing reason why, if land starts out belonging equally to everyone, I somehow lose my right to walk on it as a result of your loudly announcing that it is yours.

It is easy enough to show reasons why the conversion of common property into private property is a good thing--why it makes us better off--but it is very much harder to derive property in land from some a priori theory of natural rights. That is why, at the beginning of this book, I conceded that the basis of property in unproduced resources such as land is shaky, and argued that it does not matter very much, since only a small fraction of the income of a modern society is derived from such resources.

The problems I have discussed so far are all associated with the definition of property rights to land. A host of similar problems arise in specifying the rules of a legal system designed to enforce libertarian rights in a libertarian way. A criminal trial rarely if ever produces a certainty of guilt. If you jail (or fine) someone after concluding that there is a ninety-eight percent chance that he has committed a crime, there remains a two percent chance that you are violating the rights of someone who is innocent. Does that mean that you can never punish anyone unless you are a hundred percent certain he is guilty? If not, how in principle do libertarian moral principles tell you what degree of proof should be necessary for conviction and punishment?

Once someone is convicted, the next question is what you can legitimately do to him. Suppose I have stolen a hundred dollars from you. If all you are allowed to do is take your money back, then theft is an attractive profession. Sometimes I am caught and give the money back, sometimes I am not caught and keep it. Heads I win, tails I break even.

In order to prevent theft, you must be able to take back more than was stolen. But how much more? When I raised that question once in a talk to a libertarian audience, I was told that it had already been answered by a prominent libertarian--you are entitled to take back exactly twice what is stolen. That was many years ago, but nobody yet has given me a reason why it should be twice. Two is a nice number, but so is three, and there may be much to be said for four, or ten, or a hundred. The problem is not to invent answers but to find some way of deriving them.

I could continue with a wide range of other problems for which the natural rights approach to libertarianism offers, so far as I can tell, no solution. I would prefer instead to suggest a different criticism of that approach. Even if we ignore situations that involve vanishingly small rights violations, the usual statements of libertarian principle imply conclusions that almost nobody, libertarian or otherwise, believes in.

Consider the following example. A madman is about to open fire on a crowd; if he does so numerous innocent people will die. The only way to prevent him is to shoot him with a rifle that is within reach of several members of the crowd. The rifle is on the private property of its legitimate owner. He is a well known misanthrope who has publicly stated on numerous occasions that he is opposed to letting anyone use his rifle without his permission, even if it would save hundreds of lives.

Two questions now arise. The first is whether members of the crowd have a right to take the rifle and use it to shoot the madman. The answer of libertarian rights theory, as I understand it, is no. The owner of the rifle is not responsible for the existence of the madman, and the fact that his rifle is, temporarily, of enormous value to other people does not give them a right to take it.

The second question is whether it is desirable that someone take the rifle and use it to shoot the madman--whether, to put it more personally, I wish that someone do so, or whether I would rather see the members of the crowd stand there and be shot down. The answer to this question seems equally unambiguous. If someone takes the rifle, there is a relatively minor violation of the legitimate rights of its owner; if no one does, there is a major violation of the legitimate rights (not to be killed) of a large number of victims--plus a substantial cost in human life and human pain. If asked which of these outcomes I would prefer to see, the answer is obviously the first.

This result is not, in any strict sense, paradoxical. An outcome may be desirable even though there is no morally legitimate way of achieving it. Indeed, this possibility is implied by the idea (due to Nozick) of viewing libertarian rights as "side constraints" within which we seek to achieve some objective; the constraints would be irrelevant unless there were some circumstances in which we could better achieve the objective by ignoring them.

While not in any strict sense paradoxical, the result is, at least to me, an uncomfortable one. It puts me in the position of saying that I very much hope someone grabs the gun, but that I disapprove of whoever does so.

One solution to this problem is to reject the idea that natural rights are absolute; potential victims have the right to commit a minor rights violation, compensating the owner of the gun afterwards to the best of their ability, in order to prevent a major one. Another is to claim that natural rights are convenient rules of thumb which correctly describe how one should act under most circumstances, but that in sufficiently unusual situations one must abandon the general rules and make decisions in terms of the ultimate objectives which the rules were intended to achieve. A third response is to assert that the situation I have described cannot occur, that there is some natural law guaranteeing that rights violations will always have bad consequences and that committing one rights violation can never decrease the total of rights violations.

All of these positions lead to the same conclusion. Under some circumstances rights violations must be evaluated on their merits, rather than rejected a priori on conventional libertarian natural rights grounds. Those who believe that rights violations are always undesirable will be sure that the result of the evaluation will be to reject the violation, but that does not mean that they can reject arguments to the contrary without first answering them. Any such argument claims to provide a counterexample to their general theorem, and if one such counterexample is true the general theorem must be false.

I have made my point so far in terms of an issue created for the purpose; whether or not to steal rifles in order to shoot madmen is not a burning issue in libertarian (or other) circles. I will now carry the argument a step further by defending one of the particular heresies which, it is widely believed, no libertarian can support--that under some conceivable circumstances a draft would be desirable.

Suppose we are threatened with military conquest by a particularly vicious totalitarian government; if the conquest is successful we shall all lose most of our freedom and many of us will lose our lives. It is claimed that only a draft can protect us. Two replies are possible. The first is that since coercion is always wrong we should reject the draft whatever the consequences. I have tried to show that that answer is not satisfactory--at the most it should lead us to refuse to enforce a draft ourselves while hoping that someone with fewer principles imposes one for us. Temporary slavery is, after all, better than permanent slavery.

The other possible reply is to deny that the draft is necessary. This can be done in many ways. The economist is inclined to argue that collecting taxes in cash and using them to hire soldiers is always more efficient than collecting taxes in labor; the moralist may claim that a society whose members will not voluntarily defend it is not worth defending. I have myself used the first argument many times; I believe that in the circumstances presently facing the U.S. it is correct. But the question I am currently concerned with is not whether under present circumstances, or even under likely circumstances, a draft is desirable. The question is whether under any conceivable circumstances it could be.

The answer is yes. Imagine a situation in which the chance of a soldier being killed is so high that a rational individual who is concerned chiefly with his own welfare will refuse to volunteer even at a very high wage. Imagine further that the percentage of the population required to defeat the enemy is so large that there are simply not enough patriotic, or altruistic, or adventure loving, or unreasonably optimistic recruits available; in order to win the war the army must also include selfish individuals with a realistic view of the costs and benefits to themselves of joining the army. Recruiters and preachers will of course point out to such individuals that "if everyone refuses to fight we will be conquered and you will be worse off than if everyone volunteers to fight." The individual will reply, correctly, that what he does does not determine what everyone else does. If everyone else volunteers, he can stay safely at home; if nobody else volunteers and he does, he will almost certainly be killed and if not killed will be enslaved.

Under such circumstances, an army could be recruited without a draft by paying very high salaries and financing them with taxes so high that anyone who does not volunteer starves to death. The coercion of a tax is then indistinguishable from the coercion of a draft. While a libertarian may still argue that to impose either a draft or a tax is immoral and that he himself would refuse to do so, I find it hard to see how he can deny that, under the circumstances I have hypothesized, he would rather see himself and everyone else temporarily enslaved by his own government than permanently enslaved by someone else's.

The point of this argument is not that we should have a draft. As it happens, I not only believe that under present circumstances a draft is a bad thing, I also believe that if the government has the power to impose a draft it is very much more likely that it will use it when it should not than that the rather unlikely circumstances I have described will occur. That is, however, a practical argument, and one that might depend on the particular circumstances of a particular time and place; it is not an argument of principle that would apply everywhere and everywhen.

Perhaps what these examples show is not that we cannot accept a simple statement of libertarian principle but only that I picked the wrong one. Perhaps we should replace a statement about what one should do ("never initiate coercion") with a statement about what objective one should seek ("do whatever minimizes the total amount of coercion"). Both seizing the rifle and imposing a draft are then, in the particular circumstances I have described, not only consistent with libertarian principle but required by it.

While I cannot speak for other libertarians, I find that this version of libertarianism does not always fit my moral intuition. Suppose the only way I can stop someone from stealing two hundred dollars from me is by stealing your hundred dollar rifle (which you are unwilling to lend or sell me) and using it to defend myself. The result is to reduce the total amount of coercion, at least if we measure amount by value of what is stolen. Yet it seems, at least to me, that stealing the rifle is still wrong.

A second problem with this approach is that it is of no help when we must choose between a small cost in coercion and an enormous cost in something else. Suppose you happen to know that everyone in the world is going to die tomorrow (by some natural catastrophe, say the earth colliding with a large asteroid), unless you prevent it. Further suppose that the only way to prevent it involves stealing a piece of equipment worth a hundred dollars from someone who, in your opinion, rightfully owns it. Your choice is simple: violate libertarian principles by stealing something or let everyone die.

What do you do? You cannot justify stealing as a way of minimizing total coercion. Being killed by an asteroid is not coercion, since it is not done by a person. After the asteroid strikes there will be no more coercion ever again, since there will be no one left to either coerce or be coerced.

Speaking for myself, the answer is that I steal. When I put such questions to other libertarians, one common response is a frantic attempt to reinterpret the problem out of existence. One example might be the reply that, since the person you are stealing from will himself be killed if you do not take the device, he would be in favor of your taking it, so you are not really stealing--you are using the device in the way he would want you to if he knew what you know. Another response might be that you should not steal the equipment because your belief that doing so will save the world may be wrong .

All such evasions are futile. I can always alter the assumptions to force the issue back to its original form. Perhaps the owner of the device agrees that using it is necessary if the world is to be saved, but he is old, tired of living, and not very fond of his fellow humans. Perhaps the situation is so clear that everyone agrees that without your act of theft we shall all die.

Our response to such questions demonstrates that we do not really believe in simple single values. Most libertarians, myself among them, believe that a libertarian society is both just and attractive. It is easy enough to claim that we are in favor of following libertarian principle whatever the consequences--given that we believe the consequences would be the most attractive society the world has ever known. But the claim that we put individual rights above everything else is, for most of us, false. Although we give some value, perhaps very great value, to individual rights, we do not give them an infinite value. We can pretend the contrary only by resolutely refusing to consider situations in which we might have to choose between individual rights and other things that are also of great value.

My purpose is not to argue that we should stop being libertarians. My purpose is to argue that libertarianism is not a collection of straightforward and unambiguous arguments establishing with certainty a set of unquestionable propositions. It is rather the attempt to apply certain economic and ethical insights to a very complicated world. The more carefully one does so, the more complications one is likely to discover and the more qualifications one must put on one's results.

Chapter 42: Where I Stand

In the previous chapter, I argued that simple statements of libertarian principle lead to unacceptable conclusions and must therefore be rejected. There is no obvious logical inconsistency in a moral principle that implies that nobody should be permitted to breathe, but it is not a principle that many people are likely to accept.

One possible response is that libertarianism is an absolute principle, an ultimate value which cannot be overridden, but that it is not adequately expressed by the simple statements I have been attacking. If those statements are only approximations to a much more complicated and subtle description of libertarian principle, it is hardly surprising that the approximation sometimes breaks down in difficult situations.

This is a view with which I have a good deal of sympathy, but it is not very useful for answering real world questions, at least until someone manages to produce an adequate statement of what libertarian principles really are. Moral philosophy is a very old enterprise and its rate of progress has not been rapid in recent centuries, so I do not plan to hold my breath while I wait.

A second response, and one with which I also have a good deal of sympathy, is that there are a number of important values in the world. They cannot be arranged in any simple hierarchy, or at least are not going to be anytime soon. Individual liberty is an important value in and of itself, not merely as a means to happiness, so we should not be willing to sacrifice large amounts of it in exchange for small amounts of happiness. But liberty is not the only value, nor is it infinitely important compared to other values, so we should not be willing to sacrifice unlimited amounts of happiness for small gains in liberty.

A third possibility is that the conflict between libertarian and utilitarian values is only apparent. Perhaps there is some deep connection between the two, so that libertarian ethics, properly understood, is the set of rules that leads to the maximum of human happiness. The counterexamples given in the previous chapter must then be interpreted as some combination of mistakes about what is possible--for some reason those situations could not arise in the real world--and mistakes about what is implied by a correct statement of libertarian principle. Something along these lines seems to be suggested by the arguments of those libertarian philosophers who claim to get their principles not by generalizing from what seems right or wrong to them but by deducing what set of rules is appropriate to the nature of man.

One argument in favor of this approach is that it fits the observation that libertarianism and utilitarianism, while quite different in principle, frequently lead to the same conclusion. Through most of this book I have used utilitarian arguments to justify libertarian conclusions. By doing so, I provided evidence that the potential conflicts between the two approaches which I discussed in the previous chapter are the exception rather than the rule. In Chapter 31, I tried to show that the institutions of anarcho-capitalism would tend to generate libertarian laws. A key step in that argument was my claim that the value to individuals of being able to run their own lives is typically greater than the value to anyone else of being able to control them--or in other words, that increases in liberty tend to increase total utility.

A fourth possibility, and the last which I will consider, is that libertarianism is wrong and we should accept utilitarianism instead. According to the strict utilitarian position, rules, actions, ethics, must be judged solely by their effect on the sum (some utilitarians would say the average) of human happiness. Whatever increases happiness is good; whatever decreases it is bad. Libertarian principles are then valued only as a means, a set of rules that frequently lead to increases in total utility and should be rejected when they do not. This again is a possible interpretation of arguments that claim to derive libertarian principles from the nature of man, although not, in my experience, an interpretation that those who make such arguments are willing to accept.

One argument against utilitarianism is that it cannot be a correct moral rule because there is no way we can tell whether we are following it. We cannot observe other people's utility and are therefore unable to judge what will increase it. Even if we could observe individual utilities, we do not know how to compare the utility of different people and so have no way of judging whether a gain in happiness to one person does or does not balance a loss to another.

I find this argument unconvincing. Consider the act of buying a present. If you really have no knowledge at all about what makes other people happy, then buying a present is pure guesswork; you might just as well open a page of the Sears catalog at random, throw a dart at it, and buy whatever you hit. Nobody believes that; if we did, we would not buy presents.

Consider a court awarding damages. If we really know nothing at all about other people's utility, how can a court decide how much someone owes me for breaking my arm? For all the judge knows, I enjoyed having my arm broken. Assuming that I disliked it, he has no way of knowing whether my disutility for a broken arm is measured by a penny or a billion dollars.

We give presents and award damages, and we do not believe that other people's utility is entirely unobservable. What we do believe, or at least what many of us believe, is that each of us knows more about his own values than most other people do, and that people are therefore usually better off deciding what they want for themselves. That is one of the main arguments in favor of a free society. It is a long step from that to the claim that we know nothing at all about other people's values.

Even if we were entirely unable to observe other people's values, that would not necessarily prevent us from constructing a society designed to maximize total utility. Each person knows his own values, so all of us put together know everybody's values. In order to maximize the total utility of the society, we would construct rules and institutions that utilized all of that information via some sort of decentralized decisionmaking system, with each person making the decisions that require the particular knowledge he has.

This is not, of course, merely an abstract possibility. One of the strongest arguments in favor of letting people interact freely in a market under property rights institutions is that it is the best known way to utilize the decentralized knowledge of the society--including the knowledge that each individual has about his own values. The field of welfare economics largely consists of the analysis of the rules that lead to optimal outcomes under specified circumstances, where the outcomes are evaluated in terms of the preferences of the individuals concerned. One originator of modern economics, including much of welfare economics, was Alfred Marshall, an economist and utilitarian who viewed economic theory in part as a way of figuring out how to maximize total utility.

Even if individual preferences can be observed, either directly or as reflected in actions, we are still left with the problem of comparing them. How can we say whether something which makes one person worse off and another better off produces a net increase in human happiness?

The answer, I believe, is that we may not be able to make such comparisons very well or describe clearly how we make them, but we still do it. When you decide to give ten dollars worth of food and clothing to someone whose house has just burned down instead of sending a ten dollar check as an unsolicited gift to a random millionaire, you are expressing an opinion about which of them values the money more. When you decide where to take your children for vacation, you are making a complicated judgement about whether their total happiness will be greater camping in a forest or wading on the seashore. We cannot reduce the decision to a matter of precise calculation, but few of us doubt that the unhappiness A gets from the prick of a pin is less than the unhappiness B gets from being tortured to death.

Utilitarianism is a possible moral rule. The difficulties of applying it to real world problems are substantial, but so are the difficulties of applying an alternative rule such as minimizing coercion. One would face very similar problems in defining and measuring the amount of coercion and in judging the tradeoff between increased coercion for one person and decreased coercion for another.