The God of the Machinery


The Machinery of Freedom, by David Friedman, New York: Harper Colophon Books, 1973, 239 pp., $2.25 (paperback).

David Friedman's THE MACHINERY OF FREEDOM is an intriguing and stimulating libertarian investigation. It is worthy of a detailed critical discussion and that is what I propose to offer here. The reader should be aware that at all points my purpose is to reach a correct understanding and evaluation of the arguments under consideration. My purpose is not to defend or attack any particular school of libertarian thought, e.g., anarcho-capitalism or limited governmentalism.


The three chapters, "Police, Courts, and Laws—on the Market," "The Stability Problem," and "Is Anarcho-Capitalism Libertarian?" (pp. 155-78) constitute the most sustained single argument in THE MACHINERY OF FREEDOM. In these chapters, Friedman attempts a special type of justification of the anarcho-capitalist view. Friedman's clear discussion of private police and defense agencies (pp. 155-58) leads him to the fundamental questions, "In…an anarchist society, who would make the laws? On what basis would the private arbitrator decide what acts were criminal and what their punishment should be?" (p. 159) Friedman's forthright answer is,

[S]ystems of law would be produced for profit on the open market, just as books and bras are produced today. There could be competition among different brands of law, just as there is competition among different brands of cars.

In such a society law is produced on the market. A court supports itself by charging for the service of arbitrating disputes. Its success depends on its reputation for honesty, reliability, and promptness and also on the desirability to potential customers of the particular set of laws it judges by. [pp. 159-60.]

Friedman's position is quite different from the usual anarcho-capitalist approach on this issue. As a means of understanding Friedman's position, let us introduce the terms, "free market law" and "libertarian law." "Free market law" means law that is produced through a market system of laws and courts. "Libertarian law" means law which accords with libertarian principles. The usual anarcho-capitalist position is to argue that, given a sufficiently large libertarian public opinion, the defenders of libertarian law (the defense agencies and courts which respond to the libertarian public opinion) will be able to "monitor" the market for law, and will, quite legitimately, prevent the emergence of nonlibertarian law or the enforcement of nonlibertarian law. Such "monitoring" of, or "interference" in, the market is seen as just as legitimate as any other "interference" to prevent criminal activity. It should be clear that the legitimacy of this type of "monitoring" or "interfering" is consistent with, indeed even required by, any plausible libertarian theory. It is by projecting such "monitoring" that the anarcho-capitalist usually explains why, given a sufficiently large libertarian public opinion, the only law which will emerge as free market laws will be libertarian laws.

The usual approach, then, is to ask and answer the question, "How (under what conditions) will the market supply libertarian laws?" Friedman's approach is to ask the question, "What law will the market supply?" Friedman is concerned with avoiding, as much as possible, arguments which rely upon claims about "rights, ethics, good and bad, right and wrong." (p. 223) Since the "monitoring" of the usual anarcho-capitalist approach is defined in terms of preventing the violation of rights, just defense, etc., it involves the introduction of the type of normative claims which Friedman would rather avoid. Instead of taking the approach of indicating how the laws that are provided by private firms may be legitimately restricted (by private firms) to libertarian law, Friedman confines his study to the dynamics of the private (i.e., free market) provision of laws and courts. He is concerned with how the machinery of freedom as such operates. The machinery of freedom—the market—is, of course, fueled by persons' desire to get what they value at the least overall cost to themselves.


What Friedman would like to show is that the market would supply, or would tend to supply, libertarian law—independently of any policy of libertarian "monitoring." Thus, Friedman attempts a type of value-free account of libertarian law. The account makes no reference to persons' acting out of libertarian attitudes or convictions. If Friedman can establish that there is a significant tendency, at least, for the machinery of freedom as such to produce libertarian laws and courts, he will have accomplished two further things: (a) a scenario for libertarian law will have been provided which does not involve the "moralism" of the usual anarcho-capitalist scenario (or of the limited governmentist scenario), and (b) a reason will have been provided for preferring anarcho-capitalism to limited governmentalism, since we know the political (nonmarket) production of law has no inherent tendency to yield libertarian law.

I turn, then, to an examination of Friedman's scenario. A distinctive feature of this account is the emphasis on the role of bargaining: two defense agencies represent their respective clients in negotiations which result in agreements about what laws, if any, will apply to the dealings of the members of one group with the members of the other group. For example, suppose that members of group A strongly desire that the rule of execution for murder apply, while the members of group B desire some less harsh rule. Defense agency A (representing group A) will offer something to defense agency B and (thereby) to the members of group B in return for their agreement to the more harsh rule. Perhaps defense agency A (acting on behalf of group A) will offer money. Or, perhaps, it will offer acceptance of some other law which members of group A would rather do without but members of group B desire. In any case, if some agreement is reached, each party gains something which it values more than what it must give up. It is extremely likely that some agreement would be reached just because it would be valuable to all concerned. The process of agreeing upon a set of laws may, of course, take the form of agreeing upon what particular court (where we imagine courts to vary in the laws which they are pledged to enforce) will have jurisdiction in specific types of cases. Thus, Friedman speaks of defense agencies agreeing upon what courts will be patronized.[2]

With this understanding of the machinery of freedom in laws and courts we can turn to the question, "Is ("unmonitored") anarcho-capitalism libertarian?" Specifically, will this machinery of freedom as such generate a libertarian legal system? Friedman's affirmative answer seems to come on two distinguishable levels: (1) he argues (pp. 161-62) that the bargaining aspect of anarcho-capitalism in laws and courts maximizes the extent to which persons can live under the laws which they want to live under; and (2) he argues (pp. 173-76) that, under anarcho-capitalism in laws and courts, the costs and benefits of laws are assigned in such a way as to create a tendency toward libertarian law.


Friedman contrasts the anarcho-capitalist bargaining conception of the generation of laws with nonmarket, i.e., political, determination of what laws will apply to the populace. The virtue of the former, compared with the latter, is that the bargaining process allows for "the customer's legal preference, his opinion as to what sort of law he wishes to live under…[being] a major factor in determining the kind of law he does live under." (p. 162.) However, a particular customer has this power to avoid living under unwelcome laws only if he is a member of a group which has something to offer in trade—either money or concessions—to the members of the groups which propose the unwelcome laws.

It is true that, insofar as such deals are made, there is an overall maximization of persons' living under laws that are desired—whatever those desired laws are. But we should note that our current political-legal processes can be understood in a way which makes their operation very much like the complex trading-off processes which, according to Friedman, are a central feature of a market in laws and courts. Persons are members of (multiple) interest groups. These interest groups form coalitions, bargain, make deals, and so on, with the aim of producing laws most compatible with the wishes of their members. Interest groups bid for laws through offers of political, financial, and moral support and through offers of concessions and alliances. (Some political scientists consider the genius of this pseudomarket arrangement to be the maximization of the satisfaction of political-legal demands.) Obviously, there is some difference between Friedman's anarcho-capitalist scenario, as sketched to this point, and the interest group picture of our current political scene. But the question is, "How significant are those differences?" The parallel between Friedman's scenario, as sketched to this point, and this picture of the current processes which determine law is, I think, extensive. For example, a person who is not a member of (a large enough number of) sufficiently strong interest groups is up the political-legal creek in much the same way as a customer for laws and courts would be if he, individually or in union with those of like interests, had nothing to offer in trade to those who proposed laws which the customer finds offensive. In both cases, the weakly positioned person would have to accept those offensive laws—eking out what minor concessions he could. Of course, in any society the relatively weak are in a precarious position. My point is simply to suggest that, if there is an important difference between market and political means of determining law, it is not the element of bargaining which makes that difference. Although the bargaining mechanism which Friedman describes would be fruitful in allowing persons to live under laws which they want to live under, this fruitfulness does not signify a tendency within the mechanism to yield libertarian law.

Clearly what is missing in the bargaining mechanism, as it has been presented to this point (and what is missing in the current political-legal context) is a limitation upon the type of bargaining positions which the parties to negotiations take. Friedman does claim that there are limitations on the bargaining positions which are taken by the negotiators which he envisions. Of course, these limitations do not proceed from the type of "monitoring" which I previously mentioned as typical of the anarcho-capitalist position. Instead, and specifically, Friedman wants to argue that, under anarcho-capitalism in laws and courts, the costs and benefits of laws accrue to individuals, and are known to accrue to individuals, in such a way as to place a damper on the demands for laws that are made by individuals. Thus, Friedman's full scenario for the operation of an unmonitored market in laws and courts does include certain limitations upon the demands for law which enter that market. We must see what these limitations are so that we can judge whether, in the light of these limitations, a market for laws and courts as such has a tendency, at least, to generate libertarian law.


Within the anarcho-capitalist situation, people's desires for the existence of this or that law are based upon their conceptions of what laws are in their interests under the circumstances that obtain. One will not have defense agencies or courts enforcing (or attempting to enforce) particular laws if the costs of having such defense agencies or courts, to those who would otherwise patronize those defense agencies and courts, would be higher than the anticipated benefits for those potential customers. According to Friedman (as I construe his argument), the costs to the potential customers can be too high in two different ways.

First, the actual monetary costs can be high relative to the benefits (usually psychic) of enforcement. To some extent this is true, for instance, in cases of laws creating victimless crimes (the prohibition of consensual conduct, such as gambling, prostitution, etc.). If those who desire such laws could not fob off, through the political process, a large percentage of the costs of enforcement, they would not purchase such laws. In effect, Friedman argues that victimless crimes are "public goods" which would not be provided on a free market for lack of individuals who would freely contract to pick up the tab. Actually, as Friedman points out, the tab would be quite low in geographical areas where the demands for drink, sex, drugs, etc., were small. So persons might be willing to support courts whose laws forbid drink, sex, drugs, etc., in those areas in return for the psychic benefit of keeping the countryside clean.

However, a court whose laws required universal cleanliness would, other things being equal, find itself without patrons. Defense agencies who committed themselves to the courts of puritanical law would incur much greater expenses than the defense agencies who committed themselves to the courts of geographically restricted puritanical law, and this would put the more puritanical defense agencies at a great competitive disadvantage in terms of the fees they would have to charge. Thus, insofar as people's desires not to be subject to certain laws does significantly increase the monetary costs of subjecting them to those particular laws, the persons with those desires will tend to be free from the application of those particular laws.

Second, the actual costs, to the patrons, of patronizing a particular court, would be high relative to the benefits which accrued to the patrons from the enforcement or attempted enforcement of that court's laws if nonpatrons would inflict high costs upon the patrons because of their patronage of this court. For instance, if defense agency A were to patronize a court whose law was that theft from persons who were not clients of defense agency A was legitimate, the nonclients of agency A would inflict high costs upon defense agency A, and directly or indirectly, upon the supporters of agency A.

In explaining the tendency for those who would be on the receiving end of victimless crime laws to be free of such laws under the anarcho-capitalist mechanism, Friedman says, "They [the would be victims] are willing to pay a much higher price to be left alone than anyone is willing to pay to push them around." (p. 174.) Actually, this remark is inappropriate to Friedman's discussion of victimless crime laws. For those cases illustrate how the high monetary cost for the would-be enforcers puts a damper on the generation of nonlibertarian law. In those cases, the would-be victims neither directly pay or bargain to be left alone, nor do they pay indirectly by hiring a larger defense agency than the one in the pay of the puritans. The claim, "They are willing to pay a much higher price to be left alone than anyone is willing to pay to push them around," would be more appropriate to the cases of would-be victims inflicting costs upon those who back certain victimizing laws. And Friedman does offer parallel claims in his discussion of the potential for Mafia defense agencies (p. 167):

The victims of theft will be willing to pay more to be protected than the thieves will be able to steal (since stolen goods are worth less to the thief than to the victim). Therefore the noncriminal protection agencies will find it profitable to spend more to defeat the criminal agency than the criminal agency could spend to defeat them.

It seems to me that the significant fact is that if the criminal agency goes into operation (perhaps with the approval of some private court), the agency (and its clients, and perhaps an approving private court) will have unacceptable (nonmonetary) costs inflicted upon them by the superior resources of the would-be victims. To summarize my construction of Friedman's points: those who would like victimless crime laws tend to be restricted by the costs of the (successful) enforcement of those laws, while those who would like the legality of theft, murder, etc., tend to be restricted by the costs of the (unsuccessful) attempts at enforcement of these laws. Of course, each manner of restriction may be a factor in the case of any specific law. Now, interesting as these arguments or suggestions are, they remain unsatisfactory on two counts.


The first problem is that it is not at all clear that Friedman has established that the operation of the machinery of freedom, even as we now understand it, would generate, or tend to generate, a legal system which was more libertarian than the one which confronts us today. After all, murder and theft which is not directed at supporting "vital public projects" is not today recognized as legal. One can easily imagine almost everyone, under anarcho-capitalist institutions, agreeing (as almost everyone agrees today) that for the support of "vital public projects" such as defense, health, welfare, education, transportation, etc., individuals should be taxed, pushed around, drafted, etc. And if almost everyone did agree to this, then the market would respond with laws that would be comparable to those we have today. It is far from obvious that if a particular state of public opinion were to be translated into law, through a market for laws and courts, a more libertarian legal system would result than would appear if that state of public opinion were to be translated into law through political institutions.

The second problem is that there seems to be an incompatibility between Friedman's emphasis on the bargaining which defense agencies would engage in to determine what courts will have jurisdiction over their clients and the view that the outcome of this bargaining will be a commitment to libertarian law. The problem involves what type of bargaining positions will be fed into the machinery of freedom. In Friedman's scenario the negotiating parties make no distinction between justified demands for laws and courts and unjustified demands. The demands which appear merely reflect the preferences of the various groups of clients without any distinction being made between those demands which would, for instance, require the violation of rights and those demands which require no such violation. It is clear from Friedman's discussions, especially his discussion of antiheroin laws (pp. 174-76), that Friedman fully realizes that nonlibertarian demands will come to the bargaining table. What I will argue is that insofar as the machinery of freedom in laws and courts includes bargaining tables of this sort—in effect, value-free bargaining tables—an antilibertarian tendency will be built into the operation of the machinery of freedom. I have two reasons for saying this:

(a) Consider the heroin example in which competitive pressures result in the patronage of courts which outlaw heroin in most areas but permit it, say, in New York City. Only courts which do not challenge those rules will survive competitively. All defense agencies will patronize one or another of the surviving courts of geographically-restricted antiheroin law. Thus, any individual who wants to be a client of a defense agency will have to sign a contract which (since the contract will specify the courts patronized by the defense agency) will involve the individual's granting his consent to the geographically-restricted antiheroin law. Now, in one sense, this is no worse than the fact that if I want to buy a copy of THE MACHINERY OF FREEDOM I have to buy a paperback copy. Those who produce the product that I want simply did not see fit to produce and offer hardbound copies. I have no just complaint in this case. The reason is that I have no moral claim to the possession of a hardbound copy of this book. But I do have a moral claim, ceteris paribus, not to be interfered with in my (possible) use of heroin. Within Friedman's value-free scenario this moral claim, and its corollary that any prohibition of heroin use is unjustified, are obscured. The demand for heroin prohibition is not identified as any less legitimate than the demand for heroin laissez-faire—and, in the absence of this comparative judgment, the compromise prohibition rule has the status of a rule to which everyone has consented. Having been consented to by all, the rule must be fair and proper. One could not say, while accepting Friedman's approach, that the believers in heroin laissez-faire were forced, through threats of complete prohibition, to accept the compromise prohibition rule—for the demand for complete prohibition is no less legitimate as a bargaining position than the demand of the laissez-faire advocates. In presenting these clashing demands, the two groups would be acting equally properly in the pursuit of their respective legal preferences.

(b) The nature of the value-free bargaining tables would tend, I think, to encourage the escalation of nonlibertarian demands. After all, what persons or defense agencies have to trade-off at the bargaining tables are concessions. The greater one's original demands, the greater the concessions one can make. Of course, some demands need not be taken too seriously, for in one way or another, the cost of implementing those demands would be more than those who put forth the demands would actually be willing to bear. However, there will always be uncertainty about whether or not those who are making certain demands are willing to risk the costs of implementation. Suppose, for instance, that in the course of the heroin negotiations the puritanical defense agency escalates its demands. It announces its strong preference for alcohol prohibition. Now the agents of the more laissez-faire defense firm are not sure how strong this preference is. But it's worth something to them to eliminate the risk of "going to the mattresses" over the alcohol demand. What do they do? Either they agree to pay the puritanical agency a sum of money to withdraw this demand or they announce some escalation (and it needn't be a libertarian escalation) in their own demands with the hope of bargaining down the puritanical position. I submit that if we project this type of tendency in negotiations over the whole range of human activities which are conceivably subject to enforced rules, and if we remember that negotiations would be going on between a multitude of defense agencies, it is very hard to be optimistic about the libertarian quality of the resulting law.

I conclude, then, that it seems unlikely that the operation of a market for laws and courts as such would produce, or significantly tend to produce, libertarian law. A scenario for the appearance of libertarian law which projects persons and defense agencies as the sources of, and the respondents to, demands for laws and courts, but as sources and respondents whose actions are not to be understood (within the scenario) as flowing, even partially, from libertarian judgments about the legitimacy or illegitimacy of demands, will not be a plausible scenario. Friedman does, sometimes, write as if he is taking what I have called the usual anarcho-capitalist approach, i.e., as if what are fed into the machinery of freedom are legitimate demands. For instance, he says at one point, (pp. 170-71),

Under a system of private protection agencies, the actual agencies, like the ideal government, are merely acting as agents for willing clients who have employed the agencies to enforce their own rights. They claim no rights over nonclients other than the right to defend their clients against coercion—the same right every individual has. They do nothing that a private individual cannot do.

However, this passage is not compatible with Friedman's typical characterization of defense agencies and their demands.

I close this part of the discussion with an indication of another problem in Friedman's value-free conception of the market for laws and courts. What bothers me is an apparent incompatibility within the scenario that he presents. Within this scenario, persons recognize rights (especially the rights of property) or the value of freedom (especially the value of freedom of ownership and contract) to such an extent that private defense agencies can exist and enter into contracts with individuals and private courts. Yet it is suggested that we exclude persons' invocations of rights or the value of freedom from any role in explaining the appearance of laws in the society which is represented by the scenario. This methodological restriction seems not merely unnecessary, but also incompatible with the conception which Friedman invites us to entertain.

Friedman's attempt to show how a libertarian society would be, or would tend to be, produced by the operation of the machinery of freedom as such fails. But Friedman's failure falls into that class of noteworthy failures from which important lessons can be drawn. The key lesson is that one cannot give an account of the structure and working of a libertarian society without referring to the libertarian attitudes of the persons within that society and the effect of those attitudes upon their preferences and actions.


Among the vices: There is another philosophical problem in Friedman's book which is important to mention. Friedman joins a good number of other libertarians in attempting to use "property" or "ownership" in order to explicate "rights". Any such attempt is doomed to failure. Specifically, Friedman attempts to use "property" and/or "ownership" to explicate the meanings of "being left alone" and "being coerced," and in turn, he utilizes "being left alone" and "being coerced" in defining "rights" (p. xiv). This added step does not circumvent the fundamental error in such an approach to the explication of "rights." For either "property" is defined as that which one has a right to control, or "property" is defined as that which one does control. If one's property is defined as that which one has a right to control, then "property" cannot be used to explicate "rights," for "rights" is being employed in the explication of "property." If, on the other hand, one's property is defined as that which one does control (cf. "Your property is that which you control the use of." p. 131), then "property" still cannot be used to explicate "rights"—for, on this definition of "property," no transfer of (control of) objects from person A to person B would ever involve B's taking over (control of) A's property. On this definition of "property," B's taking over (control of) an object makes that object B's property! Thus, on this definition of "property," there can be no such thing as the invasion of, or theft of, another person's property (even another person's property in his own body)—there can be no rights in any usual sense. Hence, if one's property is defined as that which one does control, "property" cannot be used to explicate "rights." Whether "property" is defined as that which one has a right to control or that which one does control, "property" cannot be utilized to explain the meaning of "rights." A parallel argument is easily constructed for "ownership." Q.E.D.

There are certain oddities within Friedman's discussion which I will merely take note of. There are the confused remarks in connection with the alleged belief, "in our hearts that there is only one good and that ideally everyone should pursue it." (p. 22.) And there is the argument centering on the claim that, "If we accept the concept of needs, we must also accept the appropriateness of having decisions concerning those needs made for us by someone else, mostly likely the government." (pp. 66-67.) And there is the confusion, common to many libertarians, between normative and descriptive laws (p. 163). Last, in his discussion of the liability of private protection agencies (and their operatives) for unintended harm or coercion (pp. 171-72), Friedman mistakenly assumes that persons are always liable for actions which are, after the fact, discovered to have been illegitimate. This is incorrect: there are such things as excuses.

Among the virtues: Friedman's book is replete with ingenious insights and points. These are, in fact, too numerous to indicate individually. Among the virtues of the book is an excellent discussion (pp. 3-14) of the role of property—private versus public—in determining the way decisions will be reached about the employment of scarce resources. Friedman is superb at tracing the counterintuitive consequences of various nonlibertarian policies, and at destroying antilibertarian socio-economic myths. (Cf. pp. 26-60, 133-38, 144-46, and 207-12.) Friedman's emphasis on bargaining between defense agencies to determine what court (or appeals court) will have jurisdiction is important to libertarian (and especially free market anarchist) theory.

The genuine importance of the discussion of bargaining can be appreciated in the light of a common libertarian error. It is commonly assumed by all varieties of libertarians that there is one ruling in any particular case which is demonstrably "the correct libertarian ruling." It is assumed that two rational and honest individuals (or groups of people, or organizations) who accept some basic libertarian principle of rights, or noncoercion, or nonaggression, will always, ultimately, agree on what is the proper ruling to be made in response to any particular accusation. This assumption is, unfortunately, mistaken. The application of basic libertarian principle(s) is a very complex matter, and the application may not, even in principle, yield any one particular judgment. For instance, to consider a simple case, what particular judgment follows from the application of libertarian principle(s) as to what ought to be done to, for, or with, this or that particular rapist? It seems that here, as in other types of cases, libertarian philosophical and legal reasoning can only set limits upon the range of morally permissible and/or morally appropriate responses to the criminal's action. Within that limited range, the demand for one response is just as legitimate as the demand for another. Typically we would expect the criminal to demand a less harsh response than that demanded by the victim. The possibility of a multiplicity of equally proper judgments is a problem for both the free market anarchist and the limited governmentalist. The limited governmentalist can tell us which of the equally proper judgments will be enforced, viz., the judgment made by the government will be enforced. But the justice of this decision procedure is not obvious. The free market anarchist cannot tell us which of the equally proper judgments will be enforced. But he can point out that his system allows for a fair procedure for determining which of the judgments within the range of equally proper judgments will be applied. The procedure is that of bargaining between the interested defense agencies. Here, of course, the range of legitimate bargaining positions is seen as limited by libertarian principle(s). Thus, Friedman's emphasis on bargaining can be put to use as a supplement to the usual anarcho-capitalist conception of defense, laws, and courts. It is a supplement for which there is no counterpart (though a counterpart is needed) in the limited governmentalist conception of defense, laws, and courts.

Last, I recommend Friedman's "National Defense: The Hard Problem" (pp. 185-97) as an excellent, fair-minded, and realistic (though, perhaps, overly pessimistic) look at the problem of providing defense against aggressor nations.


Friedman's last substantive chapter, "The Right Side of the Public Good Trap," strikes me as clever and important. What he does here (pp. 213-17) is to offer a political version of the public good argument, and then uses that argument as a reason for preferring free market anarchism to limited governmentalism. According to Friedman, the problem for all defenders of political law and courts lies in the fact that, under a system of political determination of the law, good (i.e., libertarian) laws are public goods. That is, although the total benefits of good law may be greater than the total costs of implementation, it does not pay individuals to expend their political energies in support of such laws. An equal expenditure of energy directed toward the passage of some special interest legislation which would favor the acting individuals would yield higher benefits for those particular individuals. Thus, there is a natural, egoistically motivated, tendency for individuals to expend their political energies in support of special interest laws of the sort which would benefit them. Thus, there exists a tendency toward bad—i.e., nonlibertarian—laws.

Even if I am just as willing to make money in a way that benefits others as in a way that harms them, the existence of governmental institutions makes it enormously easier for me to do the latter. The result is that in a society such as ours, in which most people would really rather produce than steal, we all spend a large part of our time using the laws to steal from each other. [215.]

As I understand Friedman, he is saying that this tendency toward true dog-eat-dogism is inherent in the nature of the political creation of law.

Up to this point Friedman's argument is clear, and his remarks constitute an important theory about all political processes. From this point, Friedman's argument is less clear. A summary of the whole argument runs,

[P]ublic goods are underproduced, relative to private goods. Under the institutions of government, bad laws—laws that benefit special interests at the expense of the rest of us—are private goods (more precisely, they are more nearly private goods than good laws), and good laws, laws that benefit everyone—such as laws that leave people alone—are public goods. Under an anarchy good laws are private goods and bad laws, public goods, [p. 217.]

Thus, to add the valid conclusion, under institutions of government good laws are underproduced relative to bad laws, while under an anarchy bad laws are underproduced relative to good laws. The crucial premise for determining whether the total argument is sound is the claim that, "Under an anarchy good laws are private goods and bad laws, public goods."

Two paragraphs seem to be devoted to establishing this premise (p. 216). We must, I think, largely ignore Friedman's first remark that, "Since under the institutions I have described, each individual 'buys' his own law and gets the law he buys, law itself ceases to be a public good." (p. 216.) This remark is incompatible with Friedman's own crucial premise which states that under an anarchy bad laws are public goods. Further we must assume that by "under an anarchy" Friedman does not merely have in mind a free market in laws and courts. Rather, in this argument, he has in mind a free market in laws and courts which is guided by libertarian demands. Friedman says that "the people who defend the anarchist institutions" are "individual customers insisting on laws that leave them free to run their own lives" and "members of protection agencies protecting their clients from coercers" (p. 216). Thus, Friedman is hypothesizing an anarchy characterized by, at least, a fairly strong libertarian public opinion.


Friedman's position, as I construct it, comes down to the claim that, comparing a society characterized by a market in laws and courts (and characterized by a fairly strong libertarian public opinion) and a society characterized by a limited government (and, to maintain the parallel, by a fairly strong libertarian public opinion), the former is more likely to remain libertarian than the latter. The reason is that it is more likely that in the limited government society particular individuals would find the relative benefits of pushing for improper laws (compared with the relative benefits of protecting the existing, legitimate, laws) to be high, while it is likely that in the anarcho-capitalist society particular individuals would find the relative costs of such antilibertarian activity to be high. The connection of this argument with the private versus public goods distinction lies in the fact that public goods are just those which it is relatively nonbeneficial for particular individuals to pursue, while private goods are just those which it is relatively beneficial for particular individuals to pursue. Thus, an alternative way of stating the reason for the greater libertarian stability of the anarcho-capitalist society would be: It is more likely that in the limited government society particular individuals would find improper laws to be private goods than it is likely that in the anarcho-capitalist society particular individuals would find improper laws to be private goods. This argument seems to me to be quite powerful. It deserves a place in the arsenal which the anarcho-capitalist deploys against the limited governmentalist.

I must add, in concluding my discussion of this argument, that Friedman's own defense of the premise that, "Under an anarchy good laws are private goods and bad laws, public goods" is unclear. Friedman says,

[Under an anarchy] it is…the person who wishes to reintroduce government who is caught in the public good problem. He cannot abolish anarchy and reintroduce government for himself alone; he must do it for everyone or for no one. If he does it for everyone, he himself gets but a tiny fraction of the "benefit" he expects the reintroduction of government to provide, [p. 216.]

By "reintroducing government" Friedman must have in mind reintroducing special interest, nonlibertarian, "bad" laws, for his goal is not to establish the unlikelihood, under anarcho-capitalism, of the reintroduction of government as such. Rather, his goal is to show the relative unlikelihood, under anarcho-capitalism, of the reintroduction of bad laws. Friedman's argument is that "government," i.e., bad laws, are less likely to be reintroduced under an anarchy than under a limited government because, under an anarchy, " government" would be a public good. The problem with this way of putting the argument is that it leaves the impression that "government" is a public good in the sense of being a good which does involve greater total benefits than total costs. And this view of "government," of course, contradicts the whole thrust of Friedman's position. What Friedman ought to do, when he rewrites this section, is to distinguish between the various senses of "public good" which populate his arguments. Sometimes, by "public good", he means a good which is desired by some persons but which it does not pay for them to pursue under the circumstances. And something can be a public good in this sense, and yet fail to be a public good in the more conventional sense—which Friedman sometimes employs—wherein the existence of a public good involves greater total benefits than total costs.

No one should infer from my remarks that I think THE MACHINERY OF FREEDOM is a bad book—I would not have written this analysis if I thought that. Original arguments are rare and delightful things, and THE MACHINERY OF FREEDOM has a very impressive and pleasing number of them. This is its major virtue. And this is why it strikes me as a pity that these arguments are not more fully, systematically, and self-critically, developed. Be that as it may, the original arguments and analytical forays are there. They are sufficient to make THE MACHINERY OF FREEDOM an important book.

Eric Mack is a professor of philosophy at Eisenhower College, Seneca Falls, N. Y. and an associate editor of REASON PAPERS.

[1] All page references given parenthetically are to THE MACHINERY OF FREEDOM (New York: Harper and Row 1973).
[2] Friedman claims that, "In principle, there could be a different court and a different set of law for every pair of protection agencies." (p. 159.) This is mistaken. Consider agencies A and B. There is no reason why there could not be one court (with its unique set of laws) which had jurisdiction over complaints made by clients of A against clients of B and another court (with its unique set of laws) which had jurisdiction over complaints made by clients of B against clients of A. Also, there could be a third court (with its unique laws) with jurisdiction over complaints by clients of A against other clients of A. Indeed, it should be clear that there could, in principle, be two courts for each pair of defense agencies plus a court for each agency which would be devoted to intra-agency disputes.