Reply to Jimbo on Anarchy


[This was written in response to a post by James ("Jimbo") Wales on Humanities.philosophy.objectivism, in which he argued that the sort of anarcho-capitalist legal system that I described in part III of The Machinery of Freedom would be unstable. His example involved a cartel that used force to keep out competitors. He argued, for reasons that should become clear in the following, that the cartel would succeed in getting its anti-competitive legal rules accepted by the institutions I described. Interested readers should be able to locate his original post, made in about December of 1995, using one of the Web search engines. What is posted here is a slightly edited version of what I posted in response]


I would like to thank Jimbo for making a serious effort to understand my ideas before criticising them. While I think his conclusion is wrong he has, as others in the thread have not, actually engaged the issues, and pointed the way to a real flaw in anarcho-capitalist institutions--although not, I think, a lethal flaw.

In dealing with these arguments, I am going to assume my readers are (as I believe Jimbo is) familiar with conventional economic ideas such as "efficiency" and "market failure." I realize that that assumption is false for many on this news group, but getting anywhere with this discussion is going to be hard enough without my having to explain the standard terminology of the economic theory in terms of which both of us are arguing.


That said, let me note some problems with his essay:

1. Jimbo writes: "Imagine a society set up according to Friedman's vision. ... There are many "defense agencies" all competing for customers. Each defense agency has its own set of laws although for convenience there may be some overlap or commonality between them."

This is wrong. Under my vision, as I described it in the book he is citing, defense agencies do not have laws. Defense agencies have contracts with each other specifying what private court (aka arbitration agency) will arbitrate disputes between their customers; arbitration agencies produce law codes. Thus the legal rules applying to a dispute are a function of the pair of agencies whose customers are involved.


2. Jimbo writes: "Now envision a court which decides to concentrate in protecting firms within a specific industry."

Again, that does not make sense, both because it confuses the function of courts with the function of defense agencies and because the court is being determined by a preexisting agreement between representatives of both parties to the dispute it is dealing with.

One might have a court that specialized in dealing with disputes between firms in a specific industry, but I don't think that is what Jimbo is describing. One might also have a court that specialized in dealing with disputes between firms in a specific industry and other people--but it would have to be acceptable to the defense agencies representing those other people, so describing it as concentating in "protecting" firms within the industry is misleading. Its speciality would be settling disputes involving such firms.


3. Now we come to Jimbo's main argument. I suspect from what he has just posted that he missed my response to it when he first posted it, quite a long time ago on a different group.

Simply stated, I believe that his particular example does not work, for a reason I will explain in a moment. It can, however, be modified into a version that would work, at which point it is a special case of a type of market failure in the market for law that I first described in print some years ago, in a footnote to my article "Law as a Private Good."

In discussing Jimbo's example, I am taking the liberty of modifying it slightly to correct the errors (with regard to my proposed institutions, not the economics of anarchy) mentioned above. The firms in the gasoline cartel are customers of a defense agency which specializes in serving them. That agency negotiates legal rules (in the form of agreements on arbitration agencies) with other agencies representing other people, including both potential new entrants and customers.


4. What is wrong with Jimbo's argument as given is that it assumes that I (or my representative) must choose between the legal rules "existing gas companies can shut down new competitors" and "existing gas companies cannot shut down new competitors." If that were the choice, his argument would be correct. Insisting on the latter rule would produce a benefit for all consumers larger than its cost for the would-be cartel, but I would get only a small part of that benefit while the cartel would pay all of the cost, so in negotiations over that rule the cartel would outbid me.

But I don't care whether other people are free to trade with new gas companies; what I care about is whether I am. The legal rules my defense agency will consider in negotiating with the defense agency representing the cartel are "our customers are free to buy from any producer of gasoline--and can sue and collect from anyone who stops them doing so" vs "our customers have no such right." By getting the former rule they do not produce a public good, since the cartel is still free, so far as they are concerned, to use force or the threat of force to keep other people (i.e. customers of other defense agencies) from buying from non-members of the cartel. The gain to me of being free to buy from any seller is larger than the loss to the cartel of my being free to buy from any seller, for familiar (monopoly deadweight) reasons. So my agency will be willing to spend more getting the competitive outcome for its customers than the cartel's agency will be willing to spend getting the monopoly outcome for its customers. The same will be true for all other agencies. So the equilibrium is competition, not a cartel.


5. Why is it my rule, rather than Jimbo's, that my agency will be negotiating over? For precisely the same reason that ordinary firms try, so far as practical, to produce private goods. A grocery store could conceivably do business by saying to each customer "If you want apples, pay us fifty dollars and we will put out a barrel of apples that you and everyone else can take apples from." But grocery stores, for obvious reasons, find it more profitable to offer a deal in which only the person who pays for the apples gets them. Similarly here. The defense agency wants to produce the maximum benefit for its customer at the minimum cost, and it does so by arranging, so far as possible, to negotiate agreements that provide their benefit only to its customers, not to the customers of its competitors as well. Hence my version of the choice of rules, not Jimbo's.


6. One can modify Jimbo's example, however, to make it work; I will leave that as an exercise for other people interested in thinking through these problems, and go on to discuss the general form of market failure which lies behind the problem, and some examples that I think more realistic than the cartel one.


7. The basic argument for the efficiency of the law generated by my system is that any pair of agencies, in bargaining over the law obtaining in future disputes between their customers, will be trying to maximize the summed benefit to those customers in order to maximize the profit available to be divided between them for the service they are providing. They will therefore look for efficient laws--laws that maximize the size of the pie. The argument is related to standard arguments for market efficiency although, due to the peculiar nature of the market for legal assent, it ends up relying more on Coase than on Marshall (for details see my chapter in For and Against the State.

But implicit in this argument is the assumption that the legal rule obtaining between A and B only affects them. Suppose we have a case where the legal rule between A and B has significant net effects on C, D, etc. If C, D, ... are not customers of the same agency as A and B, the effects on them will be ignored in A and B's agencies' calculation of the optimal legal rules for their customers. The result, as with other externalities, is inefficient.

My standard example of this effect is intellectual property law. My agency and Jimbo's are negotiating on whether he should be legally bound to respect my copyrights. What happens?

Suppose we start with a regime where he is not obligated to respect my copyrights, and imagine moving to a regime where he is. One cost is that he has to pay me royalties to use my books, computer programs, or whatever. That is exactly balanced by a benefit--I collect those royalties. A second cost is that he uses an inefficiently low amount of my intellectual property (assuming I can't manage perfect discriminatory pricing), since he is paying a cost for using it greater than the (zero) marginal cost of having one more person use the existing property. A third cost is the transaction and monitoring cost associated with enforcing my copyright, negotiating licenses, etc. So far it looks like net costs.

To balance these net costs there is a benefit--if I have enforceable property rights in my intellectual property, that gives me an incentive to produce more of it. Jimbo can't benefit by reading my book or using my program if I don't produce them--and in a world without copyright I may not (I am ignoring, and will continue to ignore, the interesting question of how important this effect is, since it is not relevant to the point I am making with the example--I realize that some people think copyright and patent are a bad idea).

This might be an important benefit if the question was "does everyone respect my copyright or does nobody respect it?" But that is not the question we are negotiation. Our issue is "does Jimbo respect it or doesn't he." Jimbo is a tiny part of my market, so the effect of royalties from him on my output is going to be tiny, and he will receive only a tiny fraction of the benefit from that tiny effect--"of the second order of smalls" as an earlier generation of scientific writers used to put it. So we can neglect the effect on output in our negotiations. So net costs to us of introducing intellectual property rights between us are larger than net benefits, so we don't introduce them--even if, including effects on everyone, net benefits of doing so would be larger than net costs. So the market for law produces an inefficiently low level of intellectual property protection. The same argument will apply wherever the legal rule between A and B imposes significant net externalities.

This problem is ameliorated somewhat by the fact that the negotiation is actually by my agency and Jimbo's agency, each of which will take account of effects on all of their customers, not just me and Jimbo. But as long as there are multiple agencies, this only reduces the inefficiency, it does not eliminate it. In principle it could be eliminated by multilateral negotiations involving all agencies, as per the Coase Theorem, but I do not believe that is a plausible scenario in a world of many agencies.

Why do I not regard this as a good reason to abandon my support for anarcho-capitalism? For the same reason that I do not regard the existence of market failures on ordinary markets as a good reason to abandon my support for capitalism. The market solution, for producing both law and goods, is imperfect--but I think it is probably better than any alternative.

This is already a long post, so I will end it by suggesting that people think through the analogous set of problems under ordinary political institutions. Legal rules there too are produced on a sort of a market, with political actors using votes, money, and other forms of support to bid for the outcomes they want. But on the political market, all legal rules are public goods, since what I am "buying" is not a legal rule for me but a legal rule to apply to everyone. The resulting set of inefficiencies have been worked through in some detail in a classic article by Becker; a less technical version can be found in the chapter on the political market ("Law and Sausage") in my new Hidden Order .

Simply stated, the argument for capitalism over socialism, in the market for law as in other markets, is that market failure is the exception on the private market but the rule on the public market.

Jimbo tries to deal with this sort of argument by writing:

"A good government can have built-in mechanisms to help make such schemes difficult-to-impossible.

A good constitution would provide for clear prohibitions against regulations designed to unjustly cartelize commerce. It would provide for unambiguous criminal penalties for the use of force to gain economic values. "


But that is like someone who argues for socialism explaining that of course a good socialist auto industry will produce the kinds of cars that people want in the most efficient fashion. Jimbo is describing an outcome, not a mechanism for producing it. If he applies the same ingenuity to thinking through the logic of the political market he wants that he applies to the logic of the economic market I want, and makes the same assumptions about human motivations in both, I think he will eventually conclude that his outcome is not a likely result of his institutions. Rand is not available to function as a deus ex machina outside the system, forcing "her" government to respect rights, pass good laws, etc.


David Friedman

In fn 2 to his essay, Jimbo writes:


"One glaring example is the way Friedman's terminology tends to prejudice the argument against the possibility of there existing a moral and proper government by means of *defining* government in such a way that government necessarily involves "rights violations" and being a kind of criminal gang."


That is not correct for two reasons, both of which I pointed out during the long dispute Jimbo and I waged some time back. I define a government as an agency of legitimized coercion. I define coercion as acts that would be regarded as rights violations if done by ordinary parties. I define "legitimized" as meaning that people in the society do not react to such acts when done by government in the ways in which people normally react to such acts (for details see my book)--they accept the acts in practice, whether or not they approve of them as a matter of moral theory.

The first, and less interesting, reason why my definition does not "preclude the possibility of a non-rights-violating government" is that the definition is in terms of what acts are regarded as rights violation by the population the government is in, not what acts are actually rights violations--so you might have something that met my definition not because it violated rights but because it violated what people in its society falsely believed to be rights. This point is not very important as a critique of Jimbo's claim, but I think it is important as a matter of clearly thinking through what my definition of government is--a positive, not a normative, definition.

The second, and more serious, reason why my definition does not assume that government involves rights violations is that I specify (I can't quote from the book, since I am travelling and don't have it to hand) that coercion means acts that would be considered rights violations if done by private individuals. I thus deliberately leave open the possibility that such acts might be not only legitimized but legitimate when done by government--in other words, that what I am calling "coercion" may not be right violating when done by government. That is in fact Jimbo's own position, as I understand it--that government can legitimately do things which would be rights violations if done by private parties.

It is true, of course, that I don't believe government is legitimate--as I made fairly obvious when I wrote that "Government is distinguished from other criminal gangs by being legitimized." But that was not part of my definition of government. And I believe that if Jimbo checks the chapter (or possibly one of the next few), he will find explicit mention of the possibility that government might be legitimate, along with a brief comment (I think citing Spooner's argument) to the effect that I don't myself believe it is.


David Friedman