Paul Birch's Webpage Home

Mutual Defense and The Free Rider
by John W. Goes
1/16/2003

[Please read Paul Birch's Mutual Defence before reading this article.]

It's tough being an anarchist. People roll their eyes at us, shake their heads in disbelief, and — possibly worst of all — they occasionally ask tough questions. In my opinion, the toughest of tough questions is undoubtedly the issue of national defense in anarchy. What's the incentive for people to pay for national defense without a state? Paul Birch, in his article Mutual Defence claims to have a solution to this problem.

Birch explains the free rider problem thusly:

No single person can expect to have much influence on the effectiveness of the defence of his country. Whether or not he contributes his personal share of the defence costs will almost certainly make no difference; the country will survive without him, or be defeated either way. If he doesn't pay, but most other people do, he gets a free ride. He gets to be defended for nothing. If he does pay, but most other people don't, the country goes down the tubes and he wastes his money. Unless he can somehow be confident that the others will pay up if and only if he also pays, it is not in his self-interest to do so. This is the classic public good trap.

Many attempts to solve this problem have been proposed, but in my opinion all have fallen short of a solution. I agree with Birch and others that a system of insurance companies would likely fail to provide national defense. The Private Defense Agency solution, while potentially being able to resolve small scale disputes, does not seem to have any incentive to supply national defense either. I contend that Birch's new system, while potentially being able to meet the free rider problem, probably won't.

MUTUAL DEFENSE

Birch describes his system as follows:

A mutual defence association comprises a group of persons who mutually pledge to defend each others' rights and who organise themselves accordingly. The association is jointly owned by its members. It may perform all the standard functions of an anarcho-capitalist protection agency, where appropriate subcontracting them to other agencies, especially the arbitration and court services. It can be of any convenient size and may freely enter into agreements with other like associations. It may also hire specialist staff. It stands ready to use force, where necessary, to come to the assistance of its members in emergencies; to enforce judgements against convicted offenders, both members and non-members; and to defend the association against organised opposition, rogue agencies or foreign invasion.

Birch suggests that MDAs will form in layers. Let us assume for this analysis that MDAs form territorially. For example, take three towns: Blueville, Redlawn, and Yellowpark; all of which are neigbors. They each have a 100,000, 60,000, and 40,000 people respectively. In the event that the larger Blueville is belligerent, or even as a precautionary measure, Redlawn's MDA and Yellowpark's MDA would benefit by forming a second level MDA. Birch suggests that this layering will continue until the public good of national defense has been internalized, thus solving the free rider problem.

The purpose of this essay is two-fold. I intend to further elaborate upon the effects of Birch's system, finally contending that it has the potential to solve the freerider problem, but probably will not.1 Let us begin by seeing how Birch's system would develop in a State free world.

THE EVOLUTION OF MUTUAL DEFENSE

For the rest of the article, I will assume that all first level MDAs will be essentially ultraminimal states to simplify my analysis. If anything, this stipulation is more realistic, as a system of ultraminimal states avoids challenging problems found within pure market anarchism. I will also assume, borrowing a term from Birch, that my following description of Birch's system is a dynamic system. That is, I'm explaining the evolution of the system to explain why the structure is stable once in place. I'm not suggesting, and I don't think Birch has either, that his system would naturally evolve today if the state were suddenly removed.

Imagine an area roughly the size of England. By stipulation, all of these towns are first level MDAs. These MDAs will be responsible for trying criminals, patrolling the area, preventing crime, ect. Possibly, and likely, they will also assume responsibility over other public goods such as fire protection and road maintenence. It would be in the interest of neighboring MDAs to make agreements specifying procedures for settling potential disagreements. Assuming these MDAs are generally equal in size, the risk of first level MDAs fighting "mini-wars" would be extremely small.

There are several reasons why MDAs will tend not to act belligerently. The primary reason is that the owners of the MDA bear the full costs of any potential conflict. Especially at the lower levels, each potential lost life is very valuable. Unlike today's system where politicians send strangers into battle, the heads of the MDAs would be sending family and friends into battle. When you, your family, and your friends risk death, one is much less likely to initiate a conflict. MDAs don't have a means of externalizing the costs of war onto others; this power can only be exercised by a state. The same applies to monetary losses. Besides the fact that destroying others can be expensive, getting destroyed is even more expensive. It is likely that a tit-for-tat strategy will form that will tend to discourage belligerents from forming. The belligerent MDA risks being attacked, embargoed, or even ostracized from various social groups. Waging conflict will also interfere with commerce and the careers of the belligerent populations. People tend not to shop in war zones as well, and going to war puts that MDA members at risk of being fired, or at least will prevent them from making money for a given period of time; further incentive to abstain from conflict.

But despite all of the above, there remains a possibility that an MDA could become belligerent. Perhaps it feels it can defeat certain towns without causalties, or perhaps it would attack purely based on hate; regardless, it is a possibility against which neighboring MDAs will likely want to prepare. Neighboring MDAs will tend to form alliances discouraging other MDAs from attacking. Thus, we can see that Birch's system will almost certainly create higher-level MDAs. So, a first level MDA will likely join a second level MDA (or an MDA of an MDA), a second level MDA a third, and so on to higher levels.

THE OPTIMAL LEVEL

It is important to realize that each level is less likely to form than the one below it. To show why this is so, we must explore the nature of the higher level MDAs. Each higher level MDA will have its own procedures for judging when to assert its collective authority. I foresee two general strategies: centralization and decentralization.

Centralization in this case is the strategy of "an attack on one is an attack on all." For example, if there is a 7-level MDA system (meaning the top layer is an MDA of a group of 6th level MDAs, and so on), and a town attacks one of the membering towns, all towns within this 7-level system will come to the defense of said town.

Decentralization is the other extreme. With this strategy, the level-7 MDA would not get involved in level-1 or level-2 fighting. The use of the level-7 MDA in this case would probably be reserved for specific large-scale threats.

No system could be perfectly decentralized without eliminating the higher levels, but I do not think it is likely that centralization will be the chosen strategy for the higher level MDAs. In the case of a large network of MDAs pursuing a centralized strategy, the more belligerent will be subsidized by more peaceful, increasing the potential cost of a peaceful member. Thus, we can see that either the peaceful will not join such an MDA, or that they will join an MDA only if it internalized the costs of being belligerent. There is also the additional factor of location. The lower-level MDAs surrounded by towns within their network are safer, and thus less willing to support the conflicts of the towns on the border of another network. The tendency will be to make all large networks of MDAs more peaceful, and thus more economical. The larger the given MDA network, the less likely it would be to have a centralized policy, as well as less likely to be belligerent.

So, in a completely non-statist society it appears that Birch has solved the free rider problem.2 The externality of public defense has been turned into a mostly private good. Belligerent groups will be discouraged from forming, and adequete defense will be provided against any remaining potentially belligerent groups.

INTRODUCE A STATE

The problem arrives with the introduction of states into this analysis. Presumably, any actual implementation of Birch's system would not involve the whole world, but a portion of it. An anarchist society would coexist with other states. Whether it is likely to happen or not, many states have been known to be very belligerent, despite the economic consequences of doing so. The leaders have the ability to avoid the full financial costs of war, as well as the ability to compel their subjects to fight. Some have claimed that states would actively target an anarchist society in fear of its succcess. Whatever the likelihood, we have to consider it a possibility that a State would actively seek to destroy an anarchist society. Does Birch's system internalize this externality?

The answer to this seems to be: not necessarily. There will be two optimal levels for MDA growth. One is the level demanded by threats within the meta-structure (or lack thereof) of Birch's system. This we explored and concluded that the system would meet all such threats. The second level will be the real level, or the threats within Birch's metastructure plus all other threats, namely from states. In order to have solved the free rider problem, Birch's system needs to make jumping from the "internal"-level to the real level a rational economic decision.

I think it is clear from the start that it isn't necessary in a strict sense. It is possible that the "jump" cannot be made. But this notwithstanding, can a case be made that Birch's system will probably solve the free rider problem? I'm not sure that it can. The problem is that we are now in the arena of speculation. It is very difficult to show with a high degree of apriority that Birch's system will probably solve the problem; we can only attempt an educated guess. I think this is an issue that demands more thought and discussion, but I will attempt my own answer here.

In order for Birch's MDA structure to solve the problem of national defense, the high level MDAs would have to be a considerable size. If the highest level MDA networks demanded within Birch's system are too small, they will suffer another public goods trap. There is reason to be skeptical that MDA networks would ever reach such the necessary size to make the jump to national defense a rational one. As we concluded earlier, there are strong disincentives for the formation of belligerent MDAs. The larger a given MDA network is, the more incentive it has to be peaceful and decentralized. Most people will join the cheaper and safer MDA network, causing the wasteful and dangerous MDA networks to fade out of existence. This implies that larger MDA networks will have increasingly smaller threats to worry about, severely limiting the need for the huge MDA networks needed to make a rational jump to national defense rational. After a few layers of MDAs, it is hard to conceive of a rational reason to form more alliances. With little or no threats from higher-level MDAs, there will be little need to build enormous MDA networks to combat extremely improbable threats. The burden of proof will be on those who claim MDAs will grow to very strong sizes; at this point there seems little reason to believe this would happen.

Birch has elsewhere suggested that the above problem is solvable. The jumps might be made by a series of small jumps he seems to suggest. Because states will attack the weaker areas first, all MDA networks will have an immediate incentive to increase their size above their neighbors, to make an invasion a less attractive proposition to statist neighbors. Thus, no one will want to be the most easily defeatable MDA, and the MDAs will continue to build their networks until they have effectively supplied national defense for everyone in the anarchist society. No one has an incentive to drop out of the alliance, because they are at higher risk of invasion if they drop out of their middle level defense agencies.

The problem is that this suggestion is grounded on a dubious assumption: that all other things being equal a state will tend to attack the weakest part of an anarchist society. If this assumption was correct, then presumably national defense is a public good from the very beginning, and national defense would be supplied. The unfortunate truth is that any state who used the above strategy would be extremely foolish.

There will be two general motives to attack an anarchist society. The first is for plunder. The regions with the most valuable resources might be targeted by a state. It will tend not to be in the interest of neighboring regions that lack such resources to assist their neighbors. The second general motive is to destroy the anarchist society to prevent its influence from spreading. There is no reason to suppose that the dominant strategy will be to attack the weaker regions first. The opposite strategy seems much more effective. If a state wanted to destroy an anarchist society, it could merely announce that the strongest MDAs would be targeted first, while the small ones will be left alone. The result will be that MDAs will not have incentive to make "little" jumps toward national defense. Lower level MDAs will have increased incentive to drop out of the higher levels, and will certainly not make the intermediary jumps necessary to supply national defense.

Unless the highest level MDAs demanded by Birch's metastructure are a sufficient size, which we have above shown reason to doubt, Birch's system will not be able to solve the free rider problem.

CONCLUSION

Although I am not satisfied with Birch's claim that he solved the free rider problem, I don't think it can be denied that his system has been the most successful attempt to meet it. Furthermore, the problems that I have discussed are not insurmountable. As many others have suggested, the free rider problem doesn't seem to be devasting. Especially in the earlier, more vulnerable stages of an anarchist society, people would probably be willing to create "collectively rational" MDAs. Fear and collective pressure would probably win out. I also think that the foreign threat is somewhat exaggerated. While I assent to the possibility, I really don't think it is a realistic scenario. An anarchist society as described by Birch would not by nature be hostile towards states, making any attack upon it appear to be an obvious injustice. Plus, the usually cited motivation to destroy this society suffers yet another free rider problem. While all states might wish destruction to the society, there doesn't seem to be much incentive to be the one to initiate war just for the sake of ending the experiment in anarchy.

It is my opinion that Mutual Defense is the most effective defense system yet proposed. I hope this elaboration and critique of it will give Mutual Defense the attention and scrutiny it deserves.

  1. Bob Murphy has also attempted a critique of Birch's Mutual Defense, but in my opinion Birch competently showed Murphy's argument to have been based on a strawman here, so I won't discuss Murphy's argument here.
  2. I have not addressed the issue of people who choose not to join MDAs because it is really not a major concern. As Birch has said in his article, "[those who don't join] have made defence even more of a private good, so we need not concern ourselves with them further."

John W. Goes currently resides in São Paulo, Brazil, although he usually lives in a suburb of Chicago, Illinois. He likes clipperships.

discuss this article in the ASC forum!

Paul Birch's Webpage Home