48 Cliff Road, Cowes, IOW, PO31 8BN, England and

1. Introduction

In an anarcho-capitalist society there is no state, and all the courts are private courts. There is no final court of appeal that all are obliged to recognise, and no uniform code of justice can be enforced. Laws are determined on the market. Everyone may choose which of the competing courts he will look to for protection; and he may alter his choice at will.

Since a just society is economically efficient, there is reason to hope that the hidden hand of the market may lead to justice, even (or especially) in the absence of the state. However, there is no guarantee that an anarcho-capitalist society will in fact be just, or indeed that an anarcho-capitalist society is even possible.

Even if a just anarcho-capitalist society were created, there is no guarantee that it would be stable. In particular, the restitution ramp described in my essay A Fatal Instability in Anarcho-Capitalism? might destroy it.

2. Description of an Anarcho-Capitalist Society

Descriptions of — or prescriptions for — anarcho-capitalist societies are regrettably sparse and incomplete. There are no genuine examples from history or current affairs; and although one can point to cases illuminating certain aspects of anarcho-capitalism, such as the private arbitrators of the present day, crucial questions are left unaddressed.

One fundamental question is whether anarcho-capitalism is even logically possible, that is, whether it is philosophically coherent. Can courts or protection agencies find ways of handling inter-jurisdictional disputes that are not ultimately equivalent to an agreement to form a cartel or de facto state, or else to victory by the strongest? If "might makes right", that is, if disputes are settled on the basis of the prior threat positions of the disputants, rather than by agreement or an approximation to justice, then, by definition, whatever else we may have, we do not have anarcho-capitalism. Similarly, if there is a recognised hierarchy of courts or mutual agreements tantamount to a federal or other state, or which create a uniform code of law or enforcement, then, by definition, we do not have an anarcho-capitalist society.

Suppose the protection agencies all agree to pass unsettled disputes between them on to a single court of appeal for judgement, and to enforce those judgements amongst themselves. Then the appeal court is sovereign. There may be nothing wrong with that — but it is not anarcho-capitalism.

Suppose the protection agencies agree a particular procedure for settling disputes, but threaten to destroy any dissenting agency that chooses not to abide by it, even though that agency may itself have done nothing wrong. Then the system is maintained by force majeure, not by voluntary transactions on the free market; it is not anarcho-capitalism.

Apologists for anarcho-capitalism often point out, correctly enough — in answer to criticism that protection agencies will find themselves involved in continual fighting — that since open conflict between protection agencies will be costly, it will be in their own interests to find better ways of settling disputes, and to agree such procedures in advance. Unfortunately, this is the point at which clear argument ends and hopeful hand-waving takes over. There will be an agreement, they say, but usually fail to show even that there can be an agreement (not everything that is desirable is possible), still less what forms that agreement might take.

An excuse sometimes offered is that since market transactions can't be predicted in detail, any attempt to determine the structure of anarcho-capitalist society in advance is misguided. Although there is an element of truth in this, it is basically a cop-out. We need to know, firstly, whether agreements between protection agencies can take an anarcho-capitalist form, and, secondly, whether they are in practice likely to (since non-anarcho-capitalist forms of agreement are also possible — as is continued disagreement).

Most of the hand-waved proposals (such as the rule that two concurring judgements are final) depend upon an assumption of universal consensus which is both exceedingly improbable and in any case incompatible with the core assumption of anarcho-capitalism, that no one has the authority to impose a procedural rule from which persons and associations are not free to opt out.

Even the most comprehensive works on the subject, such as David Friedman's seminal The Machinery of Freedom, have to the best of my knowledge not yet succeeded in adequately addressing this problem.

3. A Coherent Form of Anarcho-Capitalism?

After much thought I have devised what may be a possible coherent form for an anarcho-capitalist society. I cannot pretend that the description is complete; in particular, difficulties with higher-order appeals and the possibility of stubborn refusals to enter into or abide by agreements or to accept and obey adverse judgements may yet undercut its apparent anarcho-capitalist nature. Nor am I claiming that it is necessarily either a stable form or one most likely to evolve from a state of nature or any other prior state. However, it does appear to be a form of anarcho-capitalism, or something close to anarcho-capitalism, that a sufficiently able experimenter could in principle set up as a functioning society, then stand back from and watch develop. As such it may be considered useful as a thought experiment and "existence proof".

We begin by considering a pair of protection agencies. Each has its own system of justice for settling internal disputes (that is, for settling disputes between its own members or customers). In general, those systems will not be identical. Consequently, for external disputes (that is, disputes between customers of the one agency and customers of the other) there will need to be a separate agreed procedure acceptable to both agencies.

This inter-agency agreement could take a variety of forms. For example, the agencies could agree that such disputes be resolved by a specified external arbitration agency. Or that alternate cases be handled by each agency in turn. Or that the plaintiff's agency should have jurisdiction. Or the defendant's. Or that a procedure intermediate between the agencies' own procedures should be employed. Or that a panel of judges, one from each agency and one independent judge drawn randomly from a list, should rule jointly, by unanimous or majority verdict, or in some other way. Or by whatever other means and procedures they might negotiate between them. Sometimes, if one agency was large and well-trusted, and the other small and local, it might be agreed that inter-jurisdictional cases should be referred to the courts of the larger agency (but this could not be the universal rule without destroying the anarcho-capitalist nature of the society). Agreement would also be necessary on questions of appeals, permissible punishments and enforcement of judgements, and again a variety of solutions might be chosen.

Now consider another pair of protection agencies. They too will need an inter-agency agreement. But that agreement need not be the same as the agreement made by the first pair, and in general will not be. They may differ slightly, or markedly, or have almost nothing in common.

Now take one agency from the first pair and one from the second. These two will also need a bilateral inter-agency agreement. But that agreement need not be the same as the agreement made by the first pair, or by the second pair, and in general will not be. A single agency may have many different agreements with different agencies.

In general, each agency will have as many distinct agreements as there are other agencies. In all, in an anarcho-capitalist regime of n agencies there will exist n(n-1)/2 distinct bilateral agreements. Although some of those agreements might happen to be identically worded, and although some of the agencies might happen to have identical law codes, the pairwise pattern of agreements cannot be collapsed into a single uniform or dominant general agreement, and thus does not constitute a state. The fundamental features of an anarcho-capitalist regime remain intact.

4. Variations on the Theme

It may be that some of the agencies will be insurance companies relying upon external courts or arbitration and enforcement agencies. This makes no real difference to the system described. Each insurance company will have its own preferred court or procedure for internal claims, which will in general differ from company to company; each will need bilateral inter-company agreements determining how external claims will be judged and settled. Where there is a mix of insurance companies and protection agencies they will need similar bilateral agreements between them.

It may be that plaintiffs, who may be individual members of the public, will prefer to place their complaints before their chosen court directly, rather than through the medium of their insurance companies or protection agencies. Where the defendant agrees to accept the jurisdiction of that court, the court will make judgement according to its own laws and procedures (which will in general differ from court to court). However, the defendant may call his own protection agency or insurance company to his defence, or bring counter-suit in his own court of choice. Then bilateral agreement between the courts, or between the court and the insurance company or protection agency, or between plaintiff and defendant, will be necessary in order to determine how the case may be heard.

Thus, for the purposes of this analysis, the agency in need of entering into bilateral agreements for the settlement of external disputes may be a protection agency, an insurance company, an arbitration agency or a full court, or even an individual person. Any form of agency may make agreement with any other.

Difficulties may arise where disputants are customers or members of more than one agency, and seek to use both. It may then be unclear whose rules apply. Within the constraints of anarcho-capitalism it is not possible to prohibit such multiple claims; however, agencies, and pairs of agencies, may have their own rules for dealing with them. In general, we should expect the existence of concurrent or prior claims to be taken into account within the court proceedings. However, a simple "double jeopardy" rule would not be appropriate, since a case may well be brought before a second court, or through another agency, as a means of appeal against the perceived inadequacies or injustices of the first.

5. Reducing the Burden

For an anarcho-capitalist regime with a reasonable number of competing agencies (Friedman suggests around 10,000 for a territory comparable to that of the USA), the number of bilateral agreements required, O(n2), would quickly become unwieldy. Can we cut down on the amount of negotiation, without losing the essential character of the regime?

To some degree, certainly, once it is understood that most of those agencies are likely to transact most of their business within a single locality, within which effective competition may be limited to perhaps half a dozen agencies, such that bilateral negotiated agreements between them would remain practicable. Further afield, agencies might specify a set of acceptable standard forms from one or more published lists, which could be matched against the equivalent set for each other agency in turn, obviating the need for individual negotiation of the details of every pairwise agreement. Moreover, at some risk to peaceful relations, agencies might choose to forgo the making of agreements with distant agencies unless and until their respective customers came into actual dispute.

Another approach, less securely anarcho-capitalist, would be for agencies to make agreement with a limited number of meta-agencies, granting them jurisdiction over disputes with other agencies where no direct bilateral arrangement exists but which have comparable agreements with the same meta-agency. Further bilateral agreements between meta-agencies could extend such jurisdiction to agencies with agreements with different meta-agencies. Meta-meta-agencies, and so forth, could if necessary be created by recursive application of the same procedure.

However, we should observe that such methods severely constrain the freedom of agencies to make agreements compatible with their own promises and procedures — agreements are forced into more and more standardised forms, weakening anarcho-capitalism's defining principle of free competitive choice. At some point along this road we would have to say that the regime had ceased to be truly anarcho-capitalist and had begun to partake more of the nature of an imperial or federal state, or of a regime with significant statelike qualities.

Pure anarcho-capitalism, therefore, obtains when agencies enter into pure bilateral agreements only. Some derogation from pure anarcho-capitalism may however be acceptable without substantially changing anarcho-capitalism's essential character.

6. Is Bilateral Anarcho-Capitalism Stable?

The previous sections have described a form of anarcho-capitalism that appears to be philosophically coherent (at least to the depth that the analysis has so far penetrated) and which could in theory be brought into being. The question we must now address is whether such a regime would be stable.

In a previous essay A Fatal Instability in Anarcho-Capitalism? I examined the problem of the restitution ratio, the tendency of competing courts to offer ever more advantageous terms to the plaintiff in return for his custom. I found that the likely result of such a restitution ramp was the rapid destruction of the anarcho-capitalist regime.

However, in that essay I was considering the behaviour of courts as single, albeit composite, entities; it is now necessary to ask whether the system of pairwise inter-agency agreements makes any difference to the dynamics. I conclude that it does; the restitution ramp instability is weakened; but not enough to prevent failure.

First note that for internal disputes, the previous analysis holds; each agency may compete in offering higher rates of restitution to its customers. For external disputes, the possibilities for offering improved rates depend upon the terms of the bilateral agreements (bearing in mind the additional possibility of renegotiation of those terms). In general, such agreements will approximate to some sort of average or intermediate position; neither agency will get all it wants, but each will move some way towards the other. Unless agreement breaks down completely, customers of the agency offering super-restitution will gain from this, even against offenders from another agency, albeit to a lesser extent than would be true for purely internal claims. There is then obvious pressure upon the other agency to follow suit and offer super-restitution itself — and the ramp takes off.

Not all forms of bilateral agreement will be equally susceptible to the hazards of the restitution ramp and indeed some may be highly resistant (for example, where it is the defendant who chooses the court). However, such agreements will inevitably prove less satisfactory to plaintiffs, who will prefer to switch their custom to those agencies not bound by such restrictive agreements, or else to put pressure on their agencies to renegotiate. Those agencies that champion the cause of the convicted offender will not be popular with plaintiffs; but they will be popular with criminals.

One might wonder whether the regime might not separate into several kinds of agency; one biased towards the plaintiff, the other to the defendant, with the cross-faction agreements forced into a rough approximation to neutrality. But this would be to assume an unreal symmetry between plaintiffs and defendants, or between criminals and the victims of crime. Far more people are likely to choose their courts or protection agencies on the basis of expecting to be plaintiffs or successful defendants than of expecting to be convicted — interestingly, if the opposite were the case, the restitution ramp would still happen, but in reverse, with the courts competing to offer exceptionally lenient sentences to criminals!

Those courts or agencies seen to be offering criminals an easy ride will come under considerable pressure to conform. Indeed, once it becomes apparent that a substantial or dominant proportion of their subscribers are drawn from the criminal classes, it is likely that they will be declared "rogue" agencies, threatened with violent suppression unless they submit to the general trend — even if their alleged criminal bias is nothing more than a dedication to strict restitutional justice. Clearly, such threats violate the fundamental principles of anarcho-capitalism, but we cannot thereby assume that they will not occur or that they will not succeed.

We should note further that super-restitution need not be uniform. In particular, it is likely that agencies will seek to provide higher rates of compensation for criminal offences than for accidental damages; ordinary third-party liabilities, which will normally be covered by the customer's insurance policies, will as far as possible be excluded from the ramp. Agencies will try to ensure that their honest customers do not face the burden of heavy punishments or excessive insurance premiums. This somewhat weakens the scope and the incentives for super-restitution, but at the same time it weakens the motives for rejecting it.

In summary, I would argue that although the restitution ramp instability may be less strong than I originally believed, it is nonetheless powerful enough to destroy this form of bilateral anarcho-capitalism within a short space of time. I see no practicable way of specifying or manipulating the pairwise agreements sufficiently to stabilise the system, short of imposing a rule incompatible with strict anarcho-capitalism and in effect reconstituting a state.

In another essay, Anarcho-Capitalism Dissolves into City States, I similarly addressed the growth of territoriality in an anarcho-capitalist regime, driven by the higher costs of inter-jurisdictional or external disputes over internal disputes. Although the system of bilateral agreements provides a means of settling external disputes less costly than going to war, it remains true that external disputes are messier and more expensive to deal with than internal disputes, and thus territorial agencies, with a higher proportion of internal to external disputes, retain the commercial advantage, and so will be able to poach customers from the rest, enhancing their cost advantage even further. Consequently the probable breakdown of anarcho-capitalism into territorial city states through this and other similar mechanisms remains as previously described.

7. Conclusions

In this essay I have described a form of anarcho-capitalism that is philosophically coherent and in principle capable of being brought into existence. This is an important step forward, in my opinion, because hitherto there was no guarantee that anarcho-capitalism as a complete and consistent system compatible with its own fundamental principles was even logically conceivable; previous descriptions fell a long way short; and even now it must be admitted that there remain parts of the theory where rigorous proof is still lacking.

A free-market regime characterised by a pattern of bilateral agreements between competing agencies fulfils all the requirements of an anarcho-capitalist society. Pure bilateral anarcho-capitalism thus provides a convenient ideal for the purposes of philosophical, ethical, political, logical and mathematical analysis, and may also form the basis of practical experiment.

However, I have to reiterate that I do not believe such a regime would in fact be stable; my analysis seems to show that it would remain susceptible to devastation by the restitution ramp, and that it would also be unstable with respect to decomposition into territorial city states. No obvious means of stabilising it have been found, other than through techniques that would negate its basic anarcho-capitalist nature.

Thus my overall conclusion has to be that pure anarcho-capitalism at least is probably sufficiently unstable as to be in practical terms impossible to maintain, although — contrary to my earlier suspicions — anarcho-capitalism as a concept now seems logically tenable, in the sense of being fully describable and philosophically coherent.

© Paul Birch, 2nd May 2002.