Robert Nozick's "Anarchy, State and Utopia" is undoubtedly a highly original and intelligent work. It has three major flaws though. Firstly, his idea of a minimal state requires not only that the state itself respects individual liberty to an unparalleled degree, but also that the population uniformly respects others' liberty to a high degree. Secondly, there is no discussion of power relations and the possibility of actually achieving a minimal state; the introduction of a minimal state today would not, and perhaps could not, have the effect Nozick desires, and there is little indication of what sequence of events could lead to a minimal state which would work. Thirdly, the "entitlement theory of justice" is inherently dualistic, a distribution of holdings is either just or unjust, but to defend it against the charge that a just distribution of holdings is impossible to achieve in practice, one would probably have to introduce a variable scale of justness, which would be antithetical to the spirit of the theory.
His entitlement theory of justice says, roughly, that if the initial distribution of holdings is just and every transaction that subsequently occurs is just, then the distribution of holdings will be just at all later times. A just transaction, again roughly, is one which is mutually and voluntarily (i.e. without physical coercion) agreed upon.
The minimal state is one that serves only to provide a monopoly police and military function. The police function is to protect its citizens against violence, coercion or fraud, and to enforce contracts. The military function is obviously to protect the state against outside aggression.
Is it possible for a society to exist whose basic structures contradict the propensities of the people who make up that society? More specifically, is it possible that a minimal state could coexist with a society composed of individuals who do not respect others' liberty? It seems unlikely that it could; there would be no pressure to maintain the minimality of the state if a majority wished to infringe on the liberty of some individuals, and so a minimal state would be unstable or prone to becoming a more than minimal state. The only other way a minimal state could be achieved in a stable way would be for an external power to impose it on an essentially unwilling population, a benevolent (libertarian) dictatorship. Surely this is not a state of affairs Nozick would consider desirable?
There are two major consequences of this requirement on a society with a minimal state. Firstly, it is not as permissive as Nozick would presumably like it to be, since it imposes respect for others' liberty on the populace. Secondly, it is as utopian, in the sense that it requires an enormous change in popular consciousness, as many of the socialist systems that Nozick criticises on these grounds. If we were able to achieve a society in which people did uniformly respect others' liberty to a high degree, without doubt an incredibly desirable state of affairs, then the institutions and legislation of a less than minimal state would either not infringe upon others' liberty (because the institutions and legislation wouldn't be used in that way), or those aspects that would infringe upon them would naturally fade away anyway (because there would be no support for such institutions or legislation).
This brings us on to the second point, the lack of a discussion on how a minimal state could be achieved consistently with the principles that motivate it (mainly liberty). The major stumbling block is probably the requirement that the population uniformly respect others' liberty, and the massive change in popular consciousness that this would require.
One of the major questions that has to be answered to determine the feasibility of achieving this is whether or not is possible, in general rather than in particular cases, for an individual who considers himself or herself superior to another to respect the other's liberty to the same degree as someone he or she considers an equal. For example, if the individual has more wealth, better social status, or whatever, can they respect the liberty of individuals with less wealth, social status, etc.?
If indeed general equality is the prerequisite for a uniform respect for others' liberty then it may be that Nozick's minimal state would, to function as intended, actually require a general equality (in wealth at the very least). This would, I suspect, be much at variance with Nozick's idea of a minimal state.
Finally, Nozick's entitlement theory of justice. There are three observations that undermine the usefulness of this theory of justice. Firstly, there will always be unjust transactions to deal with. Doing so is the purpose of the police, one of the two institutions in a minimal state. Secondly, not all unjust transactions will be detected, or even if they are it will not always be possible to provide compensation (for example if the criminal is not found). Thirdly, the entitlement theory of justice is "all or nothing". Either the entire distribution of holdings is just or it is not. It follows that a single unjust transaction that is not compensated undermines the justness of the entire distribution of holdings for all time. Moreover, this state of affairs is inevitable from the first two observations. So what are we to do with a theory of justice that inevitably says that the distribution of holdings is unjust?
The obvious answer is that, for example, a small theft does not really undermine the justness of the entire distribution of holdings. However, to fix the entitlement theory of justice so that this statement is meaningful one would probably have to introduce either a variable scale of justness, or perhaps introduce the idea of a local injustice in the distribution of holdings. The second attempt at a solution to the problem would not alone be sufficient though. Once a local injustice in the distribution of holdings is introduced, it could spread through the entire system reasonably quickly (because a just transaction with someone who has gained by an unjust transaction would spread the injustice to the other party in the transaction) until the entire distribution was unjust. So, at some point the concept of a sliding scale of justness would have to be introduced. What implications does this have for the theory?
Firstly, a sliding scale of justness is antithetical to the spirit of the original theory. The purpose of the original theory was to rule out moralistic statements like "it isn't fair that some are so rich whilst others are so poor". If the initial distribution of holdings and every subsequent transaction is just then nobody can complain. Once the possibility of a variable scale of justness is introduced, there is a possibility that injustice in the distribution of holdings can accumulate. The possibility that "it isn't fair that some are so rich whilst others are so poor" is now one that cannot be avoided a priori but has to be argued empirically (perhaps an impossible task). More importantly, it is not compatible with Nozick's distaste for "patterned" conceptions of the justness of distributions of holdings. Essentially, a patterned conception of justice in holdings is one that says "From each according to X, to each according to Y" where X and Y are some criteria (e.g. X could be "their ability" and Y could be "their need"). Nozick argues against them because, according to him, they necessarily involve a constant interference in the private lives of individuals. (This is simply stated rather than argued, but that's another matter.) However, unless he could produce an argument to demonstrate that injustice in distributions would not accumulate, then it seems that some sort of interference would be necessary to stop it.
So, in conclusion, there seem to be some deep contradictions inherent in "Anarchy, State, and Utopia". The entitlement theory of justice probably cannot be fixed without introducing some form of constant interference in individuals' private transactions and the minimal state cannot survive as intended without general equality.