Utilitarians are typically taken not to have a clear theory of justice since notions that are usually taken to be important by those who want a theory of justice are regarded as fungible for utilitarians. Nozick gives an example of this when he discusses the idea of a "utilitarianism of rights". Such a form of utilitarianism would involve commitment to the notion that non-violation of rights is an ultimate goal but a consequence of this could be that a particular violation of rights in relation to a given person was itself taken to be worthwhile in principle due to the way this would safeguard rights in general. So, when a mob rushes through a town to lynch a man they think is guilty of a certain offence it might safeguard rights generally to sacrifice him.
In reply to this view Nozick develops, and is one of the first to develop, a view of deontology that works by reference to a notion of constraints. The suggestion here is that such an understanding of rights is distinct from building rights into a theory as an outcome as it was precisely building it into the outcomes that produced the paradox of the utilitarianism of rights. In response to this idea you could either indirectly promote rights rather than taking them as your direct object or you can specify them as distinct from any notion of outcome. If we take the latter view then rights are a side-constraint on any form of moral accounting we may wish to adopt.
However, whilst this argument thus far appears clear enough there are two additional steps required for it to point in the direction Nozick wishes. Firstly, the notion of rights as such a side-constraint is somewhat vacuous without a specification of the bearers of such rights. Who are they? Individuals on Nozick's view and he uses two ideas to connect the conception of individuals as bearers of rights to the notion of rights as side-constraints. The first idea is the invocation of Kant's formula of humanity understood as a reference to the specific quality of each given individual. This is further supplemented by use of John Rawls' argument that the separateness of persons has to be given due weight in moral theory.
There are a number of problems with the argument when it is presented as fully as this. Firstly, the Rawlsian link is a weak one tested against the utilitarian since there can be given arguments to the effect that the principle of the separateness of persons begs the question against utilitarianism. Secondly, the understanding of the formula of humanity as a formula intended primarily to safeguard individuals has odd aspects to it since the notion of humanity in question appears not to feature reference to any notion of particularity. Finally, even if deontological considerations are to be presented in terms of binding side-constraints on other actions it is still not as obvious as Nozick's argument suggests that rights would have this absolute standing that fitted them to fill the side-constraint in question.
Of the three possible objections I have mentioned to Nozick's argument the one that matters most in response to his overall libertarian theory is the one that indicates some problem with identifying the humanity formula as one that is concerned principally with individuals. However, this is the one that, further investigation of the humanity formula will suggest, can be most readily pressed against Nozick. This is due not just to the way the humanity formula has a generality to it that Nozick seems to miss but also to its connection with Kantian moral psychology. Expect more on this in further postings.