Since long before Rousseau put it so pithily, political theorists have thought that their task is to discuss 'men as they are and institutions as they might be.' This dualism creates a tension within every normative theory about politics. Which facts about the world are to be taken as inescapable givens, and which as correctable accidents or reformable mistakes? Which social problems are to be accommodated and which are to be overcome? The division between ideal and non-ideal theory does little to answer such questions. John Rawls, for instance, takes as given certain facts about modern individuals—that they possess limited altruism toward one another, for example—and certain facts about modern societies, such as the permanence of reasonable disagreement about religious and other fundamental questions of the good life. Yet given these constraints he sets about constructing a theory which assumes away a variety of other complications, such as the possibility of a culturally heterogeneous society, migration of persons or capital, and permanent disagreement about the right as well as of the good. Such complications he leaves for later, 'non-ideal' stages of theorizing. A cursory look at the world around us, however, suggests that ethnic and cultural pluralism are as inescapable as religious pluralism, that disagreements about justice are as prevalent as disagreements about the good life. Why are some of these taken into account in an 'ideal theory' and others not?
I say this not to criticize Rawls but to illustrate the problem at hand, and to suggest that it cannot be evaded by labeling a theory 'ideal' or 'non-ideal.' Any theory which is not foolishly utopian must assume that there are some human limitations and some aspects of the human condition which cannot be overcome, at least not in the foreseeable future. Any theory which is to have any normative bite must suppose