"In this case of unrestricted income distribution, the dimensionality of the issue space is the same as the number of individuals. Thus, as Tullock argues, political resolution of distributional issues is apt to be possible only if only a few parameters of the income distribution are under consideration, not the whole distribution. Why this restriction of the scope of choice should occur is not easy to explain on simple economic grounds. On the other hand, the restriction does conform to the long-standing view of writers on ethics, of whom Kant is perhaps most conspicuous, that decisions on distribution ought to be made as if by an impartial observer, who considers then only the mean, a measure of inequality, and perhaps one or two further parameters characterizing the income distribution, but not specifically who gets what. If voters acted like Kantian judges, they might still differ, but the chances of coming to an agreement by majority decision would be much greater than if voters consulted egoistic values only. Does this suggest that ethics may have survival value for political systems and therefore descriptive as well as prescriptive significance?"