Apportionment of representatives is a basic rule of everyday politics. By definition, this basic rule is a constitutional stage problem and should be decided behind the veil of uncertainty. To bring apportionment closer to quotas, we introduce f‐divergence for utilitarianism and Bregman divergence for consistent optimization. Even in our less restricted condition, we find that we must use α‐divergence for optimization and show that the minimization of α‐divergence induces the same divisor methods that correspond to the maximization of the Kolm–Atkinson social welfare function (or the expected utility function), which is bounded by constant relative risk aversion.