Voting power apportionments (Q2348756)

In this paper, the author presents some of the mathematical background for an apportionment method aimed at equalizing voter influence on their choice of representative. This method is based around the Banzhaf measure, a particular measure of the amount of representation afforded to a state in an apportionment, based on the classical Banzhaf power index. The paper suggests three measures of optimality for an apportionment method: pairwise optimality, the condition that transferring a representative from one state to another can only increase the discrepancy in power; minimizing global power deviation, finding the apportionment that minimizes the mean square deviation from the ``average'' apportionment; and a minimax condition that minimizes the largest discrepancy in power between two states. The paper offers a particular method (surprisingly, a classical divisor method) that achieves the first two optimality conditions for the Banzhaf measure, although not the third. The author does, however, discuss an algorithm for finding apportionments that achieve the third.