Figure Skating and the Theory of Social Choice

The rule used by the United States Figure Skating Association and the International Skating Union, hereafter the ISU Rule, to aggregate individual rankings of the skaters by the judges into a final ranking, is an interesting example of a social welfare function. This rule is examined thoroughly in this paper from the perspective of the modern theory of social choice. The ISU Rule is based on four different criteria, the first being median ranks of the skaters. Although the median rank criterion is a majority principle, it is completely at odd with another majority principle introduced in this paper and called the Extended Condorcet Criterion. It may be translated as follows: If a competitor is ranked consistently ahead of another competitor by an absolute majority of judges, he should be ahead in the final ranking. Consistency here refers to the absence of a cycle in the majority relation involving these two skaters. There are actually many cycles in the data of four Olympic Games that were examined. The Kemeny rule may be used to break these cycles. This is not only consistent with the Extended Condorcet Criterion but the latter also proves useful in finding Kemeny orders over large sets of alternatives, by allowing decomposition of these orders. The ISU, the Kemeny, the Borda rankings and the ranking according to the raw marks are then compared on 24 olympic competitions. The four rankings disagree in many instances. Finally it is shown that the ISU Rule may be very sensitive to small errors on the part of the judges and that it does not escape the numerous theorems on manipulation. Some considerations are also offered as to whether the ISU Rule is more or less prone to manipulation than others.