Revealed @ Brown. Because the weight for a judge assignment is judge_i - pairing_i, the algorithm prefers putting middle-ranked judges in middle-ranked rooms over high ranked judges, even if the high rank judge is scratched out of the top rooms. This should be changed. Maybe use something like Gale-Shapley instead of a maximum-weight-matching?
Revealed @ Brown. Because the weight for a judge assignment is
judge_i - pairing_i
, the algorithm prefers putting middle-ranked judges in middle-ranked rooms over high ranked judges, even if the high rank judge is scratched out of the top rooms. This should be changed. Maybe use something like Gale-Shapley instead of a maximum-weight-matching?