Abstract
Weighted scoring rules, perhaps the most common procedures for multi-candidate, single-winner elections, are defined by score vectors, of which the Borda Count is the best known example. The concept of Borda dominance identifies candidates who cannot win under any score vector; we assess its usefulness in the context of single-winner elections. We propose several families of score vectors that can accommodate any number of candidates, asking whether any of them can improve on Borda Count. Using large-scale simulation, we compare score vectors on Condorcet efficiency, or ability to find a Condorcet winner when one exists, on resilience to ballot truncation, and on ability to maximize the Copeland score using both complete and uniformly truncated ballots. Borda performs well on criteria related to Condorcet winner and Copeland score but, depending on the preference model, may not be optimal. Convex score vectors are the most resilient to ballot truncation.