Harmony Index: A Scalar Index to Assess and Predict Effectiveness in Multi-Agent Teaming

Darryl Roman, Noah Ari, Johnathan Mell

Abstract: Effective teams are essential to the success of any organization. However, designing and measuring effective teams is a complex task. We present the Harmony Index, a new measure of team effectiveness that is simpler, more accurate, and more useful than prior attempts. We validate the Harmony Index in a human-subjects data set with 3.9 million entries, and show its predictive power in a separate analysis. We make recommendations for its future use in additional domains.