Three referees support acceptance for the contributions. I agree with the reviewers that the analysis can help understand algorithms in game. Although there is a gap between theory and practice due to mean-field approximation, it may be a first step towards analyzing the difficult problem. Thus I recommend acceptance.