This paper develops an efficient multi-arm bandit method for determining the most useful nodes in a network. Interesting applications for improving fairness and possibly increasing robustness to adversarial attacks are mentioned. Reviewers provided three insightful reviews mentioning the pros (a cool combination of techniques with encouraging results worth the field hearing about) and cons (not the most novel method and more comparison with other methods would be beneficial) of paper acceptance. There is weak but unanimous support for acceptance.