Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
This is a study about the possibility of self-governance. We designed two versions of a step-level public good game, with or without a centralized sanctioning mechanism (CSM). In a baseline treatment participants play 14 rounds of the non-CSM game. In an automatic removal (AR) treatment participants play 7 rounds with CSM plus 7 rounds without CSM. In voted removal (VR) participants play 7 rounds with CSM followed by a voting stage to decide whether to keep CSM. All VR groups removed CSM. Contributions in AR and VR after CSM removal are dramatically higher than in the baseline. Most groups with a CSM history managed to cooperate until the last round. We do not find more cooperation in VR than in AR. Copyright Springer Science + Business Media B.V. 2007