Incentivizing Quality in ML Peer Review: A Credit System Proposal

A new position paper on arXiv (2608.14571) addresses the growing crisis in machine learning (ML) peer review. With soaring submission counts, stricter reciprocal review policies, and widespread adoption of platforms like OpenReview, the ML community faces a significant challenge: almost everyone has many unpleasant experiences with the review process. The paper argues that the lack of publication fees removes a natural deterrent to excessive submissions, exacerbating the problem. The authors propose a credit system to incentivize good reviewing and discourage bad reviewing, moving beyond simple appeals for better behavior. They assess existing mechanisms, presenting four takes on popular conference practices, and suggest two alternative design approaches. The paper aims to open a serious public discussion on what makes a review system effective and how it might be improved. The preprint was announced on arXiv with the identifier 2608.14571.