Incentives for Science

Abstract

The academic peer review system is central to scientific progress, but it has well-known flaws. Symptoms are substantial delays in getting new research published and, most seriously, the uncertain quality of published works due to non-transparent refereeing criteria and effort, as well as poor incentives to replicate. This paper reviews empirical indicators of these issues and conceptually characterizes an efficient decentralized market for refereeing. The challenge is not merely to balance the demand and supply in refereeing services. Efficiency entails, statically, that (1) the right works and properties of those works are reviewed, (2) the right experts perform the review, and (3) the right effort is expended on the review. Furthermore, efficiency entails, dynamically, that the system evolves toward cheaper and more accurate reviews. We introduce an anonymous consensus-based protocol, combined with a Vickrey-style auction, that can satisfy these demands.