Patterns · 1 author, 2 centres
This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.
This perspective piece describes two core problems in metric design—specifying measures that align with goals and minimizing perverse incentives—and proposes a structured process for building less-flawed metrics using strategies such as secrecy, randomization, diversification, and post hoc specification.
This narrative review by Manheim addresses the design of measurement and incentive systems across academia, public policy, medicine, and business. The paper identifies two interrelated problems in metric design: specifying evaluable metrics that correspond to goals, and minimizing perverse effects that undermine metrics or enable gaming. The author discusses design considerations and processes, including strategies for mitigating perverse impacts such as secrecy, randomization, diversification, and post hoc specification. The paper presents important desiderata and tradeoffs with examples of how these strategies can complement or differ from each other. The conclusion emphasizes that forethought and consideration of problems with metrics is necessary to ensure valid measurement and non-harmful evaluation. The paper frames metric design as an engineering problem involving both science and art, acknowledging that following the suggested guidelines will not make metrics impossible to exploit but represents a vast improvement over commonly used unconsidered metrics.
