Impact Drift: Evolutionary Perspectives on Mechanism Design for Unintended Consequences

Abstract

Incentive mechanisms that aim to promote positive impact often produce second-order effects that steer collective behavior away from their designers' intentions—a phenomenon we call impact drift. Drawing on evolutionary game theory, Goodhart- and Campbell-style metric failures, and performativity studies, we argue that impact drift is a near-universal dynamic in incentive mechanisms of complex adaptive systems. We review modern impact-evaluation frameworks in Web3 (Quadratic Funding, RetroPGF, Hypercerts, ReFi, Filecoin) and analyze real instances of drift such as Sybil attacks, evaluator bias, and carbon-credit arbitrage. We then formalize why perfectly aligned mechanisms are rarely evolutionarily stable through evolutionary game theory modeling and outline practical approaches to measure and mitigate drift over time: metric diversification, evolving evaluation, costly signaling, and adaptive governance.