Proxy Optimization in Complex Systems: Why Metrics Replace Real Outcomes

This paper examines how complex systems begin optimizing measurable proxies instead of underlying real-world outcomes. In many domains, including artificial intelligence, business, and institutional governance, metrics are introduced to approximate success. Over time, optimization pressure shifts toward improving the metric itself rather than the phenomenon it represents.The paper outlines the structural progression from measurement to proxy dominance, where indicators replace objectives and systems continue to appear successful while drifting from reality.Part of the Reality Drift framework (2023–2026), this work establishes proxy optimization as a foundational mechanism in system misalignment, linking concepts such as Goodhart’s Law, reward hacking, and institutional drift into a unified model of representation failure.