Change what the system is optimizing for

If the goal of the system produces the behavior you want to change, change the goal.

Why it works

In Meadows’ hierarchy, changing the goal of a system is higher-leverage than changing the rules, because the rules exist to serve the goal — if the goal stays the same, new rules will be gamed back toward the old behavior. Many intractable problems persist because everyone is actually succeeding at achieving a goal that produces the problematic behavior as a side effect; the solution is not to constrain the behavior but to change what success means.

How to do it

1. Ask: "What goal is this system actually optimizing for?" — not the stated goal, but the one revealed by the behavior.
2. Compare that to the goal you want. If they differ, identify who or what sets the current goal.
3. Design a new goal metric that would make the desired behavior the natural path to success.
4. Check for goal displacement: once the new metric is in place, is it producing unintended optimizations?

Evidence

Goal displacement — systems optimizing for measurable proxies rather than actual outcomes — is extensively documented in organizational and public policy research. Goodhart’s Law ("when a measure becomes a target, it ceases to be a good measure") is the canonical formulation. (observational)

Changing goals in social systems requires legitimacy and buy-in; technical correctness of a new goal does not guarantee adoption.

Sources

• Goodhart (1975), cited in Marilyn Strathern (1997), "‘Improving ratings’: audit in the British university system"

Common mistake

Designing a new goal metric without considering second-order effects — every goal metric will be optimized, including in ways you did not intend, so the new metric needs game-theory testing before implementation.