In 1975, the British economist Charles Goodhart observed something that economists had long intuited but rarely formalized: any statistical regularity that is used for control purposes will tend to collapse. The insight was made in the context of monetary policy — specifically, the British government's attempts to control the money supply as an intermediate target for inflation. But the principle transcended monetary economics. It described a structural feature of how systems behave under institutional observation.
Goodhart's Law, in its most commonly cited formulation (due to Marilyn Strathern rather than Goodhart himself), states: "When a measure becomes a target, it ceases to be a good measure." This formulation is tidy. The mechanism it describes is not. The collapse of a measure's reliability when it becomes a target is not a simple event but a complex social process involving strategic adaptation, perverse incentives, informational distortion, and the gradual substitution of target-performance for genuine goal-achievement across an entire institutional domain.
At the collective scale, Goodhart's Law is among the most consequential structural regularities in institutional life. Governments measure economic output through GDP and set GDP growth as the target of macroeconomic policy; within decades, GDP growth becomes achievable through mechanisms — debt-financed consumption, financial sector expansion, resource extraction — that bear no necessary relationship to population welfare. Education systems set standardized test scores as the measure of learning quality and then adopt those scores as targets for school performance accountability; within years, the scores rise while evidence of underlying cognitive skill stagnates. Central banks adopt inflation indices as targets; banks and firms learn to structure transactions in ways that fall outside the measured index, undermining the index's informational content.
The practical mechanism of Goodhart's Law operates through a specific form of collective intelligence — not the wisdom of crowds but the adaptability of crowds. When an institutional metric is announced and consequential, the distributed intelligence of the population subject to the metric immediately begins searching for ways to produce favorable readings. This search is not necessarily conscious or conspiratorial. It arises from the ordinary pressure-response dynamics of people and organizations trying to survive and succeed in institutional environments. Teachers teach to the test because teaching to the test is what administrators reward. Hospital administrators code diagnoses strategically because strategic coding is what the reimbursement system rewards. Police departments prioritize easily closed cases because closure rates are what departmental evaluations reward. No single actor decides to corrupt the metric. The corruption emerges from the aggregate of individually rational responses.
The systemic consequences accumulate across several dimensions. The informational dimension: once a metric is known to be a target, the data it generates becomes unreliable as evidence about the underlying condition it was meant to track. Policy decisions made on the basis of that data are therefore made on the basis of artifact rather than reality. The allocative dimension: resources directed by the metric flow to organizations and practices that are good at producing the target number, not necessarily at producing the underlying value. The selective dimension: over time, institutions select for practitioners skilled at target-production, further degrading actual competence in the underlying work. The cultural dimension: the target becomes the operative definition of success within the institutional field, and practitioners lose the conceptual vocabulary and evaluative practices needed to assess genuine performance.
What makes Goodhart's Law practically difficult to circumvent is that the standard response — adopting a better metric — reproduces the problem. Any new metric that is consequential enough to actually influence behavior becomes subject to the same dynamic. The solution is not a better measure but a different relationship between measurement and governance — one in which measures inform judgment rather than replace it, and in which the consequence of favorable metric performance is scrutiny rather than automatic reward.
The deeper issue Goodhart's Law identifies is about collective attention. Metrics are technologies for directing institutional attention toward specific signals. Goodhart's Law describes what happens when the signal becomes more important than what it was signaling: the signal is detached from its referent and manipulated as an end in itself. Reclaiming genuine collective attention to the underlying conditions — the actual welfare of students, patients, communities, and markets — requires institutional architectures that resist the colonization of attention by the target, preserving the capacity to see past the number to what the number was supposed to represent.