Goodhart's Law

In 1975, British economist Charles Goodhart was studying monetary policy when he noticed something that would outlive his career and colonize every measuring institution on earth: "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes."

Twenty-two years later, anthropologist Marilyn Strathern reframed it in the language that stuck: "When a measure becomes a target, it ceases to be a good measure." She was writing about university audit systems in Britain. She could have been writing about every training program you've ever designed.

The tool was built to measure distance. Not to be the destination. — Unsplash

This is not a flaw in the people doing the measuring. It is a structural property of any system that ties incentives to indicators. The indicator and the thing it was pointing at start to diverge — slowly at first, then completely. By the time you notice, you've been optimizing the wrong thing for years.

The Practitioner's Version

You have seen this play out. Not as a theory — as a Tuesday. Here is what Goodhart's Law looks like from inside the room:

Completion rates as a proxy for learning. The LMS shows 97% completion. People clicked through every slide to get the certificate. The learning took twelve minutes. The forgetting started fourteen minutes later.

NPS scores as a proxy for impact. Participants rate to be nice. They rate the facilitator's energy, the quality of the coffee, the fact that they got out of the office for a day. Kirkpatrick Level 1 — the smile sheet — was designed to be one input, not the verdict.

Competency checklists as a proxy for competence. Once people know the rubric, they learn to perform the rubric. The checklist stops pointing at the skill and starts pointing at checklist-passing behavior.

Knowledge tests as a proxy for decision quality. A person can score 91% on a post-module assessment and still make the same bad call under pressure — because pressure is not on the test.

⚠ THE COBRA EFFECT — BRITISH INDIA, 19th C.

The British colonial government, alarmed by the cobra population in Delhi, offered a bounty for every dead cobra brought to the office. Logical intervention. Clear metric. The result: entrepreneurs began breeding cobras to collect the bounty. When the government discovered the farms and cancelled the program, the breeders released their stock. The cobra population rose. The incentive created the opposite of the intended outcome — not because the people were malicious, but because the measure became the target.

Siebert, H. (2001). Der Kobra-Effekt: Wie man Irrwege der Wirtschaftspolitik vermeidet. DVA. — German economist Horst Siebert coined the term. The mechanism is universal.

Your learning programs are not growing cobras. But they may be growing completion certificates, favorable ratings, and knowledge-test scores — while the behavior change you promised your client remains unmeasured and therefore unknown.

The Literature

Goodhart's Law did not begin in L&D — Learning & Development, the industry that designs and measures workplace training. It began in monetary economics, traveled through anthropology and higher education auditing, and arrived in the training room via organizational psychology. The convergence of these lineages is the story worth understanding.

Monetary economics → anthropology → psychology → your training room. Four fields. One mechanism. — Unsplash

Alfie Kohn's Punished by Rewards (1993) arrived at the same conclusion from a different direction: incentive structures — including the meta-incentive of being rated and evaluated — undermine intrinsic motivation. When the external measure is present, the internal signal quiets. Practitioners who design for scores may be actively suppressing the autonomous motivation that makes learning durable.

Donald Campbell's 1979 formulation — now called Campbell's Law — adds a social dimension: "The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor." The institutional pressure to show results doesn't just distort measurement. It distorts the thing being measured.

Donella Meadows, writing in Thinking in Systems, frames the deeper issue as an information flow problem: the feedback loop that should correct the system is receiving the wrong signal. The system is not broken — it is working exactly as designed. It is designed around the wrong metric.

Why Smart People Still Do It

Understanding Goodhart's Law intellectually does not stop practitioners from chasing the wrong metrics. If it did, the training industry would have fixed this in 1997. The law is clear. The behavior persists. There must be a competing commitment operating beneath the surface.

KEGAN & LAHEY — IMMUNITY TO CHANGE

The competing commitment behind metric-chasing is the need to prove value. Practitioners who cannot show transformation numerically face an existential threat to their work, their budget, their seat at the table. The metric isn't just a measurement — it is a survival tool. The assumption underneath it: if I can't show the number, I don't get to keep doing the work. This competing commitment (prove value numerically) protects a big assumption (value only counts if it is countable) and it runs hot enough to override the intellectual understanding that the metric is lying.

This is why information alone does not fix Goodhart's Law in practice. The practitioners who know the law still report the same metrics — because the competing commitment (show a number that justifies your existence) runs deeper than the knowledge. Fixing this requires working at the level of identity, not data literacy.

Breaking the Goodhart Trap

The trap is structural. The counter-moves are also structural — design choices, not willpower choices. Here are five that work:

Your Three Metrics