"When a measure becomes a target, it ceases to be a good measure." — Charles Goodhart, 1975. The moment you optimize for a metric, you destroy its usefulness.
Moscow sets production targets. Watch what happens when the metric becomes the goal.
Usefulness over time as factory "optimizes":
Employees were incentivized to open new accounts. They opened 3.5 million fake accounts without customer consent.
Schools judged by test scores. Teachers narrow curriculum to tested material only. Critical thinking and creativity suffer.
UK hospitals measured on A&E wait times. Solution? Patients held in ambulances outside (not "in A&E" yet).
Platforms optimize for "engagement." Result: outrage, misinformation, and addiction maximize clicks.
Metrics are proxies for what we actually care about. When we target the proxy, people optimize for it—often in ways that undermine the underlying goal. The metric stays good while reality degrades. This is why Campbell's Law warns: "The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures."
Multiple metrics: Harder to game several at once. But beware—people optimize for the weighted combination.
Qualitative judgment: Include human assessment alongside numbers. But this doesn't scale.
Rotate metrics: Change targets before gaming becomes entrenched. But this creates confusion.
Measure outcomes, not outputs: Focus on actual goals (patient health) not proxies (procedures performed). But outcomes are harder to measure.
Charles Goodhart formulated this in 1975 regarding monetary policy. The insight applies everywhere: policing statistics, academic citations, app store rankings, SEO, KPIs...