← Back to Paradoxes

The McNamara Fallacy

"What gets measured gets managed"—and gamed

The Paradox: Measuring performance should improve it—but when metrics become targets, they stop measuring what matters. The more a measure is used for control, the more it gets gamed, until it measures nothing but the gaming itself.

Interactive Metric Gaming Simulator

12,547
Enemy Body Count
MEASURED ✓
10:1
Kill Ratio
MEASURED ✓
?
Civilian Support
IGNORED ✗
?
Enemy Morale
IGNORED ✗
Outcome:
The metrics look great! But unmeasured factors—civilian resentment, guerrilla resilience, political will—were what actually determined victory.
72%
Average Test Score
MEASURED ✓
85%
Graduation Rate
MEASURED ✓
?
Critical Thinking
IGNORED ✗
?
Love of Learning
IGNORED ✗
Gaming tactics available:
• Teach to the test only
• Classify struggling students as "disabled"
• Lower passing thresholds
• Exclude low performers from testing
4.2
Avg. Length of Stay (days)
MEASURED ✓
2.1%
Mortality Rate
MEASURED ✓
?
Patient Quality of Life
IGNORED ✗
?
Readmission Rate
IGNORED ✗
Gaming tactics available:
• Discharge patients prematurely (lower LOS)
• Transfer terminal patients (lower mortality)
• Refuse high-risk cases
4:32
Avg. Call Duration
MEASURED ✓
47
Calls/Day
MEASURED ✓
?
Problem Resolution
IGNORED ✗
?
Customer Loyalty
IGNORED ✗
Gaming tactics available:
• Hang up on difficult callers
• Rush customers off the phone
• Mark issues "resolved" without solving them
• Transfer to other departments repeatedly

Yankelovich's Four Steps to Disaster

1

Measure the Easy

Measure whatever can be easily measured

2

Disregard the Hard

Disregard what can't be easily measured

3

Assume Unimportance

Presume what can't be measured isn't important

4

Deny Existence

Say what can't be measured doesn't exist. This is suicide.

Three Related Laws

Goodhart's Law

Charles Goodhart, 1975
"When a measure becomes a target, it ceases to be a good measure."

Originally about monetary policy, now applies universally to any metric used for control.

Campbell's Law

Donald Campbell, 1969
"The more any quantitative social indicator is used for decision-making, the more subject it will be to corruption pressures."

Developed while studying standardized testing in British schools.

The Cobra Effect

Colonial India
"Incentives create their own perverse outcomes."

British offered bounties for dead cobras. Indians bred cobras for the bounty. When cancelled, they released them—more cobras than before!

Real-World Examples

🏭

Soviet Nail Factory

Target: Number of nails
Made tiny, useless nails. Switched to weight target → made giant, useless nails.
👮

Police "Juking Stats"

Target: Crime rates
Downgrade serious crimes, group multiple incidents as one, refuse to file reports.
✈️

Airlines On-Time

Target: On-time arrivals
Artificially inflate scheduled flight times. 3-hour flight scheduled as 4 hours.
📦

USPS & Amazon

Target: Delivery rates
Carriers told to scan undelivered packages at 7:15pm to avoid "late" designation.
🎓

Schools & Testing

Target: Test scores
Teach to test, exclude low performers, lower passing thresholds.
💊

Clinical Trials

Target: Progression-Free Survival
Optimizes for measurable endpoint, ignores quality of life and overall survival.

The Vietnam War Origin Story

The Numbers Looked Great...

Reported Kill Ratio (US:Enemy) 10:1
Enemy Body Count (1965-1968) 435,000+
US Troops at Peak 543,000
Predicted Victory "Imminent"
Actual Outcome DEFEAT
1962
General Edward Lansdale tells McNamara to add an "x-factor" to his metrics— the feelings of Vietnamese civilians. McNamara erases it: "I can't measure it, so it must not be important."
1965-1968
"Body count" becomes the primary success metric. Officers are promoted based on kill ratios. Estimates become wildly inflated. Civilian casualties counted as enemy.
1968
Tet Offensive: Despite "winning" by body count metrics, the US is strategically surprised. Public support collapses. McNamara resigns.
1975
Fall of Saigon. The unmeasured factors—political will, guerrilla resilience, civilian support—proved decisive. The metrics were worthless.
"The first step is to measure whatever can be easily measured. This is OK as far as it goes. The second step is to disregard that which can't easily be measured or to give it an arbitrary quantitative value. This is artificial and misleading. The third step is to presume that what can't be measured easily really isn't important. This is blindness. The fourth step is to say that what can't be easily measured really doesn't exist. This is suicide." — Daniel Yankelovich, describing the McNamara Fallacy

How to Avoid the McNamara Fallacy