The Tyranny of Metrics
Authors: Jerry Z. Muller
Publisher: Princeton University Press; Альпина Паблишер
Length: —
Jerry Muller's argument that metrics remain useful only when combined with context, professional judgment, and protection against substituting proxies for goals.
OriginalBrief
The essential idea
Jerry Muller begins with environments where reporting and key performance indicators displace substantive work, then examines the same pattern in education, health care, policing, the military, business, and philanthropy. The central risk aligns with Goodhart's law: once a measure becomes a target, pressure changes the behavior it was meant to observe.
Metric fixation produces predictable side effects: proxy optimization, short-termism, reporting cost, expanding bureaucracy, reward for luck, avoidance of valuable risk, competition between functions, and declining engagement. More transparency is not automatically better when observation distorts negotiation, judgment, or cooperation.
The answer is not to reject data but to limit what numbers are asked to decide. Leaders should separate measures used by a team for learning from measures used for accountability, state possible harmful incentives before launching a KPI, add qualitative context and expert review, and retire measures that no longer serve their original purpose.
Decision lens
Key takeaways
A measure can stop representing reality when people are rewarded or punished for optimizing it.
More metrics do not automatically produce better management or better decisions.
Metric pressure can shorten time horizons and discourage worthwhile risk.
Reporting and control have real operating costs that must be compared with their benefit.
Team and system measures should not be used as direct scores for individual performance.
Professional judgment and qualitative context are complements to measurement, not signs of failure.
Every KPI needs a review schedule, a known decision use, and a plan for detecting manipulation.
Workplace experiment
Apply it at work
- 1
Name the decision supported by each key metric and remove measures with no clear use.
- 2
Document the harmful behavior each proposed KPI might encourage before deployment.
- 3
Separate the team's learning dashboard from business-accountability reporting.
- 4
Add examples, limits, and expert review to consequential metric interpretations.
- 5
Schedule periodic retirement reviews for indicators that have lost their connection to the goal.
Choose one action, define the observable effect, and keep the first test small enough to reverse.
Evidence
Sources and further reading
Primary source
Additional sources
Channel, aggregator, and commentary links confirm the work; they are not the primary source.