Campbell's Law¶

When consequential social decisions depend on a quantitative indicator, measured actors game it and distort the process the indicator was meant to monitor.

Version: v2 · 2026-09-19 · History
Domain-specific #: 129
Domain group: Professional & Organizational Practice
Origin domain: Organizational & Management Science
Subdomain: Metric Distortion Under High Stakes → Organizational & Management Science
Aliases: Metric Fixation

Core Idea¶

Campbell's Law describes what happens when an authority uses a quantitative social indicator to make consequential decisions about the people or institutions being measured. The indicator stops serving only as a passive description and becomes something actors must produce. They redirect effort toward improving the number, exploit classification and reporting rules, or change participation in the measured process. The indicator becomes more vulnerable to corruption, and the social process it was meant to monitor is distorted by the attempt to govern through it.

This is a domain-specific species of Goodhart's Law. Goodhart supplies the portable proxy-under-optimization mechanism. Campbell adds the indispensable institutional structure: social decision-making, measured actors who can respond to their treatment under the indicator, and distortion of the governed process as well as degradation of measurement fidelity.

How would you explain it like I'm…

Chasing the Number Messes It Up

A teacher says, 'The class with the fewest broken crayons gets a prize.' Soon, kids hide broken crayons instead of telling. Now counting broken crayons doesn't tell the teacher much anymore, and the kids are acting differently too. When people in charge reward or punish using a number, people start chasing the number, and both the number and what it measured get messed up.

When Scores Rule, Scores Break

Campbell's Law is about what happens when people in charge use a number about people or schools to make important decisions, like rewards or punishments. Once the number matters that much, the people being measured start working to make the number look good instead of doing the real thing it was supposed to show. They might bend the counting rules, change who gets counted, or focus only on what's measured. The number becomes less trustworthy, and the real activity it was meant to track gets bent out of shape. It's a special case of Goodhart's Law, which is about any number that gets pushed on too hard.

Indicator Corruption Under Governance

Campbell's Law describes what happens when an authority uses a quantitative social indicator, such as test scores or crime statistics, to make consequential decisions about the people or institutions being measured. The indicator stops being just a passive description and becomes something those people must produce. They redirect effort toward raising the number, exploit rules for how things are classified and reported, or change who takes part in the measured process. As a result, the indicator becomes more open to corruption, and the social activity it was meant to monitor gets distorted. It differs from Goodhart's Law, which is the general pattern that optimizing a proxy weakens its link to the real target. Campbell's Law is the specifically social and institutional version: it needs a decision-making authority, a quantitative social indicator, high-stakes use of it, measured people who can respond strategically, and distortion of the governed activity.

Campbell's Law states that when an authority uses a quantitative social indicator for consequential decisions about the measured people or institutions, the indicator becomes subject to corruption pressures and distorts the social process it was meant to monitor. Consequential use changes the indicator's role from passive description to something actors must produce. Measured actors respond strategically: they redirect effort toward improving the number, exploit classification and reporting rules, or alter participation in the measured process. The relationship to Goodhart's Law is taxonomic. Goodhart supplies the substrate-portable mechanism, that optimization pressure on a proxy degrades its relationship to the target. Campbell's Law is the social-governance species of that mechanism, adding required roles: a decision-making authority, a quantitative social indicator, consequential use, measured actors capable of strategic response, and distortion of the governed social activity. If those institutional roles are removed, what remains is a generic optimizer-versus-proxy pattern, which is Goodhart's Law rather than Campbell's.

Scope of Application¶

Education: when test scores determine pay or funding, instruction reshapes to the test (curriculum narrowing, score inflation, cheating); the score rises while learning does not.
Healthcare: surgical mortality league tables incentivize patient selection over better surgery; readmission penalties incentivize reclassification.
Policing: accountability for reported crime numbers produces reclassification and discouraged reporting, diverging from victimization surveys.
Corporate KPIs: quarterly-revenue targets generate channel stuffing; engagement targets generate engagement-bait.
Academic publishing: citation counts used for hiring generate citation cartels, salami-slicing, and p-hacking.

Clarity¶

Its clarifying move is to separate social objective (what the authority cares about), indicator (the observable number), decision use (what consequences depend on the number), and measured actors (who can adapt). That makes two failures visible at once: the indicator becomes less trustworthy, and the governed activity itself changes around the indicator.

Manages Complexity¶

It turns a vague warning about "metric gaming" into a governance audit. Before attaching consequences, list who will be affected, what cheap actions can move the indicator, which unmeasured activities those actions displace, and how independently the underlying objective can still be checked.

Abstract Reasoning¶

The load-bearing reasoning is reflexive: using an indicator to govern a social process intervenes on the process that generates the indicator. The validity evidence gathered before high-stakes use does not automatically transport into the regime created by that use, because measured actors now have reasons to reshape both their behavior and the record of it.

Knowledge Transfer¶

Across social institutions: separate indicators used for management from independent indicators used for evaluation; triangulate with measures that respond to different gaming moves; audit the underlying objective outside the reporting chain; and reduce or diversify the consequences attached to any one indicator.
Boundary of transfer: reward hacking in a machine-learning system carries the Goodhart mechanism, but without a quantitative social indicator used to govern people or institutions it is not literally Campbell's Law.

Example¶

When standardized-test results determine school sanctions, funding, and personnel consequences, the test score becomes a high-stakes social indicator. Schools can raise it through genuine learning, but also through curriculum narrowing, intensive rehearsal of tested formats, exclusion or reclassification of students, and sometimes fraud. Scores may improve while independent low-stakes assessments show smaller gains. The accountability system has therefore corrupted the indicator and altered the educational process it was intended to monitor.

This entry is a kind of Goodhart's Law.

Goodhart's Law is the strict parent: Campbell's Law is the social-governance species in which a consequential quantitative indicator changes measured actors' behavior.
Proxy-Target Divergence is inherited through Goodhart's Law rather than repeated as a flattened direct parent.
Measurement, Incentive, and Accountability supply important roles, but none alone captures the full indicator-use-and-process-distortion mechanism.

Domain Classification¶

Campbell's Law remains broadly useful across education, healthcare, policing, administration, firms, and science policy, but these are social-institutional settings sharing the same governance substrate. Removing authorities, consequential social decisions, and measured actors leaves Goodhart's more general proxy-collapse mechanism. The Campbell identity therefore belongs in the domain-specific layer even though its parent prime transfers much farther.

Relationships to Other Abstractions¶

Current abstraction Campbell's Law Domain-specific

Parents (1) — outgoing typed relations

Campbell's Law is a kind of Goodhart's Law Prime

Campbell's Law is Goodhart's Law specialized to consequential use of quantitative social indicators, where measured actors game the indicator and distort the social process it governs.

Hierarchy paths (7) — routes to 5 parentless roots

Campbell's Law → is a kind of Goodhart's Law → is a kind of Intervention-Induced Model Invalidation → is a kind of Concept Drift → is a kind of Calibrated Rule versus Moving World → is a kind of Temporal Decay and Degradation → presupposes Entropy (Thermodynamic Sense)

Show alternative paths (6)

Neighborhood in Abstraction Space¶

Campbell's Law sits in a sparse region of the domain-specific corpus (99^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2769 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25

Not to Be Confused With¶

Campbell's Law is not Competition because Campbell's law is the corruption of a proxy once a stake is attached (a monopolist with no rival falls fully into it), whereas competition is rivalry among agents over a scarce prize.
Campbell's Law is not coextensive with Goodhart's Law. Every Campbell case is a Goodhart case, but Goodhart also covers non-social optimization such as mechanical reward hacking. Campbell requires consequential use of a quantitative social indicator and distortion of the monitored social process.
Campbell's Law is not the Observer Effect because Campbell's law requires an adaptive agent gaming its own fate under the measure, whereas the observer effect is a non-strategic perturbation from the act of measuring.