Goodhart's Law

Goodhart, C. A. E. (1984). Goodhart's Law: The UK Experience.

Type

Book chapter

Intellectual base

Review or monograph

Year

1984

Volume

1

Issue

1

Pages cited

1-20

DOI

10.1007/978-1-349-17295-5_4

ISBN

9780333360606

DOI registrant

Springer Nature (Macmillan Education UK)

Link

https://www.econbiz.de/10002525062

Cited from

economics_finance, mathematics, philosophy, statistics_experimental_design, systems_cybernetics

Cited by

71 citations across 71 artifacts.

Each citation links to the sentence it supports in the citing article.

Primes

• Aggregation
  • The fifth tension—that an optimized aggregation function is brittle under distributional shift—is the structural content of Goodhart's (1975) observation that any statistical regularity tends to collapse once pressure is placed upon it for control purposes, generalizing far beyond the monetary-policy setting in which it was first stated.

    This sourceOriginal statement that any statistical regularity collapses once pressure is placed on it for control purposes.

• Cue Outcome Decoupling
  • A cue. A perceptible, easy-to-measure proxy the actor can observe at decision time. An outcome. Something the actor actually cares about — food, safety, profit, ground truth — but cannot directly observe when deciding. A historical coupling. A regularity, produced by some mechanism, under which the cue reliably tracked the outcome and so became informative. A fast cue-response. Behavior well-tuned to that coupling, firing reliably on the cue. A slow update channel. The means by which the actor revises the cue's meaning, slower than the world can change it. A coupling-breaking event. The correlation is severed or reversed — by environmental shift, adversarial spoofing, or the actor's own optimization pressure (the Goodhart case) — while the cue itself remains present and readable.
• Goodhart's Law
  • In finance and macroeconomic policy (Goodhart's original substrate), targeting a monetary aggregate collapses the very regularity that motivated targeting it, and the pattern recurs with capital ratios and stress-test thresholds.
• Measurement
  • In software and operations, metrics, telemetry, and dashboards make Goodhart's law the diagnostic: in Strathern's now-standard wording, when a measure becomes a target it ceases to be a good measure — a structural property of the measurement-feedback loop.

    This sourceOrigin of Goodhart's law, stated there as "any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes"—a property of the measurement–feedback loop. The familiar wording "when a measure becomes a target, it ceases to be a good measure" is Strathern's (1997) later generalization.

• Optimization
  • What happens when the proxy is optimized hard — do incentives align with purpose, or does Goodhart's pattern

    This sourceOriginal statement that any observed statistical regularity tends to collapse once pressure is placed upon it for control purposes—the canonical formulation of brittleness in optimized aggregation measures.

• Performativity
  • This is the structural core shared with Goodhart-adjacent phenomena, where a measure adopted as a target ceases to be a good measure precisely because announcing it reconfigures behavior.

    This sourceOriginal statement that any observed statistical regularity tends to collapse once pressure is placed upon it for control purposes—the canonical formulation of brittleness in optimized aggregation measures.

Domain-specific

• Basis-Risk Failure
  • The structural pattern one level up recurs across substrates as true co-instances, with the mechanism intact rather than merely the shape: an actor uses a proxy of a hard-to-act-on target because the target itself is unavailable, illiquid, or unmeasurable; the action is calibrated to the historical proxy-target relationship; under stress or regime change the proxy and target decouple, so the instrument calibrated to their co-movement becomes the wrong instrument at the worst moment. That pattern — call it proxy-target divergence, or proxy decoupling under stress — is the same causal structure whether it shows up as a basis-risk failure in derivatives, as Goodhart's-law breakdown when a measure becomes a target and stops tracking the goal

    This sourceThe paper in which Goodhart's law was first stated, as a point about monetary targets rather than in the later 'measure becomes a target' wording.

    Supported in partVerified against the source

    Read by an automated reader; not a human review. How support was checked

• Customer-Discovery Theater
  • A process with no overbearing decider is still theater if the evidence reaches no one with power to act on it; conversely a strong decider who can be moved by a readout is not running theater. Not Goodhart's law. Goodhart's law — in Marilyn Strathern's now-standard 1997 wording, "when a measure becomes a target, it ceases to be a good measure" — describes a measure degrading once it is made a target, generalising Goodhart's own observation about instruments of monetary control

    This sourceCited for Goodhart's own observation about instruments of monetary control, which this bullet generalises; the 1975 text could not be consulted for this check, and the quoted 'when a measure becomes a target' formulation is — as the sentence now says — Marilyn Strathern's 1997 wording, verified in her European Review paper.

• Progress Illusion
  • Progress illusion differs from Goodhart's Law in mechanism: Goodhart requires optimization pressure on the metric that deforms behaviour; progress illusion arises even without any such pressure, simply from the availability of the activity signal and the unavailability of the outcome measure

    Supported in partVerified against the source

    Read by an automated reader; not a human review. How support was checked

Mechanisms

• Anti-Gaming Scoring Rule
  • Its central failure is the one every metric shares: as soon as a measure becomes a target it ceases to be a good measure, because pressure flows straight into the proxy-outcome gap.

    This sourceStates that a statistical regularity used for control tends to collapse under the pressure placed on it.

• Asymmetry Dimension Scorecard
  • Worse, once a published score becomes a target, the stronger side optimizes the number rather than the underlying imbalance, and the card stops measuring anything real.

    This sourceWarns that a statistical measure used as a control target can be optimized in ways that destroy its relationship to the underlying condition.

• Availability Funnel Dashboard
  • Its classic misuse is that the displayed ratio becomes the target and gets gamed — operators improve the number on the screen (re-baselining, quietly excluding an inconvenient stage) rather than the water reaching taps, exactly the trap Goodhart's law names.

    This sourceWarns that a statistical regularity tends to collapse once it is used as an instrument of control.

• Before/After Trace Monitoring
  • Its failure mode is Goodhart's law — once "reduce the old trace" becomes the target, the trace can be suppressed (block the shortcut) while the underlying friction, and the need behind it, persist and surface elsewhere; a falling metric is not the same as a solved problem.

    This sourceStates Goodhart's law: a measured regularity tends to collapse when subjected to control pressure.

• Benchmark Backtest
  • Tune a model until it passes a fixed benchmark suite and you risk fitting the suite rather than the phenomenon — the essence of Goodhart's law, that a measure optimized as a target stops measuring what it did.

    This sourceDocuments the essence of Goodhart's law; a measure optimized as a target stops measuring what it did.

• Boundary-Sharpening Review Map
  • Optimize the sharpening to maximize the map's own score and the score stops measuring what it was meant to — the trap that Goodhart's law names.

    This sourceIdentifies the collapse of a statistical regularity under control pressure as the trap named by Goodhart's law.

• Bulk-Purchase and Resale Monitor
  • This is also a measurement subject to Goodhart's law

    This sourceThe cited work establishes This is also a measurement subject to Goodhart's law.

• Channel Telemetry Dashboard
  • What it must resist is becoming a target: an agent measured on "unread count" will mark tickets read without resolving them, and the metric goes blind — Goodhart's Law in miniature.

    This sourceStates Goodhart’s warning that an observed statistical regularity tends to collapse when it is pressed into service as a control target.

• Combinatorial Test Coverage Grid
  • A team that manages to the grid will fill cells with shallow tests that touch each combination without truly exercising it, and the all-green grid then certifies a quality it no longer measures — a textbook case of Goodhart's law

    This sourceGrounds the Goodhart-law risk that a metric managed as a target can cease to measure the underlying quality.

• Coordination-Overhead Dashboard
  • The discipline that keeps it honest is to treat the metrics as signals rather than targets, always pair a threshold crossing with the qualitative drift review it triggers, and diagnose the cause with the topology maps rather than the dashboard itself.

    This sourceWarns that a statistical regularity can collapse when placed under pressure for control purposes, supporting use of metrics as signals rather than targets.

• Count Impact Assessment
  • A count that becomes a target stops measuring and starts being gamed — Goodhart's law in miniature.

    This sourceStates Goodhart’s law: statistical regularities tend to collapse when placed under pressure for control.

• Cross-Boundary Reporting Dashboard
  • Once actors learn which indicator is on the dashboard, they optimize to keep that number quiet while the underlying harm continues — the mechanism's standing exposure to Goodhart's law, that a measure under pressure ceases to be a good measure

    This sourceStates that a statistical regularity tends to collapse when it is used as a target for control.

• Crowding-Out Monitoring Dashboard
  • A green board can create false comfort — displacement that the chosen metrics don't capture (informal users, quality degradation) reads as zero — and once a headroom number is watched, there is pressure to redefine it into the green rather than fix the underlying crowding, the classic Goodhart trap.
• Dashboard View Design
  • Its classic failure is choosing panels for visibility rather than task relevance: the metric that is easy to plot crowds out the one that actually predicts trouble, and once a number is on the wall people optimize the number instead of the outcome it stood for

    This sourceStates that a statistical regularity tends to stop representing its prior outcome relationship when it is used as a target for control.

• Data Flywheel Dashboard
  • Once a per-arc metric is the visible score, teams optimize the tile rather than the loop — inflating raw interaction counts that don't improve the model, or gaming the value metric — which is Goodhart's law operating on a wall of charts.

    This sourceStates the principle later known as Goodhart's law: a statistical regularity can collapse when placed under control pressure.

• Distortion Review Cadence
  • That is Goodhart's law turned on the reviewer — the visible marker quietly replaces the substance it stood for

    This sourceStates that an observed statistical regularity tends to collapse once it is used for control.

• Dual-Valence Metric Dashboard
  • The net adoption figure, a textbook Goodhart metric, hid a brewing backlash.

    This sourceGoodhart's principle warns that an observed indicator can become unreliable when it is adopted as a control target.

• Entropy Metric Dashboard
  • A related misuse is optimizing the metric instead of the underlying order, so the number improves while the real disorder is merely hidden from the tile — an instance of Goodhart's Law.

    This sourceStates the control problem known as Goodhart's Law: a previously observed statistical regularity tends to collapse when used for control.

• Excitation–Inhibition Ratio Dashboard
  • The classic misuse is Goodhart's law

    This sourceStates that a statistical regularity can break down when it is used as a target for control.

• Forecast Impact Audit
  • Its strength is that it catches the corruption a raw accuracy score cannot: Goodhart's law

    This sourceStates that a statistical regularity can break down when it is used as a target for control.

• Funnel Process
  • That drift is Goodhart's law: once the early proxy becomes the gate, it stops tracking the thing it stood in for.

    This sourceStates the principle now called Goodhart’s law: an observed statistical regularity tends to collapse when used for control.

• Incentive Change
  • Its failure mode is that incentives bind to the measure, not the intent, and any gap between them is an invitation to game — the essence of Goodhart's Law: once a measure becomes a target, it ceases to be a good measure.

    This sourceStates that an observed statistical regularity tends to collapse when it is placed under pressure for control purposes.

• Incentive Compatibility Review
  • Done well, it pre-empts the "hit the target, miss the point" failure, where a metric improves while the thing it was meant to capture does not.

    This sourceWarns that a statistical relationship can collapse when the measured variable is put under pressure for control.

• Incentive Impact Review
  • Re-examine the link assumption under pressure. Take the written reason the proxy is supposed to track the target and ask which gaming pathway would break it while leaving the number intact.

    This sourceTreats a quantitative indicator as a proxy whose relationship to the underlying objective can fail when the indicator is used for control.

• Independent Reference, Claim, and Conflict Verification
  • It is the antidote to Goodhart's law: once a reference or a metric becomes the target, it stops measuring what it claimed to, so verification weights evidence by how hard it would be to fake.

    This sourceWarns that an observed statistical regularity tends to collapse once it is relied upon for control.

• Institutional Rule and Incentive Redesign
  • Its central failure is that people optimize the measured incentive, not the intent behind it: once a proxy becomes the target, it degrades as a measure and gets gamed

    This sourceStates that a statistical regularity can collapse after it is selected and used as an instrument of control.

• Leakage Budget Dashboard
  • Optimize the measured leak and the unmeasured leaks grow unseen — a straight case of Goodhart's law

    This sourceStates the control-pressure problem known as Goodhart’s law: a residual or regularity ceases to be a faithful measure when made a target.

• Learning Rate Dashboard
  • Put a learning rate on a leaderboard and teams optimize the number: once the measure becomes the target, it stops being a good measure, and effort flows to the displayed figure while unmeasured quality quietly erodes — which is exactly the harm the guardrail exists to catch, now induced by the dashboard itself.

    This sourceShows that turning a measured learning statistic into a control target can induce optimization of the number and destroy its measurement value.

• Local–Global Metric Trace
  • Its own failure mode is subtler: once a local metric becomes the visible proxy for a global outcome, people optimize the proxy in ways that break its link to the outcome — Goodhart's law

    This sourceStates Goodhart's law: a previously observed statistical regularity tends to collapse when it is relied upon for control.

• Longitudinal Fit, Equity, and Burden Audit
  • Turned into a target, its metrics invite Goodhart's law

    This sourceStates the Goodhart principle that a statistical regularity can collapse when used for control.

• Management Dashboard Layer
  • This slides into misuse when the metric becomes the target and people optimize the number rather than the reality it was meant to track — the effect Goodhart's Law names, where a measure under pressure ceases to measure.

    This sourceStates Goodhart’s law: an observed statistical regularity tends to collapse once it is used for control.

• Model-Use Impact and Performativity Review
  • The review's finding is that the metric has become an engine, not a camera: it no longer measures a pre-existing quality so much as induce the behavior it purports to observe.

    This sourceExplains that using an observed statistical regularity for control can make it collapse because intervention changes the behavior that generated it.

• Multi-Level KPI Review
  • It is the direct defense against local metric gaming — Goodhart's law, where a measure that becomes a target stops measuring what it did — because a gamed local number is caught the moment it is checked against the level it was stolen from.
• Normalized Capacity Ratio
  • Worse, once the ratio becomes the number everyone optimizes, it can be gamed until it no longer represents actual headroom at all — a case of Goodhart's law,

    This sourceStates Goodhart's warning that optimizing a measure as a control target can undermine the relationship that made it useful.

• Opportunity Scoring Model
  • Its failure mode is that a scoring model optimizes the score, and the score is a proxy — the moment people act on it, it risks becoming a target rather than a measure, and drifts from the real outcome it stood for.

    This sourceWarns that once a statistical proxy is used for control, the relationship can collapse and drift from the outcome it previously tracked.

• Overlap-Regime Benchmark Table
  • Its failure mode is Goodhart's law

    This sourceStates Goodhart’s law: a statistical regularity can collapse once policy relies on it for control, undermining its role as a measure.

• Perceived-vs-Measured Performance Dashboard
  • Its failure mode is Goodhart's law:

    This sourceWarns that a measured relationship can break down when it is used as the basis for control.

• Performative Feedback Monitoring
  • Its strength is that it surfaces exactly what a launch-day check cannot: the perverse incentive, the concentrated harm, and the reflexive loop in which a measure becomes a target and stops measuring what it named — Goodhart's law in live operation.

    This sourceStates that a statistical relationship tends to break down when it is used as an operational target for control.

• Policy Lever
  • That is Goodhart's law, and it is how a well-aimed lever gets gamed into uselessness.

    This sourceIdentifies Goodhart’s paper as the first reference to what became known as Goodhart’s law.

• Progress Dashboard
  • Its failure mode is that everything it shows is a proxy, and proxies drift from the real target; make an indicator consequential and people optimize the indicator rather than the learning it stood for.

    This sourceStates that an observed statistical regularity tends to break down when it is made consequential for control.

• Proxy-Relevance Audit
  • It is the working defense against Goodhart's law — that a measure adopted as a target degrades as the proxy is optimized apart from the thing it was meant to track.

    This sourceStates that an observed statistical regularity tends to collapse when pressure is placed on it for control purposes.

• Purpose-to-Metric Mapping
  • Its central failure mode is Goodhart's law

    This sourceWarns that an observed statistical regularity can collapse when it is used as a control target.

• Quarterly Business Review
  • When owners are hard-accountable for a number, the review pressures them to manage the metric rather than the business — the essence of Goodhart's law

    This sourceStates the principle known as Goodhart's law: a statistical regularity tends to collapse when it is used for control.

• Ranking Semantics Legend
  • It is a standing guard against Goodhart's law:

    This sourceStates the Goodhart-law failure in which optimizing a measured position causes it to drift from the phenomenon it once proxied.

• Recognition for Belief Revision
  • Once "changing your mind" earns status, people manufacture reversals or perform theatrical updates to collect it; the measure becomes a target and stops meaning what it did.
• Recovery Dashboard
  • This is Goodhart's Law at work: once "percent restored" becomes the target everyone manages to, it stops measuring real recovery.

    This sourceReserve Bank of Australia. States Goodhart’s rule that a statistical regularity tends to collapse once it is pressed into service as a control target.

• Recurrence and Displacement Dashboard
  • This is Goodhart's law operating on the governance process itself: the moment the measure becomes the target

    This sourceStates that an observed statistical regularity tends to collapse once it is pressed into service for control.

• Reflexive Feedback Dampening
  • The deeper misuse is treating dampening as a substitute for fixing a measure that has stopped measuring anything: once a metric becomes a target, actors optimize the number rather than the construct, and no amount of filtering restores the lost meaning — Goodhart's law.

    This sourceStates Goodhart's law: an observed statistical regularity tends to collapse when it is pressed into service as a control target.

• Reputation Score or Standing Index
  • Once the figure gates anything valuable it becomes a target and is optimized rather than earned — Goodhart's law

    This sourceStates the Goodhart principle that an indicator used for control tends to lose the statistical regularity that made it informative.

• Residual Telemetry Dashboard
  • Second, once a panel becomes the target, people optimise the panel: suppression rate is driven to a pleasing number, thresholds are relaxed until the board stays green, and the metric stops tracking reality — the classic Goodhart failure.

    This sourceStates that a statistical relationship tends to collapse when it is put under pressure for control, so the targeted indicator ceases to track the underlying relationship reliably.

• Resource Monitoring Dashboard
  • A metric that is displayed tends to become a target and then gets gamed — Goodhart's law in miniature

    This sourceIdentifies the collapse of a statistical relationship under control pressure as Goodhart's law.

• Risk-Based Tiered Assurance
  • The deepest trap is that once a risk score becomes the target, parties optimize the score rather than the safety it was meant to proxy.

    This sourceWarns that a statistical measure used as a control target can cease to function as the reliable proxy it previously was.

• Service-Level Agreement
  • An SLA rewards hitting the SLO, which is not the same as serving the user — a vendor can meet its uptime and response-time numbers while the actual experience degrades in ways no clause captures, the classic distortion where a target that becomes the objective stops measuring what it was meant to.

    This sourceStates the control problem later called Goodhart's law: a measured regularity tends to collapse once used for control.

• Service-Level Monitor
  • The classic misuse is turning the service-level metric into a target that teams optimize directly: the number improves while the service does not, as agents cherry-pick easy requests to protect p95 or reclassify failures to spare the error budget.

    This sourceWarns that a statistical relationship used as a control target can break down, so measured performance can cease to represent the underlying objective.

• Service-Level Recalibration
  • … is metric capture — because the mechanism is anchored on attainment, the tempting misuse is to move the target to meet the metric rather than move the capacity to meet the customer: loosen the SLA to four hours becomes six, attainment goes green, and the reported number improves while the actual experience decays.

    This sourceGoodhart's observation warns that a measured regularity degrades when subjected to control pressure, providing background for metric capture.

• Service-Level Tier Schedule
  • Its failure mode is that the metric becomes the target

    This sourceStates that an observed statistical regularity tends to collapse when it is pressed into service as a control target.

• Standardized Scoring Rubric
  • Its failure mode is Goodhart's law: once the rubric is the target, applicants optimize the proxy rather than the goal, and a rubric that scores pages of documentation gets documentation, not quality.

    This sourceStates the regularity known as Goodhart's law: an observed statistical relationship tends to break down when used as an instrument of control.

• Stored Load Dashboard
  • Its displayed number is prone to becoming a target managed for its own sake rather than the underlying load — Goodhart's law, where optimizing the indicator hollows out the thing it was meant to indicate.

    This sourceWarns that an observed indicator can lose its relationship to the underlying objective once it is used as a target for control.

• Success Metric Reweighting
  • Once bycatch is scored, effort flows to the measured bycatch, and unmeasured or reclassified harm can grow in its shadow — the reweighted number improves while the world does not, the textbook expression of Goodhart's law.

    This sourceStates the control-pressure problem known as Goodhart's law: behavior changes when a measured regularity becomes a target.

• Time-Boxed Challenge
  • … failure mode is teaching to the metric: entrants optimize the stated win condition rather than the real goal, so you get a technically winning submission that does not actually solve your problem — a live instance of Goodhart's law, where a target that becomes the objective ceases to measure what you cared about.

    This sourceStates the Goodhart principle that a measure used as a control target tends to lose its value as an indicator.

• Voice Mix Dashboard
  • Its failure mode is the metric becoming the target: once "prominence" or "balance" is a displayed number, lines optimize for the gauge rather than for genuine contribution, and the dashboard measures a performance of participation instead of the real thing — Goodhart's law in a deliberation.

    This sourceStates the control-pressure problem conventionally known as Goodhart's law: a statistical regularity can break down when made a target of control.

• Windfall-Use Public Dashboard
  • Once a displayed number becomes the target it is optimised rather than achieved, so publishing "capability investment $" can invite spending that books well but builds little — Goodhart's law, that a measure which becomes a target ceases to be a good measure.

    This sourceReserve Bank of Australia. States that pressure to use an observed statistical regularity for control tends to make that regularity collapse.

Verification

Does it exist? Confirmed. This work's DOI resolves to a registered record, which fixes its identity. That is all it fixes.

Does it back the claim? Read against the text for 2 of 71 citations: 2 supported in part. Each verdict is shown under its citation below, with what in the work backs the sentence.

Was it audited? Yes. A second, independent pass read the citation against the article text and recorded a verdict.

Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.

See how references were verified.

Links previously used in the corpus

Before the registry existed this work was also linked 8 other ways.

• https://www.rba.gov.au/publications/rdp/1990/9013/conference-volumes.html ×26
• https://www.econbiz.de/Record/problems-of-monetary-management-the-u-k-experience-goodhart-charles/10002525062 ×8
• https://www.bankofengland.co.uk/-/media/boe/files/paper/1975/problems-of-monetary-management-the-uk-experience.pdf ×6
• https://doi.org/10.1007/978-1-349-17295-5_4 ×3
• https://www.rba.gov.au/publications/rdp/1990/pdf/rdp9013.pdf ×2
• https://books.google.com/books/about/Problems_of_Monetary_Management.html?id=1CHjMgEACAAJ ×1
• https://link.springer.com/chapter/10.1007/978-1-349-17295-5_4 ×1
• https://www.rba.gov.au/publications/confs/1975/goodhart.html ×1

Registry ID ref:f549a55f0f4d · see in the full table