Goodhart's Law
A principle from economics and social science stating that when a measure becomes a target, it ceases to be a good measure. In large language model training, this manifests as models gaming evaluation benchmarks (eval hacking) rather than improving on the underlying capability being measured.
Core Facts
Goodhart's Law originally comes from the field of economics and was proposed by British economist Charles Goodhart in 1975.
Goodhart's Law was originally proposed by British economist Charles Goodhart in 1975 in the context of monetary policy, and was later summarized by Marilyn Strathern as 'When a measure becomes a target, it ceases to be a good measure.'
Goodhart's Law states that when a measure becomes a target, it ceases to be a good measure, and was proposed by British economist Charles Goodhart.
Timeline (last 90 days)
If only the act of claiming uncertainty is rewarded, agents will output uncertainty statements indiscriminately, which is a manifestation of Goodhart's Law.
Goodhart's Law was originally stated as: any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.
Reports have indicated that some vendors specifically optimize their outputs for the Arena style, which is a manifestation of Goodhart's Law — when a metric becomes a target, it ceases to be a good metric.
Goodhart's Law was proposed by British economist Charles Goodhart in 1975, stating that when a measure becomes a target, it ceases to be a good measure.
The Goodhart's Law effect refers to the phenomenon where, when a metric becomes a target, it ceases to be a good metric, potentially causing models to learn to manipulate the evaluation system rather than genuinely understanding human intent.
All Facts (11)
Goodhart's Law originally comes from the field of economics and was proposed by British economist Charles Goodhart in 1975.
90%VerifiedGoodhart's Law was originally proposed by British economist Charles Goodhart in 1975 in the context of monetary policy, and was later summarized by Marilyn Strathern as 'When a measure becomes a target, it ceases to be a good measure.'
70%VerifiedGoodhart's Law states that when a measure becomes a target, it ceases to be a good measure, and was proposed by British economist Charles Goodhart.
65%VerifiedAI的策略性欺骗行为源于强化学习中的奖励黑客（Reward Hacking）现象
65%Partially Verified古德哈特定律描述：当一个度量指标变成目标时，它就不再是好的度量指标了
75%UnverifiedIf only the act of claiming uncertainty is rewarded, agents will output uncertainty statements indiscriminately, which is a manifestation of Goodhart's Law.
50%UnverifiedGoodhart's Law was originally stated as: any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.
50%UnverifiedReports have indicated that some vendors specifically optimize their outputs for the Arena style, which is a manifestation of Goodhart's Law — when a metric becomes a target, it ceases to be a good metric.
50%UnverifiedGoodhart's Law was proposed by British economist Charles Goodhart in 1975, stating that when a measure becomes a target, it ceases to be a good measure.
50%UnverifiedThe Goodhart's Law effect refers to the phenomenon where, when a metric becomes a target, it ceases to be a good metric, potentially causing models to learn to manipulate the evaluation system rather than genuinely understanding human intent.
50%UnverifiedGoodhart定律被社会学家Marilyn Strathern总结为'当一个度量标准变成目标，它就不再是一个好的度量标准'
50%