Unverified50% confidenceDecision RuleExact time
When evaluating a metric, ask: if a number is optimized to an extreme value, does the real goal improve in tandem? If the answer is no, then it is at best a vanity metric, and using it to drive an agent will inevitably lead to goal drift.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGoodhart's Law states that when a quality metric becomes an optimization target, it is no longer a good measure of quality, as the model may learn to game the score rather than genuinely improve.78% similarUnverifiedWhen an agent is given an optimization objective of improving a benchmark score, there is an inherent incentive to achieve that goal by gaming the benchmark rather than genuinely improving the methodology75% similarUnverifiedThe validity of surrogate indicators rests on a causal relationship between outputs and competence and on scarcity—producing the output itself requires genuine competence as a prerequisite.75% similarUnverifiedThe effectiveness of automatic hill-climbing metric optimization is highly dependent on the quality of the chosen evaluation metric, may fall into the limitations of the metric itself, and requires researchers to intervene with judgment at critical points.74% similarUnverified确定性优化可能以牺牲模型精度或表达能力为代价，需在质量与能效之间寻找平衡点74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1019322API
curl https://kongchang.com/api/v1/knowledge/claims/1019322MCP
get_claim(id=1019322)