Unverified50% confidenceConceptExact time
The theoretical premise of self-consistency voting is that errors across paths are independent of each other; but when candidates are generated by the same model, the same prompt, and the same context, errors are often highly correlated, and voting merely repeats the same error.
1
Sources
50%
Confidence
Long-term
Relevance
10/24/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedStatistically, majority voting can effectively reduce error rates only when errors are mutually independent; when errors are highly correlated, consensus instead reinforces mistakes79% similarUnverifiedThe statistical basis for the credibility of majority voting, the Condorcet Jury Theorem, requires that voters be mutually independent, and the error correlations among LLM reviewers violate this premise.76% similarUnverifiedA race condition is when multiple execution flows read, modify, and write back the same shared state at almost the same time, with the final result depending on the interleaving order, which can lead to defects such as overselling.72% similarUnverifiedarXiv paper 2609.22512 points out that treating the consensus of multiple LLM reviewers as strong evidence presupposes that the errors of each reviewer are mutually independent, whereas in reality LLM reviewers make the same errors due to similar training data and evaluation approaches, causing consensus to exaggerate evidential strength.71% similarUnverifiedThe framework introduces two complementary signals: decorrelation in error patterns and divergence in predictive behavior.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1092547API
curl https://kongchang.com/api/v1/knowledge/claims/1092547MCP
get_claim(id=1092547)