False-acceptance frequency of external decision mechanisms

Determine how frequently an external decision mechanism incorrectly accepts a candidate as ready, including under requirement-based sensor-coordinate and polarity binding tasks.

Background

The paper emphasizes that moving release authority outside a language-model generator does not by itself establish the reliability of the resulting acceptance layer. An external gate may still produce incorrect acceptances, so its false-release frequency must be assessed independently of the model’s confidence or self-assessment. The paper reports a benchmark-specific measurement, but the broader question identified in the literature remains unresolved.

References

However, two questions are left unresolved: whether a generator's output can be checked reliably by the generator itself, and how often an incorrect acceptance is produced by the external decision mechanism.

These finite comparisons identify an empirical dependence within the panel and leave its internal mechanism open.

— Adversarial Cues in Decision Models Used as Judges: The Role of Request Presentation  (2610.11436 - Liu, 8 Oct 2026) in Section 4.2, “Readable cues and dependence on the grading rule” (Appendix, subsection labeled sec:lexical)