The Silent Three Percent

A companion analysis of scholarly rejection (Vuong & Nguyen, 2024) argues, through the lens of information theory, that editorial rejection acts as a noisy filter: it reduces the entropy of the published record, but the evaluators doing the filtering are themselves fallible. A later empirical study (Nguyen & Vuong, 2025) examined 304 actual rejection letters and found that over 97% attributed the outcome solely to the authors or to the journal's own standards, while fewer than 3% acknowledged any journal-side limitation. This note asks what an honest, noise-aware attribution process would predict, and compares it with what is observed. We show that whenever an evaluator's signal contains any noise at all, a rational observer should assign strictly positive, and exactly computable, probability that a rejected paper was in fact sound (Proposition 1). We then show that if acknowledging fault carries even a small cost to the journal while blaming the author carries none, always blaming the author is a dominant strategy regardless of that honest benchmark (Proposition 2), which is enough to explain the gap on its own. A corollary, reusing an existing result on optimal rule reversal (Vuong, 2026), adds that whatever criteria a rejection letter cites, a reading of those same criteria can always be found that portrays any submission, however strong, as maximally deficient. A further section demonstrates, in a deliberately small and self-contained way, that the same noisy-quality framework extends to a genuine multi-stage game across desk review, peer review, and revise-and-resubmit, solved gate by gate for a Nash equilibrium (Nash, 1951) under specific, exhibited conditions (Propositions 3 and 4). An appendix fits the observed acknowledgment rate to the 304-letter record directly and compares it with the honest benchmark at plausible noise levels.