BayesCore

BayesCore とは？

Drop any document with a claim. Bayescore extracts its IS(subject, criterion) hypothesis, derives falsifiable predicates, and scores each using a two-pass adversarial LLM evaluation. Absence of evidence reduces confidence toward zero. Formula locked: score = Σ(weight × confidence) × 100. We ran it on ourselves. 46/100, Grade D. Five failing predicates. Published because it's right, not flattering. bayescore.com/self-eval