Drop any document with a claim. Bayescore extracts its IS(subject, criterion) hypothesis — IS(startup, launch_ready), IS(proposal, fundable), IS(campaign, worth_running) — derives the predicates, and scores each using a two-pass adversarial LLM evaluation. Absence of evidence is treated as evidence against.
We ran it on ourselves. Bayescore scored 46/100, Grade D. Five failing predicates. Full breakdown at bayescore.com/self-eval — published because it's right, not flattering.
