Keeps the reasons, then grades the result
Each significant decision is saved with its confidence and several independent reasons. Later it is reviewed against the outcome, and Brier scores (the average squared gap between confidence and outcome; lower is better) show whether its confidence was right.



