Self-Assessment Calibration Simulator

Self-assessment is only useful when it's accurate: a learner who thinks they understand something they don't will never go back and fix it. This simulator models a learner answering practice questions of varying difficulty, judging their own confidence after each one, and compares that self-judgment to what actually happened. Three sliders control the learner's true ability, a systematic self-assessment bias (over- or under-confidence), and metacognitive noise (how consistently their self-judgment tracks their real performance). Every answered item lands as a green or red sphere in a 3D reliability diagram bucketed by reported confidence, with a live purple accuracy bar per bucket compared against the grey perfect-calibration diagonal, plus running Brier score and Expected Calibration Error — the same metrics used to study real students' self-assessment, goal-setting, and strength/weakness identification.