Guidance and mistake-reflection update — October 28, 2026

Behavior

Home combines demonstrated low-accuracy topics (at least five answers), missing science/CARS evidence, and recent concept mix-ups. Missing evidence is not treated as proven weakness or strength. Ranking weights remain explicit study heuristics.
The daily passage assignment is stable across refreshes. Completion requires that subject, and a mini-test assignment requires a mini test. Another subject's passage cannot mark CARS complete.
Topic metadata influences passage selection among unused items. It is a relevance preference, not a guarantee that every passage question matches the weakest subtopic.
Daily recall separates first-seen base facts from review attempts. Repeats and alternate forms do not increase new-fact coverage. Historical answers are retained; earlier new/review splits are unknown rather than fabricated.
Once new and due-review goals are met, extra Smart Sessions are optional. Broad content pace is labeled separately from test readiness.
A baseline diagnostic/review is recommended within a week when none is recorded and the exam is not imminent. An outside diagnostic can be marked reviewed. The AAMC study-plan guide supports diagnosing weaknesses and working on them between exams.
Score-report weighting eligibility is no longer labeled syllabus coverage.

Incorrect answers only

Selecting an answer submits immediately. Correct recall/tutor answers do not ask for another option and are not included in alternative-answer analysis. Incorrect answers optionally prompt after feedback. Timed mini/full-length tests prompt only during incorrect-answer review, never during the timed questions.

Feedback distinguishes an alternative that was correct from one that was also incorrect. Because the question is asked after feedback, the app explicitly treats it as retrospective reflection, not reliable evidence of prior elimination, scientific closeness, or partial credit. A later selection replaces the same reflection record. Old before-feedback and correct-answer records are not included in the new metric; old exam alternatives are not mislabeled as new retrospective responses.

Regression coverage

The learning-guide fixture recreates the reported pattern: Atomic Structure 3/6, sparse Physics/Biochemistry/Organic Chemistry, P/S passage completion, zero CARS, and 125 earlier attempts. Checks verify chemistry priority, CARS guidance, matching completion, assignment stability, distinct new/review counting, diagnostic timing, optional extras, and honest score wording. Other tests exercise immediate submission, no correct-answer prompt/analysis, incorrect-only full-length review, mini-test delayed feedback, identity preservation, and content quality. Browser CI verifies the changed controls in the actual page.