FSRS schedules the card. The rubric decides whether seeing it teaches anything.

FSRS stands for Free Spaced Repetition Scheduler. It shipped natively in Anki 23.10 (October 2023) and is the default for new collections in current Anki releases. The model has three latent variables per card (Difficulty, Stability, Retrievability, the DSR triple) and is fit by gradient descent on your own review history. SM-2 is the older Anki default, a fixed heuristic with an ease-factor that drifts under bad ratings. The practical difference: FSRS asks 'what interval keeps this card at my chosen retention probability', SM-2 asks 'what does the ease factor say to multiply the last interval by'. FSRS is more efficient at the same retention target and recovers from rating mistakes more gracefully.

Desired retention is the probability that you will recall a card on review day. The FSRS default is 0.90 (90 percent). Pushing it to 0.95 or 0.97 buys higher single-review accuracy at the cost of roughly 1.5x to 3x more reviews per day. Pushing it down to 0.85 saves reviews but means you will forget more. The honest answer for medical, dental, nursing, pharmacy, vet exams: leave it at 0.90 unless you have empirical evidence (from the Anki FSRS evaluator) that a different value works better for your collection. The bigger lever for deep mastery is upstream of FSRS, in whether the card itself is testing concept recall or just stem-pattern recall.

No. FSRS sees one signal per review: your Again/Hard/Good/Easy rating. It cannot see WHY you got the card right. If the stem reads 'The patient presents with crushing substernal chest pain and ST elevation in II, III, aVF...' and you have seen the exact wording five times, you will rate it Good in 1.2 seconds because the first ten words are a unique key into your mental cache, not because you understand inferior MI. FSRS will then push the interval out, mark you as mastering it, and you will fail the same concept on test day when the wording changes. This is the failure mode the MCQ rubric layer exists to catch.

Auto-rephrasing means the stem is rewritten on each surfacing while the underlying concept and answer stay fixed. The day-1 stem might be 'inferior wall MI typically presents with ST elevation in which leads', the day-8 stem might be 'a 58-year-old with crushing chest pain shows ST changes in II, III, and aVF, which artery is most likely occluded'. Same fact, different verbal surface. When FSRS surfaces the card on the day-8 interval, your retrieval has to actually reach the concept; the stem-cache is not enough. The Again/Hard/Good/Easy rating you give back is now measuring concept recall, which is what FSRS thinks it is measuring. Without rephrasing, the rating is measuring a verbal lookup, and FSRS schedules accordingly.

Anki ships with Basic and Cloze note types and not much else. You can build an MCQ template by hand: add four option fields, write the front template to render the stem and the four options as a list, and write the back template to highlight the correct one. The .apkg files Studyly exports include a custom note type (studyly_mcq) with the four-option fields, an Explanation field, and Source fields, plus the front and back templates pre-wired. Once imported, you get MCQ cards as native Anki notes, schedulable by FSRS like any other card.

Two options. (1) Re-export from Studyly periodically; the export carries a fresh stem variant on each generation, so syncing a new .apkg into the same Studyly-namespaced deck rotates the wording. The cards keep their FSRS scheduling state because the import matches on the studyly_card_id field. (2) Use the Studyly app for review instead of Anki for the cards you want stem-rotation on, and let Anki handle the cards where the rote wording is the point (drug names, vocabulary, anatomy labels). The two are complementary, not exclusive.

Almost none. Default desired retention 0.90, FSRS on, the optimizer run once after you have around 1000 reviews in the collection (Tools, FSRS, Optimize). If you import a Studyly .apkg into an existing collection, leave the new deck on the inherited preset; do not create a separate FSRS profile per imported deck. The DSR model averages across cards in the collection, more cards per profile means a tighter fit. The full Anki documentation on this lives at docs.ankiweb.net/deck-options.html.

Yes. FSRS treats every card the same: a sequence of ratings with timestamps, a DSR fit, a next-interval prediction. Image-occlusion cards on the studyly_image_occlusion note type schedule under FSRS exactly like an MCQ note. The mask varies between cards in a set (different muscles masked from the same brachial-plexus diagram); each mask is its own card with its own DSR triple. Deep mastery here means rating yourself on whether you could name the structure unprompted, not whether you can recognize it once revealed.

Cramming is fundamentally outside what FSRS optimizes for. FSRS optimizes the long-term review schedule given a target retention; the day before an exam, the schedule is already as compressed as it gets, and you are review-density-bound, not algorithm-bound. The relevant question on cram day is whether the cards you are drilling test what the exam will test. A 200-card deck with strong distractors and rotating stems gives you a tighter signal in two hours than a 400-card deck of recognition-only basic cards. The deep-mastery framing on this page applies to the eight days before exam day; on day-of, just retrieve.

It is not, directly. The 81.3 score on the held-out three-document eval measures four things about the cards themselves (factual correctness, clarity, distractor quality, question-type coverage). FSRS measures none of those. The point of mentioning both on the same page is that they sit on different layers: FSRS optimizes the schedule of a deck, the eval measures the quality of a deck, and deep mastery requires both to be strong at once. Methodology for the eval is at studyly.io/quality.