How does the 'New Cards Per Day' limit interact with your future FSRS-6 review workload?

In steady state, each new card introduced per day generates roughly 7 to 10 total reviews over its first year (at 90% retention), meaning 20 new cards/day stabilizes to ~140–200 daily reviews.

What is a 'Leech Card' in spaced repetition?

A card that has lapsed (been rated 'Again') repeatedly (typically ≥ 4 to 8 times), consuming disproportionate review time while failing to stabilize in long-term memory.

With w_20 = 0.5 (F = 19/81), calculate the interval multiplier m(R_d) if you lower Desired Retention from 90% to 80% (R_d = 0.80).

m(0.80) = (0.80^(-2) - 1) / (19/81) = (1.5625 - 1) / 0.23457 ≈ 2.40—meaning intervals become 2.4× longer than at 90% retention.

With w_20 = 0.5 (F = 19/81), calculate the interval multiplier m(R_d) if you raise Desired Retention from 90% to 95% (R_d = 0.95).

m(0.95) = (0.95^(-2) - 1) / (19/81) ≈ (1.10803 - 1) / 0.23457 ≈ 0.4605—meaning every interval is cut to less than half (46%) of its 90%-retention length!

What happens in FSRS-6 if you press 'Easy' (G = 4) on a card that was already 30 days overdue?

You receive a double compound boost: (1) the large overdue retrievability bonus from low R, AND (2) the Easy multiplier w_16 > 1, plus a reduction in Difficulty D.

Conversely, when is sorting overdue cards by 'Lowest Retrievability First' (most overdue first) useful?

When your backlog is small enough to clear over a few days and you want to maximally exploit the overdue stability bonus exp(w_10 · (1 - R)) - 1 on cards you still remember.

State the 'Rule of 7–10×' for estimating steady-state daily reviews from daily new cards.

At 90% Desired Retention, your long-term daily review count stabilizes at approximately 7× to 10× your daily new-card intake (e.g., 15 new cards/day → ~110–150 reviews/day).

Using the FSRS-6 interval equation I(R_d, S) = (S / F) · (R_d^(-1 / w_20) - 1), what does I(0.90, S) simplify to when R_d = 0.90?

When R_d = 0.90, the term (0.90^(-1 / w_20) - 1) is identically equal to F, so F / F cancels out and I(0.90, S) = S (the next interval equals Stability in days!).

What is the difference between 'Target (Desired) Retention' and 'True (Average) Retention' in a deck?

Desired Retention (R_d) is the threshold probability at which a card becomes due (the floor of its review cycle), whereas Average Retrievability across all cards in your deck is always higher than R_d because most cards were reviewed more recently than their due date.

Write the exact FSRS-6 formula that converts a card's Stability (S) and your Desired Retention (R_d) into the next scheduled interval I(R_d, S).

Solving R_d = (1 + F · I / S)^(-w_20) for I gives: I(R_d, S) = (S / F) · (R_d^(-1 / w_20) - 1), where F = 0.9^(-1 / w_20) - 1.

State the golden rule for choosing between 'Again (1)' and 'Hard (2)' in FSRS-6.

If your retrieval was incorrect or incomplete, ALWAYS press 'Again (1)'; press 'Hard (2)' ONLY when your retrieved answer was 100% correct but took uncomfortable mental strain or extra seconds to surface.

Does capping the displayed Maximum Interval at 365 days cap the internal Stability (S) state variable?

No. Internal Stability (S) reflects your true psychological memory trace; when reviewed at day 365, FSRS-6 uses the elapsed time t = 365 and internal S to compute R(365, S) accurately.

Summarize the mathematical relationship between Desired Retention (R_d), Interval Length (I), and Daily Workload (W).

As Desired Retention R_d increases toward 100%, Interval Length I shrinks non-linearly toward 0 and Daily Workload W rises exponentially; R_d = 90% sits at the knee of the efficiency curve.

Why do Leech Cards occur even under an optimal scheduler like FSRS-6?

Because no scheduling algorithm can fix a poorly formulated flashcard—leeches are almost always caused by ambiguous prompts, multi-fact lists, lack of conceptual understanding, or interference between similar cards.

Why does lowering Desired Retention below 75% often INCREASE total daily study time despite producing longer intervals?

Because when more than 25% of reviews end in a lapse ('Again'), the time spent re-learning forgotten cards and completing extra post-lapse repetitions outweighs the time saved from longer intervals.

Why does raising Desired Retention from 90% to 97% cause an explosive, non-linear surge in daily review workload?

Because as R_d approaches 1.0, the interval multiplier (R_d^(-1/w_20) - 1) / F approaches 0, shrinking intervals toward 1 day and forcing you to review almost your entire collection every few days.

What is the 'Hard Button Misuse' anti-pattern that learners carry over from SM-2 or bad habits?

Pressing 'Hard' when you actually forgot the card (out of reluctance to press 'Again'): in FSRS-6, rating a forgotten card as 'Hard' falsely tells the algorithm that recall succeeded, inflating Stability S and causing guaranteed future lapses.

How does FSRS-6's Maximum Interval cap work, and what is a sensible setting?

The Maximum Interval clamps scheduled intervals to an upper ceiling (e.g., 365 days for a 1-year exam timeline, or 3,650–36,500 days for lifelong learning) so cards are reviewed at least once within your chosen horizon.

If you want to cut your daily review time in half next week without changing Desired Retention, what is the single fastest lever?

Temporarily set 'New Cards Per Day' to 0: because recent cards in their first 1–14 days account for the majority of daily reviews, pausing new cards rapidly shrinks the daily queue within 7 days.

When should a learner intentionally choose a lower Desired Retention of 82%–85%?

When maintaining a massive lifelong deck (e.g., 20,000+ general vocabulary or trivia cards) under strict daily time limits where occasional lapses are acceptable.

What is the 'Optimal Retention Sweet Spot' recommended for 95% of learners in FSRS-6?

Between 85% and 92% (with 90% as the universal gold-standard default), which maximizes total knowledge retained per minute of daily study.

If your Desired Retention is set to 90% (R_d = 0.90), what will your overall deck's average Retrievability typically be?

Approximately 94% to 96%, because cards start at R = 100% right after review and decay gradually down to 90% on their due date.

Why is the interval multiplier m(R_d) = (R_d^(-1 / w_20) - 1) / F strictly less than 1.0 whenever Desired Retention R_d > 0.90?

Because requiring a recall probability higher than 90% means you must review the card earlier than its 90%-decay point S.

When should a learner temporarily raise Desired Retention to 93%–95%?

During the final 2 to 3 weeks before a high-stakes board or licensing exam (e.g., USMLE, Bar Exam, JLPT) on a bounded, high-yield core deck.

What is the optimal sort order when clearing a large backlog of overdue cards in FSRS-6?

Sort by Retrievability descending (highest R / least overdue first) OR by Stability ascending when you want to rescue cards that are still close to 80%–89% retrievability before they decay further, rather than spending your entire session on deeply forgotten cards.

Why can reviewing 'Least Overdue First' (Retrievability close to R_d) be superior during an exam crunch with a huge backlog?

Cards near R = 85% can still be successfully recalled in seconds and pushed months into the future, whereas cards at R = 20% will mostly lapse and flood tomorrow's relearning queue.

Why is reviewing consistently for 15 minutes every single day mathematically superior to reviewing for 105 minutes once every Sunday?

Because young cards with Stability S = 1–3 days decay far below R = 90% over a 7-day gap, turning quick 5-second maintenance recalls into slow, frustrating lapses.

Assuming a standard forgetting curve exponent w_20 = 0.5 (equivalent to FSRS-4.5/5's -0.5 power law), what is the constant F = 0.9^(-1 / 0.5) - 1?

F = 0.9^(-2) - 1 = (10/9)^2 - 1 = 100/81 - 1 = 19/81 ≈ 0.234568.

How does FSRS-6 help you plan an 'Exam Countdown Schedule' with a fixed test date?

By matching your new-card completion date so all cards graduate at least 21–30 days before the exam, then raising Desired Retention to 93%–95% during the final two weeks so every card's Retrievability is ≥ 93% on exam morning.

What are the four steps to remediate a Leech Card once flagged?

1. Split complex cards into atomic single-fact cards; 2. Add a concrete mnemonic, visual diagram, or personal example; 3. Use Feynman self-explanation to fix underlying conceptual gaps; 4. Add disambiguating context cues to the front prompt.

What is 'Load Balancing / Fuzzing' across days of the week in FSRS-6?

Distributing scheduled intervals within a narrow ±5% window around the exact target date so daily review counts remain smooth rather than spiking after bulk study days.

What happens in FSRS-6 if you press 'Hard' (G = 2) on a card?

1. Difficulty D increases (making future intervals grow slower); 2. The current stability gain is multiplied by the Hard penalty w_15 < 1 (yielding a modest interval increase rather than a full Good jump).

FSRS-6 & Spaced Repetition — Desired Retention, Interval Scaling, Lapses & Overdue Reviews

Mastering the practical scheduling controls of FSRS-6: deriving next intervals from Desired Retention, workload vs. retention trade-offs, handling backlogs, and optimizing post-lapse recovery.

Level: intermediate
Languages: en → en
Cards: 32
Markdown: FSRS-6 & Spaced Repetition — Desired Retention, Interval Scaling, Lapses & Overdue Reviews (.md)

Flashcards (32)

#	Front (Term / Question)	Back (Definition / Answer)
1	How does the 'New Cards Per Day' limit interact with your future FSRS-6 review workload?	In steady state, each new card introduced per day generates roughly 7 to 10 total reviews over its first year (at 90% retention), meaning 20 new cards/day stabilizes to ~140–200 daily reviews.
2	What is a 'Leech Card' in spaced repetition?	A card that has lapsed (been rated 'Again') repeatedly (typically ≥ 4 to 8 times), consuming disproportionate review time while failing to stabilize in long-term memory.
3	With w_20 = 0.5 (F = 19/81), calculate the interval multiplier m(R_d) if you lower Desired Retention from 90% to 80% (R_d = 0.80).	m(0.80) = (0.80^(-2) - 1) / (19/81) = (1.5625 - 1) / 0.23457 ≈ 2.40—meaning intervals become 2.4× longer than at 90% retention.
4	With w_20 = 0.5 (F = 19/81), calculate the interval multiplier m(R_d) if you raise Desired Retention from 90% to 95% (R_d = 0.95).	m(0.95) = (0.95^(-2) - 1) / (19/81) ≈ (1.10803 - 1) / 0.23457 ≈ 0.4605—meaning every interval is cut to less than half (46%) of its 90%-retention length!
5	What happens in FSRS-6 if you press 'Easy' (G = 4) on a card that was already 30 days overdue?	You receive a double compound boost: (1) the large overdue retrievability bonus from low R, AND (2) the Easy multiplier w_16 > 1, plus a reduction in Difficulty D.
6	Conversely, when is sorting overdue cards by 'Lowest Retrievability First' (most overdue first) useful?	When your backlog is small enough to clear over a few days and you want to maximally exploit the overdue stability bonus exp(w_10 · (1 - R)) - 1 on cards you still remember.
7	State the 'Rule of 7–10×' for estimating steady-state daily reviews from daily new cards.	At 90% Desired Retention, your long-term daily review count stabilizes at approximately 7× to 10× your daily new-card intake (e.g., 15 new cards/day → ~110–150 reviews/day).
8	Using the FSRS-6 interval equation I(R_d, S) = (S / F) · (R_d^(-1 / w_20) - 1), what does I(0.90, S) simplify to when R_d = 0.90?	When R_d = 0.90, the term (0.90^(-1 / w_20) - 1) is identically equal to F, so F / F cancels out and I(0.90, S) = S (the next interval equals Stability in days!).
9	What is the difference between 'Target (Desired) Retention' and 'True (Average) Retention' in a deck?	Desired Retention (R_d) is the threshold probability at which a card becomes due (the floor of its review cycle), whereas Average Retrievability across all cards in your deck is always higher than R_d because most cards were reviewed more recently than their due date.
10	Write the exact FSRS-6 formula that converts a card's Stability (S) and your Desired Retention (R_d) into the next scheduled interval I(R_d, S).	Solving R_d = (1 + F · I / S)^(-w_20) for I gives: I(R_d, S) = (S / F) · (R_d^(-1 / w_20) - 1), where F = 0.9^(-1 / w_20) - 1.
11	State the golden rule for choosing between 'Again (1)' and 'Hard (2)' in FSRS-6.	If your retrieval was incorrect or incomplete, ALWAYS press 'Again (1)'; press 'Hard (2)' ONLY when your retrieved answer was 100% correct but took uncomfortable mental strain or extra seconds to surface.
12	Does capping the displayed Maximum Interval at 365 days cap the internal Stability (S) state variable?	No. Internal Stability (S) reflects your true psychological memory trace; when reviewed at day 365, FSRS-6 uses the elapsed time t = 365 and internal S to compute R(365, S) accurately.
13	Summarize the mathematical relationship between Desired Retention (R_d), Interval Length (I), and Daily Workload (W).	As Desired Retention R_d increases toward 100%, Interval Length I shrinks non-linearly toward 0 and Daily Workload W rises exponentially; R_d = 90% sits at the knee of the efficiency curve.
14	Why do Leech Cards occur even under an optimal scheduler like FSRS-6?	Because no scheduling algorithm can fix a poorly formulated flashcard—leeches are almost always caused by ambiguous prompts, multi-fact lists, lack of conceptual understanding, or interference between similar cards.
15	Why does lowering Desired Retention below 75% often INCREASE total daily study time despite producing longer intervals?	Because when more than 25% of reviews end in a lapse ('Again'), the time spent re-learning forgotten cards and completing extra post-lapse repetitions outweighs the time saved from longer intervals.
16	Why does raising Desired Retention from 90% to 97% cause an explosive, non-linear surge in daily review workload?	Because as R_d approaches 1.0, the interval multiplier (R_d^(-1/w_20) - 1) / F approaches 0, shrinking intervals toward 1 day and forcing you to review almost your entire collection every few days.
17	What is the 'Hard Button Misuse' anti-pattern that learners carry over from SM-2 or bad habits?	Pressing 'Hard' when you actually forgot the card (out of reluctance to press 'Again'): in FSRS-6, rating a forgotten card as 'Hard' falsely tells the algorithm that recall succeeded, inflating Stability S and causing guaranteed future lapses.
18	How does FSRS-6's Maximum Interval cap work, and what is a sensible setting?	The Maximum Interval clamps scheduled intervals to an upper ceiling (e.g., 365 days for a 1-year exam timeline, or 3,650–36,500 days for lifelong learning) so cards are reviewed at least once within your chosen horizon.
19	If you want to cut your daily review time in half next week without changing Desired Retention, what is the single fastest lever?	Temporarily set 'New Cards Per Day' to 0: because recent cards in their first 1–14 days account for the majority of daily reviews, pausing new cards rapidly shrinks the daily queue within 7 days.
20	When should a learner intentionally choose a lower Desired Retention of 82%–85%?	When maintaining a massive lifelong deck (e.g., 20,000+ general vocabulary or trivia cards) under strict daily time limits where occasional lapses are acceptable.
21	What is the 'Optimal Retention Sweet Spot' recommended for 95% of learners in FSRS-6?	Between 85% and 92% (with 90% as the universal gold-standard default), which maximizes total knowledge retained per minute of daily study.
22	If your Desired Retention is set to 90% (R_d = 0.90), what will your overall deck's average Retrievability typically be?	Approximately 94% to 96%, because cards start at R = 100% right after review and decay gradually down to 90% on their due date.
23	Why is the interval multiplier m(R_d) = (R_d^(-1 / w_20) - 1) / F strictly less than 1.0 whenever Desired Retention R_d > 0.90?	Because requiring a recall probability higher than 90% means you must review the card earlier than its 90%-decay point S.
24	When should a learner temporarily raise Desired Retention to 93%–95%?	During the final 2 to 3 weeks before a high-stakes board or licensing exam (e.g., USMLE, Bar Exam, JLPT) on a bounded, high-yield core deck.
25	What is the optimal sort order when clearing a large backlog of overdue cards in FSRS-6?	Sort by Retrievability descending (highest R / least overdue first) OR by Stability ascending when you want to rescue cards that are still close to 80%–89% retrievability before they decay further, rather than spending your entire session on deeply forgotten cards.
26	Why can reviewing 'Least Overdue First' (Retrievability close to R_d) be superior during an exam crunch with a huge backlog?	Cards near R = 85% can still be successfully recalled in seconds and pushed months into the future, whereas cards at R = 20% will mostly lapse and flood tomorrow's relearning queue.
27	Why is reviewing consistently for 15 minutes every single day mathematically superior to reviewing for 105 minutes once every Sunday?	Because young cards with Stability S = 1–3 days decay far below R = 90% over a 7-day gap, turning quick 5-second maintenance recalls into slow, frustrating lapses.
28	Assuming a standard forgetting curve exponent w_20 = 0.5 (equivalent to FSRS-4.5/5's -0.5 power law), what is the constant F = 0.9^(-1 / 0.5) - 1?	F = 0.9^(-2) - 1 = (10/9)^2 - 1 = 100/81 - 1 = 19/81 ≈ 0.234568.
29	How does FSRS-6 help you plan an 'Exam Countdown Schedule' with a fixed test date?	By matching your new-card completion date so all cards graduate at least 21–30 days before the exam, then raising Desired Retention to 93%–95% during the final two weeks so every card's Retrievability is ≥ 93% on exam morning.
30	What are the four steps to remediate a Leech Card once flagged?	1. Split complex cards into atomic single-fact cards; 2. Add a concrete mnemonic, visual diagram, or personal example; 3. Use Feynman self-explanation to fix underlying conceptual gaps; 4. Add disambiguating context cues to the front prompt.
31	What is 'Load Balancing / Fuzzing' across days of the week in FSRS-6?	Distributing scheduled intervals within a narrow ±5% window around the exact target date so daily review counts remain smooth rather than spiking after bulk study days.
32	What happens in FSRS-6 if you press 'Hard' (G = 2) on a card?	1. Difficulty D increases (making future intervals grow slower); 2. The current stability gain is multiplied by the Hard penalty w_15 < 1 (yielding a modest interval increase rather than a full Good jump).