Flashcards are terrible at inventing understanding and excellent at maintaining it. Use the study guide to build the schema — then hand off weak joints to spaced retrieval so the algorithm schedules memory, not fog.
A familiar failure: open a dense guide, highlight everything, make forty cards, schedule them, feel responsible, fail a comparison question that the guide explained in paragraph two.
The handoff failed. Cards were asked to do encoding. Guides were treated as a card mine instead of a schema builder.
Mayer's pretraining idea and Sweller's load limits both support a sequence: clarify → generative encode → atomize weak points → space → transfer. Skip the middle and the algorithm schedules fog. FSRS cannot rescue a deck built from highlighter ink without a model behind it.
Why the handoff exists at all
Guides and lessons build stories: causal chains, part–whole structure, boundary conditions. Cards store atoms: cues that demand production of a piece of that story under spacing.
Confuse the jobs and you get trivia with timestamps — or a beautiful guide that never returns to memory. The handoff is the skill that connects clarity to durable access. Chi's self-explanation work and Fiorella & Mayer's generative strategies sit upstream of any scheduler: organize and integrate first, then maintain.
The handoff checklist
| Before cards | Gate | If you fail the gate |
|---|---|---|
| Read for clarity | Can state core claim | Keep guiding / self-explain |
| Generative note / map | Organize parts + links | Don't atomize yet |
| Teach-back 60–90s | Production without page | Repair, don't card-spam |
| Identify weak joints | 5–12 high-value cues | Prefer quality over quantity |
| Rate & space | Honest retrieval | Add mixed questions later |
Cards store atoms; guides build stories
If the story isn't yours yet, cards become trivia with timestamps. Earn the handoff with teach-back. The gate is production, not 'I finished the PDF.'
What to turn into cards (and what not to)
Card: definitions you must produce, easy-to-confuse pairs, mechanism steps, boundary conditions, diagram labels you forget, when does this rule fail?
Don't card (yet): entire essays, every example, anything you can't explain, system-level items before parts exist
Higher-order cues transfer better than one-word telegrams — write cards that ask why/compare/apply when the exam will. Atomic does not mean trivial. A comparison cue that forces discrimination is often worth five isolated cloze deletes.
A 35-minute guide → deck session
- Guide section + self-explanation (12)
- Closed-book teach-back (3)
- Fix blanks in notes (8)
- Create 6–10 cards on blanks and twins (8)
- Immediate quick retrieve once (4)
Then walk away. Let spacing work tomorrow. The urge to make forty more cards is usually anxiety, not pedagogy — Sweller would call that extraneous load dressed as productivity.
Highlight → card pipeline
Highlighting is selection, not encoding. Cards made from highlights without teach-back often train recognition of your highlighter, not exam transfer.
Common handoff failures
Carding mid-first-read splits attention between understanding and cue craft. Mark future cards lightly; build the deck after teach-back.
Carding every sentence creates bloat. High-weight, high-confusion, high-reuse ideas first — Hattie's spirit applied to your own syllabus: effect size of the card, not volume.
Skipping mixed questions after a green deck leaves cue dependence intact. The handoff continues into transfer practice — it does not end when the deck is created. Bjork's retrieval strength can look high for familiar cues while storage for transfer stays thin.
After the handoff
Protect retrieval blocks. Interleave related topics. Sit mixed questions so cue variety grows. If papers expose misconceptions, return to the guide — don't only batter the card.
A chronic Again card is often a thin schema or a bad cue. Re-encode or rewrite before you schedule more punishment. Spacing is a retention tool, not a substitute for diagnosis.
Return tickets
The guide is not a one-way airport. When transfer fails, fly back for clarity. Cards maintain; they do not replace diagnosis.
How this fits the wider system
Clarity (guide) → encode (notes/map/teach-back) → retrieve (cards) → transfer (mixed questions/papers). Each stage has a job. Asking cards to invent the schema is like asking a calendar to write the essay.
Pretraining (Mayer) installs names and parts so later germane load can build relations. Generative encode installs the relations. Cards schedule access to the joints that still slip. Papers test whether access survives novel phrasing.
Quality control before you hit save
Before a new card enters the deck, ask:
- Can I teach the parent idea without this card? If no, encode more first.
- Does the exam ever ask this demand (define / why / compare / apply)?
- Is there a near-twin that needs a comparison cue instead of two isolated definitions?
- Is the answer key scorable by a stranger in under thirty seconds of reading?
Four nos should delete the card draft. Four yeses earn a slot in FSRS. The handoff is a filter, not a factory.
When to reopen the guide after cards exist
Reopen when:
- Chronic Again after honest study
- Papers expose a misconception cards never asked
- Comparison items fail while definitions are green
Do not reopen only because anxiety wants another highlight pass. Targeted return beats ritual rereading. The guide earns another visit when production fails, not when vibes wobble.
Guide then cards
Sukrat Study Guide builds clarity; Flashcards maintain retrieval on syllabus concepts — use them in that order. Guide: /products/features/study-guide · Cards: /products/features/flashcards
Stop rates that mean the handoff worked
You will know the handoff is healthy when: teach-backs precede new cards, deck growth is modest, mixed questions improve, and chronic Agains trigger re-encode rather than rage reviews. You will know it failed when the deck balloons after a highlight binge and papers still surprise you with cousins of "easy" cards.
Track one crude ratio for two weeks: new cards created after a teach-back pass versus new cards created from highlights alone. If the second number dominates, the handoff gate is theater. Restore production before atomization — Mayer's pretraining and generative encode are not optional garnish for FSRS.
A healthy handoff also shows up as quieter Sundays: fewer emergency card factories, more scheduled mixed sets, and fewer papers that feel like strangers wearing your deck's vocabulary.
FAQ
Can I make cards while reading?
Light future-card marks yes; full card creation mid-first-read often splits attention. Better after teach-back.
How many cards per guide section?
Often under a dozen high-value cues. Bloat is a smell. Prefer fewer higher-order cues over fifty shallow clozes.
What if teach-back is fine but cards still fail later?
Normal forgetting — space them. If failure is conceptual, the schema was thinner than the teach-back suggested; repair notes.
Should every guide section become a deck?
No. Prioritize high-weight and high-confusion sections. Some clarity only needs a teach-back and a mixed question later.
When is the handoff "done"?
Never permanently. It is done for this week when teach-back passed, weak joints are carded, and a mixed set is scheduled — then the loop restarts when papers or Agains expose new gaps.
Related reading
- Clarity Before Practice
- Pretraining Principle
- How to Write Flashcards That Transfer
- When Flashcards Aren't Enough
- Self-Explanation While You Read
- FSRS vs Cramming
Keep going across Sukrat
Earn the deck
Guides create understanding. Cards maintain access. Confuse the jobs and you will own a busy scheduler and a surprised exam face.
Teach it back. Card the weak joints. Space honestly. Remix with papers.
The handoff is a skill — practice it until it feels obvious.