Flashcards work. Making them does not scale. Writing a good card takes a few minutes, most people write them badly — too long, two facts at once, cued by the wording rather than the idea — and the time comes out of studying.
Downloading someone else's deck solves the time problem and creates a worse one. A card written by a stranger carries none of your context. You meet an isolated fact with no memory of the question that made it matter, and isolated facts are exactly what does not survive.
What makes a card worth having
- One idea. If the answer has an "and" in it, it is two cards.
- Cued by the concept, not the phrasing. If you can answer from the shape of the sentence, you have learned the sentence.
- Answerable in a few seconds. A card that needs a paragraph is a topic, and topics belong in questions.
- Traceable. You should be able to get back to where it came from, because the context is half the memory.
The best card is one you would have written yourself, generated from material you have already engaged with.
Deriving rather than authoring
An explanation you have just read already contains the single fact that matters — usually stated explicitly, because a good explanation leads with it. Turning that into a card is extraction, not composition, which is both faster and much safer than asking a model to invent study material.
It also preserves the link. A card derived from question #50097 can carry you back to the stem, the figure and the full explanation when the bare fact is not enough — which is the recovery route a downloaded deck cannot offer.
When a card is the wrong tool
The temptation once card-making is cheap is to make one for everything, and the deck then becomes a second syllabus competing with the question bank for the same hours. Cards are for material that has to be simply available — values, criteria, classifications, drug classes. Anything requiring a judgement between plausible alternatives is a question, and turning it into a card either gives the answer away in the prompt or is too long to drill.
How many is too many
A deck grows faster than the time to review it, and the arithmetic catches people out: every card added is not one review but a review every few days for months. A few hundred well-chosen cards is manageable alongside question practice. A few thousand is a second job, and the usual outcome is that the deck is abandoned and takes the habit with it.
Every card you add is a small recurring commitment, not a one-off. Add them as though they cost something, because they do.
Cards made from your own errors are worth most
A card derived from a question you got wrong carries a memory of getting it wrong, and that is a strong retrieval cue. It is also self-limiting in a useful way: you only make cards where you actually failed, so the deck stays small and stays aimed at your real gaps rather than at the syllabus in general.
Review them in the gaps, not in study time
Cards suit the fragments — waiting for a list, between cases, on a bus. They are short by nature and lose nothing from being interrupted. Question practice needs a clear head and an uninterrupted run, so spending a rare good hour on card drill is close to a waste of it.
Cards and questions do different jobs
Do not replace question practice with cards. A card rehearses a fact you have been told is relevant. A question makes you notice that it is relevant, among four plausible alternatives, under time. The exam asks the second thing. Cards are for the arbitrary material — values, criteria, classifications — that has to be available before reasoning can start.
Our Memory cards feature works this way, drawing a card from the explanation and handing it to the same scheduler as everything else. It is in pilot, not generally available; Bookmarks and Due Today, which cover much of the same ground manually, are live.