I made 300 flashcards and reviewed 40 of them

Sophomore year, psych midterm. I spent an entire Sunday building a deck. Three hundred cards, every term from six chapters, color coded, beautiful.

I reviewed about forty of them. Once. The deck was so big that opening it felt like starting a project rather than doing a task, so I kept not opening it, and by Thursday it was easier to just reread the slides.

The Sunday wasn't wasted, exactly — typing things out is a mild form of processing. But I'd spent seven hours on transcription and called it studying, which is a specific kind of self-deception that flashcards are unusually good at enabling.

The two failure modes

Almost every dead deck died of one of these.

Cards too big to grade. The front says "the central limit theorem." The back is a paragraph. You look at the front, think "yeah, sampling distributions, normal, something about n," flip it over, read the paragraph, think "right, that's what I said," and click through.

But you didn't say that. You said a blurry fraction of it, and because there was no specific thing to be wrong about, you couldn't fail the card. A card you can't fail teaches you nothing — it just runs the recognition loop that made rereading useless in the first place.

Too many cards. A deck you can't get through is a deck you stop opening. Eighty cards reviewed four times each beats three hundred reviewed once, and it isn't close, and the eighty-card deck takes an hour to build instead of a day.

What a good card looks like

One question, one answer. If the answer has three parts, that's three cards. Yes, this feels like it's making more work. It isn't — it's making the same work gradeable.

The front is a real question, not a topic. "Central limit theorem" is a topic. "CLT: what becomes normal as n grows?" is a question, and it has exactly one right answer, and you'll know immediately whether you had it.

The answer should be short enough to say out loud. If you can't say it in a breath, split it.

Use your own words. A card copied verbatim from the slides gets memorized as a string of words. You'll recognize the phrasing and feel fine, then fail to produce the idea when the exam words it differently.

Cards for relationships, not just terms. The best cards in a deck are usually the ones that aren't definitions:

• What's the difference between X and Y?
• When would you use X instead of Y?
• What breaks if condition Z doesn't hold?
• Why does X follow from Y?

Those are the ones that show up on exams. A deck that's a hundred percent definitions is a deck for a vocabulary quiz.

What not to make cards for

Anything you understand but can't do. Cards test recall. If the problem is that you can't execute the procedure, you need practice problems, not cards. This is the single most common misuse — people make flashcards for calculus, and calculus is not a recall problem.

Anything you'll never be asked. Not every bolded term is exam material. If you're making a card for every bold word, you're transcribing the textbook and the deck will be too big to use.

Things you already know cold. Feels good, wastes review time. Cut them.

The number

For a typical unit, I'd aim for something like 15 to 25 cards per week of material. Across a semester that's maybe 150 to 250 per course — but you're building them incrementally, reviewing continuously, and never facing a wall of three hundred on a Sunday.

If you're much above that, you're carding things that aren't recall problems.

Build them at the right time

Not the night before. Not all at once.

Make cards when you're already reviewing — during the same-day review after a lecture, when you've just found out what you didn't remember. Those gaps are exactly what should become cards, and they're identified for free by the review you were doing anyway.

That's the good version of the workflow: try to recall, find the holes, card the holes. Rather than: transcribe every term in the chapter and hope.

About the apps

Anki does real spaced scheduling — each card gets its own interval based on how you did. Free on desktop and Android, paid once on iOS. It's the right tool if you're carrying a lot of cards across a long time.

Quizlet is what most people already have and it's fine for drilling a set. Two things to know: it removed its actual spaced-repetition mode in 2020, and Learn and Test went behind a subscription in 2022. So a Quizlet set is a stack of cards, not a scheduling system, and it's worth being clear about which one you're getting.

And a warning about pre-made decks: there are thousands for common courses and most of them are somebody else's cards from a different professor's version of the course. The making is a real part of the value — you're deciding what matters, which is a form of processing. Using a stranger's deck skips that and gives you cards calibrated to an exam you're not taking.

The version I'd give my sophomore self

Twenty cards a week. Short answers. Made from the things you got wrong during review, not from the bold terms in the chapter.

That deck is boring, small, and finishable, which is why you'll actually open it in week nine. The beautiful three-hundred-card deck was a Sunday I spent feeling productive.