Pre-made flashcard decks optimize for the average learner’s gaps. Yours are not average — they are yours. A card that tests knowledge you already have is a card that wastes your time. A card that skips knowledge you don’t have is a card that reinforces a hidden gap.
Brainback cards are generated from problems you got wrong and concepts you explained back poorly. They are rephrased by Haiku so you can’t rely on surface recognition — but the underlying concept is one you actually failed to demonstrate. Every review is a test of a gap you actually have.
SM-2, in plain English
SM-2 is a scheduling algorithm from 1990. It works. Every recall product worth using descends from it. Here’s how it applies inside Brainback:
- New card: reviewed the next day.
- Rated Good or better: next interval = previous × ease factor. Ease starts at 2.5.
- Rated Again: ease drops, interval resets to 1 day.
- Rated Easy: ease climbs, interval multiplies by a larger factor.
Over weeks, cards you know well drift to longer intervals. Cards you struggle with come back often. The math does the tuning; you don’t.
The review UX
Cards flip on tap or spacebar. Before flipping, you’re asked for your confidence — one to five. That data feeds calibration. After flipping, four buttons: Again, Hard, Good, Easy. Any more granularity and users froze; any less and the scheduling got noisy.
The whole review is designed for one thing: speed. Median time to review a card in Brainback is nine seconds. You’ll do ten cards in less time than it takes to make coffee.