A Spaced repetition memory system depends on the Testing effect: you’ll remember something longer if you have to actually retrieve it from memory than if you’re simply re-exposed to it. So when writing prompts, make sure that the prompt requires the reader to actually retrieve the memory you want to reinforce.
Avoid “leading questions”—questions which imply their own answer—because then reviewers won’t be retrieving the answers from memory. For instance, this is a leading question: “In human-scale situations, is it usually better to ask ‘Is X true?,’ or ‘In what sense in X true?’” The question obviously wants you to pick the latter choice. It’s also bad for several other reasons: it’s too pushy, too confident, too vague about context. A better alternative: “Q: What’s the characteristically meta-rational rejoinder to ‘Is X true?’; A: ‘In what sense?’” One common special case to watch out for: Avoid yes-no spaced repetition memory prompts.
More generally, think of spaced repetition as slowly wearing “grooves” in the mind. When you write questions, you shape the grooves: you’ll reinforce whatever paths light up in a person’s mind when they answer the question. In this sense, true/false questions are weak: they mostly reinforce one binary detail. It’s better to phrase questions to produce vivid, detailed, and specific memory retrieval. Instead of “Which is generally taller: a nimbus cloud or a cumulonimbus cloud?”, you might ask: “Which cloud altitude region(s) do cumulonimbus clouds inhabit?” and “Which is the largest type of cloud?”
Related: Desirable difficulties, after Bjork