My notes

• Brevity
  • If I don't understand it, I can't break it down.
  • Set 20 card review limit
  • Max 9 words for most cards. At absolute max, 3 logical bullets with 18 words max
• Redundant cards may be good (eg. reverse a card)
• Name it for the purpose
  • E.g., not "length of whale intestine" but "animal fun facts". (otherwise, when the time is ripe to impress your friend with an animal fun fact, you might not even recall that you have a related Anki card)
  • "when I hear/read the name somewhere, I want to quickly recall what his basic deal was” – if the expected "real-life prompt" demands a longer card, consider making one.
  • put all info you want to remember in the answer: 24% example (It's more likely that you'll ask yourself, "What do I know about the history of agriculture?" than "What do I know about 1950s agriculture in Germany?".)
  • Recalling "Victorian period" is useless without remembering what century this refers to (was it the 18th or the 19th?),
  • "cloze deletion" may be a bad idea bc the recall may be anchored to that visual thing and not meaning
• I don't tend to make many two-bullet cards anymore, at least.
• 2-bullet cards are usually faster to recall as 1 bullet (if not too long).

Comments

• I probably agree with Nielsen/Matuschak that you should make "more cards than you think you should." While there are diminishing returns, there are also diminishing costs. A tightly interconnected web of facts has better retention than isolated ones.
• Redundancy is a Feature: Hard agree with this point you make. Cards reinforce the pattern "circumstances → solution." Multiple cards with slight prompt variations (epsilon changes) help you recognize solution spaces more generally. This is underappreciated.
• Cloze Deletion: Format vs. Application I disagree that clozes are bad. Every card type is technically cloze deletion (if there is any point in getting pedantic). Your real issue therefore (I assume) isn't the format but excessive context, which leads to remembering "visual shape → answer" instead of "semantic meaning → answer."