Separate What to Memorize from What to Look Up
Added 2026-11-05
Not everything deserves memory; some only needs a pointer.
I am learning this field: [field or subject]. Split what I need into two groups: 1) must be in memory because it cannot be looked up on the spot — judgment frameworks, interface semantics, common counter-examples; 2) fine to look up, so all I need is where to find it — specific parameters, commands, values; 3) give the criterion for each: frequency of use, cost of getting it wrong, cost of looking it up; 4) flag what I may be misremembering — details that become risky once memorized; 5) give a minimum memory set of no more than twenty items plus a lookup index; 6) say how often this split needs revisiting.
Site original
Chinese original
Replace these placeholders 1
Swap the bracketed parts for your own details before sending.
[field or subject]
How to use
Memorize the frame, look up the detail; a wrong detail costs more than a lookup.
Try tweaking it
- Ask for one diagnostic question first to locate where you are stuck.
- Replace "explain it" with "give three progressive exercises; correct me after I answer".
What you should get
- A diagnosis of exactly where you are stuck
- A small set of self-checkable exercises
Common mistakes
- Asking it to explain from scratch, and still not knowing where you went wrong.
- Memorising the explanation as the answer, then failing on any variant.
How to check the output
- It names exactly where you are stuck instead of re-explaining everything.
- Exercises are graded and self-checkable.
Where it works
- Good for: diagnosing where you are stuck, then giving the smallest exercise instead of a lecture.
- Watch out: ask it to flag uncertainty rather than memorising a confident but wrong explanation.