Generating a Quiz
Test exactly the places most likely to be confused
Write 8 questions from the material below: 4 single-choice, 2 multiple-choice, 2 short-answer. Requirements: 1) only test details that are easy to confuse or easy to miss — no rote numbers; 2) give the answer and a one-sentence explanation for each; 3) mark what each question tests (recall / understanding / application). [Paste material]
How to use
Answer before reading the explanations — the ones you miss are the ones you keep.
Good for
- You have read the material once, the exam is in three days, and you want to find what you think you know but do not
- You are onboarding a team and need a baseline test that surfaces the usual confusions
Not for
- The material is under a page (a 200-word product note), so eight questions can only be repetition or invention
- It is a domain where you cannot judge whether an answer is right and have no source to check, so wrong items get baked into your notes
Example
Input
Cache-Control: no-store means the response must not be stored; no-cache means it may be stored but must be validated with the origin server before each reuse. max-age and s-maxage both set a freshness lifetime, but s-maxage applies only to shared caches (CDN, proxy) and overrides max-age.
What you get
Questions should cluster on no-cache versus no-store and on s-maxage having no effect in the browser, not on the unit of max-age. Each answer's one-line explanation should point back to the sentence in the source, and the ability tags should separate recall (directive names) from application (pick a directive for a CDN scenario). At least one distractor should be wrong by a single word.
Common mistakes
- Using the quiz instead of the reading: answer without the source, score well, assume you know it
- Asking for harder questions until they test trivia or numbers, which is memory noise rather than confusion, and you end up muddier
Try tweaking it like this
- Change it to 'only the items I will get wrong, with the wrong option I would likely pick and why': produces a mistake log directly
- Split it into four questions before reading and four after: same concepts, so you can see what the reading actually changed