The one idea
Big questions are not answerable. Small questions are.
A question like “is AI good for jobs?” cannot be researched, because it contains at least six different questions wearing one coat. Until you separate them, every source you find is answering a different one, which is exactly why they all seem to disagree.
Decomposition is the move: break the question into parts small enough that each part has a findable, checkable answer. Then reassemble.
Split one big question into several small ones you could actually look up.
Scope the question by who, what, where and when, then identify which sub-questions actually change the decision being made.
Separate empirical questions (what is measurably true) from definitional questions (what do we mean by this term) from value questions (what should we do about it). Only the first kind is resolved by evidence.
The mental model
Three kinds of question hide inside most big ones. Telling them apart is most of the work, because each needs a completely different response.
| Kind | Looks like | Resolved by |
|---|---|---|
| Definitional | “What counts as productivity?” | Agreeing on a definition, not by searching |
| Empirical | “Did output per hour rise?” | Evidence, data, measurement |
| Value | “Is that trade-off worth it?” | Judgement, priorities, discussion |
If you try to answer a value question with evidence, you will search forever. If you try to answer an empirical question with opinion, you will be wrong.
The four cuts
Take any oversized question and cut it four ways.
Cutting down 'Should we be worried about AI taking our jobs?'
1 of 5Cut by WHO. Whose jobs? “Jobs” is not a group. Radiologists, warehouse pickers, junior copywriters and plumbers face completely different situations.
Narrowed: our team's jobs — say, six people doing customer support and reporting.
Examples
Vague: “Find out if we should switch to a different payroll provider.”
Decomposed:
- What is our current provider costing us, all in?
- What specifically is going wrong now? (Is there a real problem, or a feeling?)
- What would switching cost — money, time, risk during transition?
- What do two comparable alternatives charge for our headcount?
- What breaks if we switch mid-financial-year?
Notice the second question. Often nobody has asked it, and answering it ends the project early — which is a completely legitimate result.
Vague: “Write about the impact of social media on mental health.”
Decomposed: Whose mental health — which age group? Which platforms? Which outcomes — anxiety, sleep, self-image, loneliness? Correlation or causation? Over what period? What do the strongest studies actually claim, and what do their authors say the limits are?
That is now a paper you can write, instead of a topic you can drown in.
Vague: “Is this health insurance policy any good?”
Decomposed: What exactly is excluded? What is the waiting period for pre-existing conditions? Is there a room-rent cap? What is the claim settlement record? What happens to the premium as I age? Which of these matters most for my situation specifically?
The last question is the one that turns research into a decision.
Try this
Decompose this one before reading further:
“Should our company invest in employee training?”
Write at least four smaller questions, and mark each one as definitional, empirical or value.
Your challenge
Level 3 · IndependentSomeone forwards you this message:
“Can you look into whether we should be using AI for our customer emails? Some of the team are already doing it and I'm not sure how I feel about it. Let me know what you think by Thursday.”
Do not research anything yet. Produce only the decomposition:
- Split it into at least five sub-questions.
- Mark each as definitional, empirical or value.
- Identify which one you cannot answer without asking the sender.
- Say which single sub-question, if answered, would most change the decision.
- Write the one clarifying question you would send back — just one.
The judgement being tested is point 5. A beginner asks six questions and looks confused. Someone experienced asks one and looks like they understand the problem.
What people usually get wrong
- Starting to search before decomposing. The most common failure. You end up collecting material for a question you have not defined.
- Splitting into more parts than anyone needs. Twelve sub-questions is not rigour, it is avoidance. Cut anything that would not change the decision.
- Not noticing the definitional question. If “productivity”, “better”, “safe” or “worth it” is undefined, no amount of evidence resolves anything.
- Trying to research a value question. “Should we?” is decided, not found. You can inform it with evidence; you cannot look up the answer.
- Forgetting the internal questions. Some sub-questions are answered by your own data or by asking a colleague, not by any search engine.
How someone experienced does it
Experienced people decompose out loud, in the reply. Something like: “Before I dig in — I read this as three questions: what it costs us today, what the alternatives cost, and what breaks if we switch mid-year. Is that the right framing, or is there a fourth thing driving this?”
That single message does three things at once. It shows you understood the request, it surfaces the unstated reason behind it, and it stops you spending two days on the wrong question. It also makes you look considerably more senior than saying “sure, I'll look into it”.
They also decide what would end the investigation early. If the first sub-question comes back with a clear answer that makes the rest irrelevant, they stop. Beginners complete the whole plan because it was the plan.
When not to use this
Do not decompose when the question is already small and the answer is one lookup. “What is the GST rate on this category?” does not need a research plan.
Do not decompose in front of someone who is stressed and wants a fast answer. Do it in your head, come back with the answer and the one caveat that matters. The framework is for your thinking, not for performing thoroughness.
Why sources 'disagree' when they are all correct
Most apparent disagreement between good sources is not disagreement about facts. It is two studies answering two different questions and being reported as though they answered the same one.
Two studies on remote work productivity can both be right and reach opposite conclusions if one measured self-reported output for knowledge workers over three months and the other measured call-handling volume for support staff over two years.
This is why decomposition is a verification skill as much as a planning skill. Once you know exactly which question you are asking, you can check whether a source is answering that question or a neighbouring one wearing similar words. That is the whole subject of the next lesson.
Prove it
Take a real question you are facing this week — at work, in study, or a purchase you are unsure about. Decompose it before you research anything.
Keep the decomposition. When you finish researching, look back and check whether the question you actually answered was one of the ones you wrote down. Often it will not be, and noticing that is the lesson.
Keep learning this
Paste this into any AI assistant. It turns the assistant into a tutor that tests you instead of just answering you.
Act as an experienced practitioner who is good at teaching. I have just learned breaking a large question into researchable sub-questions. Assume I am intelligent but relatively new to this — treat me as beginner level. Work through this in order, and wait for my reply at each step: 1. Ask me 5 questions that test whether I actually understood breaking a large question into researchable sub-questions. Do not reveal the answers yet. 2. After I answer, tell me which parts I got right, which I got wrong, and which I only half-understand. Explain only what I misunderstood — do not re-teach what I already know. 3. Give me one practical challenge based on something I could genuinely encounter at work or in daily life. Do not solve it for me. 4. Evaluate my solution the way an experienced person would judge it, including what a professional would have done differently. 5. Tell me what to learn next, and why that comes next. 6. Give me trustworthy sources for deeper study — prefer official documentation, primary research or standards bodies over blogs and videos. Rules for you: no buzzwords. No motivational filler. Say "I'm not certain" when you are not certain, and tell me which parts of your answer I should verify myself. Clearly separate facts from your recommendations and your opinions.
Become independent at this
Use this when you want a path from where you are to actually good, with checkpoints you can test yourself against.
I want to become independently capable at scoping and planning research — not permanently dependent on AI, tutorials or step-by-step guides. Design a progression for me with five stages: Beginner, Guided practice, Independent practice, Real-world application, Professional level. For each stage tell me: - what I must know - what I must be able to do without help - the mistakes people make at this stage - one practical challenge - one real project that would prove I reached this stage - one way I can test myself honestly Then tell me the signals that I am ready to move to the next stage, and the signals that I have skipped ahead too early. Keep the theory to the minimum I actually need. Focus on ability I can transfer to situations you and I have not discussed.