1. A before-and-after
Weak: something is off at home lately, it feels strange, can you tell me what it is?
Better: Question: there is a persistent sour smell in the kitchen — what is the most likely source?
Clues:
· Duration: about two weeks, worse in the evening
· Location: near the sink, close to the base of the cabinet
· Smell: sour, not sharp
· Change: stronger while cooking, weaker with the window open
· Ruled out: rubbish is emptied daily; the drain has been cleaned
Every phrase in the second version narrows the space: timing, location, character of the smell, how it changes, what is excluded. Credibility comes from checkable specifics.
2. The checklist
- Time — how long? continuous or intermittent? a fixed window?
- Location — where exactly? how large an area? tied to an object or room?
- Frequency — how often? is the interval regular? (regularity often points to a mechanism or a program.)
- Change — what makes it worse or better? window, appliance, day vs night?
- Exclusions — what you have already confirmed it is not. This is powerful and usually under-used.
- Context — any recent change: move, renovation, new device, season.
3. Three ways to raise quality
Replace adjectives with numbers. "Quite hot" is worse than "about 45°C, hot to touch but brief contact is fine"; "often" is worse than "2–3 times a day".
One clue per line. It lets the inference quote each clue and lets you spot what is missing.
Phrase the question as one sentence. "What is most likely causing X?" works better than "can you look into this".
4. Why exclusions matter so much
Ruled-out items bound the search space. Doctors and engineers both ask "what has already been checked or replaced?" for the same reason: removing one common cause sharply raises the probability of the rest.
5. Then run it
Once the checklist is done, take it to the bench. To understand what happens next, see the method; unfamiliar terms are in the glossary.