brainstorm

A good answer starts with the right question. Stuck thinking is usually a problem framed wrong. A first idea gets mistaken for the only idea, or a simple need arrives dressed up as a hard one. The skill acts as a thinking partner. Pull the real question to the surface and open the space of answers wide before judging any of them. Convergence comes last, onto the simplest direction that works.

The order does the work. Frame before generating, and judge only once the options are on the page; the plainest option that survives wins. Converging early and worshipping complexity are the two ways this thinking goes wrong.

Steps

1. State the problem behind the problem. Write the problem as one sentence, then ask what outcome the user is after underneath the first-named solution. A request for "a faster cache" may be a request for a faster page. The step ends once the underlying question is written down and the user confirms that solving it would be enough.

2. Diverge: generate options without judging. List at least three distinct directions, holding back every objection while the list grows; a critique voiced now kills the option it lands on before it is understood. Reach for the thinking moves (first principles, inversion, the 5 whys, constraints as a lever) when the list stalls below three. The step ends once three or more options that differ at the root sit on the page, unjudged.

3. Probe for the trivial version. Ask what would make this problem trivial: the constraint that, if dropped or added, collapses the work to almost nothing. A reframed problem often beats the best solution to the original. The step ends once the cheapest framing is named, even when the answer is that none collapses it.

4. Converge: pick the simplest option that works. Score the options on what the framed question needs, then choose the one that meets it with the least moving parts. Simplest-that-works ties to least-code: smallest thing covering the need in full, never a subset that only demos. The step ends once one direction is chosen and the reason it beat the runners-up is stated in a sentence.

5. Stress-test the choice by inversion. Ask what would make the chosen direction fail, and name the one assumption it rests on most heavily. A choice with no known failure mode has not been examined yet. The step ends once the top failure mode is written and the choice either survives it or is swapped for one that does.

6. Stop on a clear next step. Convergence is done when a concrete next action exists (a file to open, a spike to run). Thinking past that point is analysis paralysis. The step ends once the chosen direction and its first checkable next step are both written in one or two lines.

See also: the thinking moves for the techniques, the failure modes, the red flags, and a worked example.