Generating ideas with structured techniques — questions

Because they run under opposite rules, and the rules cancel each other out.

What happens when they are mixed is observable within a minute: somebody says "that will never work", and from that point on the group defends ideas instead of producing them. Every further idea is now offered with a justification attached, the wild ones stop coming — and the wild ones are the reason for the exercise, because a wild idea can be tamed while a tame one cannot be enlivened.

So the separation is in time: a diverging hour with its own rules, announced as such, and a separate session later in which the criteria are applied. In this course the second half is the next module, weighted scoring.

Points the answer must contain:

Six people, three ideas each, five minutes per round; then the sheet is passed one seat on and the next person reads what is there before writing. Six rounds, so 6 · 3 · 6 = 108 written ideas in half an hour.

Four properties are built into the procedure rather than left to discipline:

The honest expectation: not 108 good ideas, but perhaps five worth pursuing. The other hundred are the price of finding those five, and rounds three to six produce them — round one is everybody’s obvious first thought.

Points the answer must contain:

The middle column is the part worth remembering, because it is what decides the choice. Three of the six work by association and differ only in the social setting: spoken in a group, written in a group, alone with a structure. Morphologischer Kasten is the only one aiming at completeness rather than inspiration, Bionik is the only one that imports a solution from outside the problem domain, and Delphi is the only one whose purpose is to narrow a spread rather than widen it.

The four rules of brainstorming are worth being able to recite: no criticism, quantity before quality, wild ideas welcome, build on what has been said.

Points the answer must contain:

Break the problem into independent characteristics, list the possible values of each, then read a path across the rows.

Every path is a candidate solution: member + phone + before training + club database is one, trainer + tablet + during + club database is another. The count is the product of the row lengths — 4 · 4 · 3 · 4 = 192 here — which is far more than any group proposes freely. That is the whole point of the technique: it surfaces the combinations nobody would have thought of, because it does not depend on anybody thinking of them.

Two rules keep it honest. The characteristics must be genuinely independent — if choosing "card reader" already fixes the device row, those two rows are one characteristic and the 192 is fiction. And five to seven rows is the practical limit; beyond that the combination count stops carrying information.

Reading it out is not automatic either: most of the 192 are nonsense, and the technique’s job is finished when it has put three or four plausible ones in front of you that would otherwise not have been on the list.

Points the answer must contain:

Brainstorming, 6-3-5 and mind mapping all draw on what the people in the room already know, and organise how it comes out. The other two do something different.

Bionik imports a solution from outside the problem domain. It works through an abstraction: the technical problem is turned into a functional question ("how does a surface stay clean?"), an answer is looked for in a living system (the lotus leaf), and the principle — not the shape — is transferred back. The burdock burr became the hook-and-loop fastener, shark skin became drag-reducing riblet films, termite mounds became passive building ventilation. Skipping the abstraction is the standard failure: copying the shape of a wing gives you a bad wing.

Delphi narrows an assessment instead of widening a field. Experts answer a written question anonymously, the answers are summarised with their reasons and fed back, and everybody answers again. Anonymity removes the effect that the loudest or most senior voice decides; the fed-back summary lets people revise on reasons rather than on status, which is why the spread usually narrows over two or three rounds. It costs days rather than minutes and needs people who actually know the subject.

So: five techniques diverge, Delphi converges; five look inward, Bionik looks outward.

Points the answer must contain:

Because what comes back is the average of everything that has been written about the subject. It is fluent, it is plausible, and it is nobody’s idea in particular — including nobody in your room.

That matters for a reason that has nothing to do with quality: a year project has to survive February, and what carries it through February is that somebody in the team actually wants this thing to exist. An idea the team produced has that attachment; an idea from a list does not, and the list has no way of knowing which of the fifty anybody cares about.

The techniques also produce something the list does not: written raw material with a history. The 6-3-5 sheets show which ideas came from building on which others, the morphological box shows the combinations that were considered and dropped — and that record feeds directly into the proposals and the weighted scoring in the next module.

The reasonable use is afterwards, in the converging half: as a second opinion on criteria, or to check whether an obvious candidate has been missed.

Points the answer must contain: