↩ ExploRational · A Catalogue of Divergent Rationality
explorational · a divergent move
How do you find out what you should have been asking? The answering reflex is so strong that the question never gets its turn — state a problem in a room and the room starts solving. The Question Formulation Technique forbids answers entirely: produce only questions, then improve the questions themselves — and the assignment changes under you.
Lawrence, Massachusetts, 1990. Dan Rothstein and Luz Santana kept hearing one sentence from parents who had stopped coming to school meetings: we don't even know what to ask. The technique they built from that sentence — the Question Formulation Technique — is almost insultingly mechanical. A facilitator offers a focus: a statement, never a question. The group storms: as many questions as possible, no discussing, no judging, no answering, every question written down exactly as asked. Then the group studies its own output — which questions are closed, which open — rewrites some against their grain, and ranks three. Built for classrooms; since run in boardrooms, clinics, labs, and city halls.
Notice what the deliverable is. A better question — a different thing than an answer, and for a stuck problem usually the scarcer one, because most stuck problems are answers to the wrong question, held in place by the reflex. The QFT's wager is that the reflex is too fast for good intentions and must be beaten by a rule with no exceptions. The bench below enforces the one rule a machine can check.
the signature instrument
Five stations: focus → storm → transform → prioritize → the sheet. The bench's only power is refusal — entries that fail the form checks go to a visible answers jail, kept and counted. A worked storm is loaded so you can see the machinery; reset it and bring a stuckness of your own. The displayed values are counts of form: entries, rejections, open/closed tallies, rewrites, picks. Nothing grades quality.
The question bench
…
The bench enforces one rule here, and really checks it: the focus must not end in a question mark. Everything else that makes a good focus — flat, concrete, no verdict smuggled in — is yours to keep.
focus
detector, form only — must end in ? · jailed openers: · rough lexical check
Closed buys answerability — it can be settled with a yes, a no, a number. Open buys reach — it cannot be settled yet. Rewrite every closed question open, and close exactly one open question, to feel what each form purchases. The bench re-reads each rewrite live.
Pick the three whose pursuit would change the assignment — one line each on why. Pick order is rank order; pick again to unpick.
the answering reflex
Answering is the performance every institution rewards. School grades answers; meetings reward the first confident fix; a question held open too long starts to read as incompetence. So the reflex is fast, trained, and prestigious — by the time a problem statement has finished being said, three solutions are already queued, each of them an answer to whatever question happened to be embedded in the framing. Nobody chose that question. It arrived inside the sentence.
Rothstein and Santana's insight is that the reflex cannot be negotiated with; it has to be forbidden. During the storm the rules are absolute: no answering, no discussing, no judging — praise included, since "great question" is a verdict too. Any crack and the reflex pours through it. The bench takes the one enforceable rule literally: everything that is not a question, and everything that opens like an answer, goes to jail — kept and counted, never deleted — because the tally is information. How often your hand reached for an answer while answers were forbidden is worth knowing.
what to try
Storm a real stuckness. Reset the bench and state something from your actual week, flat — the bench checks only that it is a statement. Then produce fifteen questions against the clock of your patience. The first five will be your stock questions; somewhere past eight, the interesting ones start.
Watch your jail fill. When an entry gets jailed, notice that it felt like a question while you typed it — that is the reflex from the inside. The tally on the sheet is real: n grabs in m entries, counted and kept.
Transform your best closed question. Rewrite it open — the bench re-reads the rewrite live — then read the two side by side. "Can we fix onboarding?" invites a vote. "What does the second month ask of a customer?" invites a discovery. Same territory; different assignment. That shift is the move.
the question is the deliverable
"Better," for a question, is structural. A better question redraws what the work is: pursue it and you would do different things, involve different people, count different evidence. That is why the transformation pass matters more than it looks. The closed form buys answerability — "Is the standup necessary?" can be settled by a vote before lunch. The open form buys reach — "What is the standup for, here, specifically?" cannot be settled yet, and that unsettledness is what a stuck problem needs, because the stuckness lives in what the current framing forecloses. Neither form is superior; each purchase costs the other. The discipline is choosing the form on purpose instead of inheriting it.
So the storm ends in a ranking, and the ranking is a reassignment. The old task — fix meetings — becomes another one — find out what a meeting is for, here — and it changed while you were forbidden to answer. The top three are the new assignment. Answering resumes tomorrow, aimed differently.
the honest bench
Two detectors, both lexical, both rough, both disclosed. The answer-detector checks that an entry ends in a question mark and does not open with one of eleven declarative frames — the exact list is printed under the input, rendered from the same array the code checks. The open/closed classifier matches a question's first word against two short lists, plus three opening phrases. That is all. "Right?" sails through; "How so." goes to jail. The detectors enforce form — and form is the discipline the whole technique runs on, which is why a dumb check is enough to change behaviour. Every number on the bench is a count: produced, jailed, open, closed, rewritten, picked. The one judgment that matters — which question deserves your week — stays where it belongs, with you, at the ranking.
the move ↔ the bench
| The move | The bench |
|---|---|
| the focus | state the stuckness as a sentence, because a question would already be aiming the storm. |
| a stormed question | one reopening of the stuckness — one direction the assignment could move. |
| the answers jail | the reflex, caught in the act and counted; kept visible, never deleted. |
| closed form | answerable, narrow — it can be settled with a yes, a no, a number. |
| open form | reaching, unanswerable-yet — it cannot be settled, which is its value. |
| the top 3 | the new assignment — what the work is now, pending an owner and a first move. |
how this opening fails
risk · endless deferral
The catalogue's named risk for this move. Question storming can become the way a group avoids ever answering anything — there is always a finer transformation, a fresh storm, a re-ranking, and the deferral feels rigorous the whole time. The technique's own shape refuses this: it ends in a ranked three, and the three exist to be pursued. A storm that closes without an owner and a first move is theater with excellent production values. The sheet flags an unowned top question for exactly this reason — presence checked, nothing more.
risk · a detector you can dress for
Add a question mark to a verdict and the bench waves it through: "Right?" passes, and so does "Don't you think we should just ship it?" — a verdict phrased as a question, opening with a word the frame list doesn't watch. The detector catches form because form is what a lexical check can catch; eleven frames are not a mind, and the same roughness will occasionally jail a question that opens like a verdict. Both errors are the price of a rule simple enough to enforce. The rule's spirit — during the storm, produce nothing that already knows its answer — is kept by you or not at all.
Hold the answers in jail until the question has finished becoming itself.
can you use it?
RECOGNITION — Which is question storming? A: brainstorming answers to a fixed question. B: generating only questions — answers forbidden — until the question itself gets sharper. C: ranking the team's proposed solutions.
B. Producing and improving questions before any answering, because a stuck problem is often an answer to the wrong question. A and C both answer.
THE NEAREST NEIGHBOR — Surfacing the assumptions also probes beneath a problem. What separates question storming from it?
Output. Question storming produces a ranked set of better questions; surfacing the assumptions produces a map of premises under a given claim. One improves the question, the other exposes what a claim rests on.
PRODUCTION — Your team is stuck on 'how do we increase retention?' Produce four questions — no answers — that reframe it, and mark which is sharper than the original.
A version: Who churns and when? What did the ones who stayed have in common? Is retention even the metric, or is it activation? What would make leaving feel like a loss? The third is sharper. Yours works if at least one question challenges the original's framing rather than its scope.