Diverge, Then Converge

You now have a shelf full of lateral tools: reframing, dropping assumptions, provocation, random entry. But a tool is only as good as the moment you use it. Reach for a provocation while you’re busy judging every idea and it dies on contact. The tools need a rhythm — a beat that tells you when to open wide and when to close down. That beat is: diverge, then converge. This lesson is about running the two cleanly, in order, and never at the same time.

Two engines: divergent and convergent thinking

Picture two very different mental engines. One fans out: it throws off options, associations, wild tangents, and never stops to ask “is this any good?” The other funnels in: it lines options up, tests them, and cuts the weak ones. You need both. The mistake is running them at the same time.

Divergent thinking is generating wide. Its job is to produce many options: quantity over quality, judgment deferred, wild ideas welcome. This is the movement not judgment stance from the last lesson made into a whole phase — you follow ideas to see where they lead instead of grading them on arrival. Success looks like a long, messy, uneven list.

Convergent thinking is narrowing down. Its job is to evaluate and select: apply rigour, compare against constraints, stress-test, and kill the weak options so one strong option survives. This is ordinary vertical thinking doing exactly what it’s good at. Success looks like a decision.

Neither is “better.” Divergent thinking without convergent thinking is a pile of sticky notes and no plan. Convergent thinking without divergent thinking is picking the first adequate idea and calling it a day. Lateral thinking needs both — the whole point is to widen the field before you narrow it, so that when you narrow, you’re choosing from more than the obvious.

Why mixing them kills ideas: the “yes-but” reflex

Here’s the analogy that does all the work: you can’t drive with one foot on the accelerator and one on the brake. You’ll burn the engine, go nowhere, and wonder why it smells like smoke. Generating is the accelerator; judging is the brake. Press both and you get noise and no motion.

The specific failure has a name: the “yes-but” reflex. Someone offers an idea, and before it’s even finished, a voice — theirs or yours — says “yes, but that’ll never work because…” That “but” is the brake slamming on. And ideas arrive fragile. A good idea rarely shows up fully formed and defensible; it shows up as a weird fragment that would be good after two more steps of development. Judge it at step one and it never reaches step three.

Worked example. A café is losing the morning rush to a slow queue. Brainstorm, judged as it goes:

• “Hire a second barista.” — Yes, but we can’t afford the wages. (killed)
• “Let people pay by phone.” — Yes, but half our regulars are older and won’t. (killed)
• “Let people order the night before.” — Yes, but who orders coffee at 10pm? (killed)

Three ideas, three instant deaths, dead brainstorm. Now run it without the brake, deferring every judgment, and that third line gets to keep moving: order-the-night-before → or order on the way in → a pre-order app that fires the drink when your phone crosses a geofence near the shop, so it’s ready as you walk in. That was the winner — faster queue, no extra staff. It only existed because nobody was allowed to say “but who orders coffee at 10pm?” while the idea was still a fragile fragment. The judgment (“older regulars, geofence reliability, cost”) is real — it just belongs in the convergent phase, applied to a developed idea, not fired at a newborn one.

Brainstorming done right (and the way it’s usually done wrong)

Brainstorming is the classic diverge-then-converge ritual, and it’s usually run so badly that people conclude “brainstorming doesn’t work.” What doesn’t work is the free-for-all where everyone talks over each other and judges as they go. Done properly, it has rules — and they all serve one goal: keep the accelerator and the brake in separate phases.

The real rules of the generative phase:

1. Separate generation from evaluation in time. Generate first. Evaluate later. Draw a hard line between the two — different phase, sometimes a different session.
2. Defer judgment. In the generative phase, no “yes-buts,” no scoring, no “we already tried that.” Judgment is not banned forever, just parked.
3. Go for quantity. More ideas, not better ideas. Quality is a convergent-phase concern. A long list has more raw material to develop.
4. Welcome wild ideas. The absurd ones are cheap to tame later and often carry the reframe that breaks the problem open.
5. Build on others’ ideas — “yes-and,” not “yes-but.” Take a fragment and add to it. “Yes-and” is the accelerator; “yes-but” is the brake wearing a helpful smile.

The honest research caveat: solo-first beats the free-for-all

Now the part most brainstorming pep-talks skip. Decades of research find that naive group brainstorming often underperforms the same people generating ideas alone and then pooling them. Three reasons:

• Production blocking — only one person can talk at a time, so while you wait your turn you forget your idea or stop generating.
• Anchoring — the first few ideas voiced set a groove, and everyone’s later ideas cluster around them instead of ranging wide. The group converges by accident, exactly when it should be diverging.
• Social loafing — in a group it’s easy to coast and let others carry it.

The fix is brainwriting: everyone generates alone and silently first, on paper or in a doc, then the ideas are combined and built on together. Solo-first sidesteps production blocking (nobody’s waiting to speak) and defeats anchoring (you commit your own ideas before you’ve heard anyone else’s groove to fall into). It’s still diverge-then-converge — you’ve just made the divergent phase private so it stays genuinely wide.

Practical drills you can run today

You don’t need a workshop to practise the rhythm. These four drills each force a clean divergent phase before you’re allowed to converge.

Drill 1 — the alternatives-quota

Rule: before you’re allowed to choose any solution, generate N different ones — say five. No picking until you have five. This is a direct antidote to grabbing the first idea that’s merely good enough (the convergent brake slamming on at idea number one). The quota keeps the accelerator down until the field is actually wide.

Mini-example. Your notifications keep interrupting deep work. First adequate idea: “turn on Do Not Disturb.” Fine — but the quota says keep going: (2) batch all notifications to three fixed times a day; (3) move the phone to another room while working; (4) auto-reply “in focus block, back at 3” so people aren’t left hanging; (5) separate work and personal devices entirely. Now you choose — and options 2 and 4 together beat the first idea, because DND alone left people wondering if you were ignoring them. You’d never have reached them if you’d stopped at one.

Drill 2 — assumption-reversal as a generator

Callback to the reframing lesson: list the assumptions baked into the problem, then deliberately reverse one and generate from the reversal. It’s a machine for producing wild-but-useful ideas on demand.

Mini-example. Problem: “the restaurant needs more tables to serve more diners.” Assumption: diners eat at the restaurant. Reverse it — the restaurant comes to the diners — and you’ve generated delivery, meal kits, a pop-up at the office park. The reversal isn’t the answer; it’s the accelerator that gets you off the obvious road.

Drill 3 — “what would have to be true?”

Take a wished-for option you’d normally dismiss as unrealistic and, instead of asking “will this work?” (a convergent, brake question), ask “what would have to be true for this to work?” This keeps a shaky idea alive and turns it into a list of conditions to explore — divergent fuel instead of an early funeral.

Drill 4 — “what problem am I actually solving?”

Before you generate a single option, reframe: “what problem am I actually solving?” Often the stated problem is a solution in disguise. “We need a bigger meeting room” assumes the answer is a room; the actual problem might be “too many people in sync meetings.” Fix the real problem and the option space changes completely. Reframe first, then diverge — otherwise you generate brilliantly wide answers to the wrong question.

The switch discipline: know which mode you’re in

The master skill isn’t divergent thinking or convergent thinking — it’s knowing which mode you’re in right now and not letting them bleed. Two failure modes sit at the edges:

• Stuck divergent forever: endless ideas, no decision. The whiteboard fills, the meeting ends, nothing ships. At some point you have to switch feet and converge — evaluate, cut, choose.
• Converging too early: the very first mediocre idea wins because you never really diverged. This is the far more common failure, and it’s what all four drills above are built to prevent.

Run cleanly and the rhythm is simple: diverge, THEN converge. Open wide with judgment off. Then close down with judgment on. Same two engines everyone has — the lateral edge comes entirely from refusing to run them at the same time.

Where this goes next

You now have the whole kit — reframing, dropping assumptions, provocation, random entry — and the rhythm that makes them productive: diverge with judgment off, then converge with judgment on, cleanly and in order. That’s the engine of lateral thinking running properly.

The final lesson is the guardrail. Lateral thinking is powerful, which is exactly why it’s dangerous when misapplied. Lesson 6 — Where Lateral Thinking Misleads covers when not to reach for it: the problems that just want a careful vertical answer, the moments a “creative reframe” is really just avoidance, and how to tell the difference. Knowing when to switch the tool off is the last thing that turns it from a party trick into a genuine skill.