Generating Creative Ideas

Given a problem, don't answer immediately. Run it through this sequence:

1. Restate the problem 3 different ways (literal, metaphorical, inverted)
2. List the assumed constraints — then ask which are real vs. imagined
3. Generate 10 ideas fast, no filtering (quantity over quality first pass)
4. Combine/mutate the 2-3 most interesting ones
5. Pick the one that's useful AND surprising — not just novel for novelty's sake

Example prompt: "How do we get more people to use the stairs instead of the elevator?"

• Literal: incentivize stairs
• Inverted: make the elevator worse (slower, less convenient) — but this harms some users
• Metaphorical: "how do games get people to choose the harder path?" → gamify stairs (the famous "piano stairs" effect — each step makes a sound, stairs become fun)

Progress:

• Clarify the real problem (not just the stated one)
• Surface and challenge hidden assumptions/constraints
• Diverge: generate many raw ideas without judgment
• Cross-pollinate: combine ideas, borrow from unrelated domains
• Converge: filter for usefulness, feasibility, and surprise
• Stress-test the top idea against the original problem

Step 1: Reframe the problem. Ask "what is this really a problem of?" Strip it to its function, not its form. A "better can opener" is really a problem of "opening sealed containers."

Step 2: Challenge constraints. List every assumed rule ("it has to be digital," "it must cost nothing," "users won't change behavior"). For each, ask: is this actually true, or just how it's always been done?

Step 3: Diverge without filtering. Use techniques to force volume and range:

• Analogical transfer: how does nature/another industry/a game solve this?
• Inversion: what would guarantee failure? Now flip it.
• Combination: force two unrelated concepts together (SCAMPER: Substitute, Combine, Adapt, Modify, Put to other use, Eliminate, Reverse)
• Extreme scaling: what if you had infinite budget? Zero budget? One second to solve it?

Step 4: Cross-pollinate. Take the strongest 2-3 ideas and merge them, or borrow a mechanism from an unrelated field (biology, games, architecture, history).

Step 5: Converge and filter. Pick ideas on: does it solve the real problem, is it feasible enough to try, does it still feel surprising (not the obvious first answer anyone would give).

Step 6: Pressure-test. Ask "why would this fail?" before presenting it. Fix the obvious failure mode or note it as a known tradeoff.

Example 1: Input: "We need a new feature to increase user engagement on our app." Output: Reframe from "add a feature" to "what makes people return to things voluntarily?" → look at habit loops in gaming/social apps → idea: instead of a new feature, introduce a visible "streak" of a simple existing action, paired with a once-a-week surprise (variable reward) rather than constant notifications.

Example 2: Input: "Our office wants to reduce paper waste but people keep printing documents." Output: Challenge the constraint "people must be told not to print" → invert to "what if printing became slightly harder, not forbidden?" → idea: default print queue requires a 10-second manual release at the printer (like badge-release printing), introducing friction without a ban. Borrowed from "frictionless vs. friction-based" behavior design in UX.

Example 3: Input: "How can a small bakery compete with a new chain store nearby?" Output: Reframe "compete on price/convenience" (a losing game) to "compete on what the chain structurally cannot offer" → idea: real-time custom orders texted to the baker, and a visible "name a pastry" contest where customers' ideas get baked — chains can't personalize at that level.