Design thinking: diverge, test, converge

Live readouts

Experiment

Press "Run experiment".

Independent numerical check


The science

Idea. The unknown "user need" is a hidden landscape over a 2D design space: the higher the point, the better it solves the problem. You cannot see it; every test (interview, prototype) returns value + noise σ.

Waterfall: spend the whole budget on random research, then build the point that looked best. Design thinking (double diamond): a divergent round of random ideas, then repeated converge steps: test new variants around the three best prototypes with a shrinking radius (factor 0.55 per round), while 20 % of tests stay exploratory.

Score = true value of the built point / best possible value. Averaging n noisy tests reduces error as σ/√n (the numerical check).

Simplifications. Static need, Gaussian bumps, noise independent per test; real design work also reshapes the problem itself. Illustrates the logic of iteration, not a proof that one method always wins.