Prototyping and Testing
Prototypes Answer Questions
A prototype is a physical or virtual artifact built to resolve a specific uncertainty. The type should match the question:
- Proof-of-concept / looks-like / works-like - separate prototypes for form, function, and feel.
- Breadboard - quick, cheap, functional, ugly.
- Alpha / beta - progressively closer to production for internal and external evaluation.
- Virtual prototypes - simulations, useful when physical iteration is slow or dangerous.
Testing discipline: define the test objective, the measurement method, the acceptance criterion, and the sample size before running. Distinguish destructive from non-destructive tests, and accelerated life testing from single-point functional tests.
Traps
- Confusing prototype fidelity with validity. A beautiful prototype can still test the wrong thing.
- Sample size of one. A single unit passing does not establish reliability; variation between units is real.
- Test-article bias: the prototype is hand-built and better than production will be.
- Changing the test after seeing results - that converts a test into a search for confirmation.
- Ignoring the environment: testing at room temperature when the product ships to −20 °C.
Rule of Thumb
Test the riskiest assumption first. Rank uncertainties by (probability of being wrong) × (cost if wrong) and prototype the top item. Building the easy part first feels productive and teaches little.
Worked example: A team is unsure (a) whether the electronics fit in the housing and (b) whether the latch survives 10,000 cycles. (a) is cheap to check with a cardboard mock-up; (b) needs a cycling rig and real material. Do (a) today, start (b) immediately because it has the long lead time - lead time, not just risk, orders the queue.