Fortune #38

James Dyson was frustrated that vacuum bags clogged and caused suction to fall. After seeing an industrial cyclone at a sawmill, he wondered whether the principle could be reduced for a vacuum cleaner. The answer did not emerge from one sketch. He built 5,127 prototypes across several years; the first 5,126 were not yet the product he wanted.

The number is often used only as a lesson in persistence. More importantly, each prototype had to produce information. Dyson changed airflow, form, materials, and components, observed the result, and carried the error into the next version. Repeating an identical attempt five thousand times would not have created the same learning.

Perfection was replaced by a visible loop: build, test, observe, and correct. Releasing an experiment from his hands early enough allowed reality to help. Standards still mattered, but they defined what to test rather than becoming a reason to postpone every beginning until certainty appeared.

Practice one skill again while changing only one variable: shorten the opening, rehearse one transition, or ask one audience for feedback. Record the result without judging your identity: “This method caused…” instead of “I am bad.” A mistake becomes useful when it points toward the design of the next prototype.