How Errors Can Lead to Breakthrough Innovations
- by Paul Schoemaker
In 1961, Edward Lorenz, an academic meteorologist, made a small mistake with remarkable consequences. Lorenz was working on weather prediction models using computer simulation in the era when simulations were timely and expensive. One day, Lorenz had completed a large round of simulations and wanted to repeat the experiment over a longer time frame. Rather than waste valuable processing time, he used the printout from the first simulation and manually typed in numbers.
The results of the second simulation diverged radically from Lorenz’s expectation. He backtracked, looking for where he must have gone astray. He puzzled about it for days. Then it struck him: He had entered numbers using a computer printout that rounded all numbers to three places of precision after the decimal. In the computer’s memory, however, numbers were stored to six decimal places. This rounding error of 0.000127 pushed the second simulation onto quite a different path, with vastly different results.
Lorenz was struck by the idea that small changes in a complex system with nonlinear feedback loops could yield such tremendous consequences. As he put it in a 1972 presentation to the American Association for the Advancement of Science: “Does the flap of a butterfly’s wings in Brazil set off a tornado in Texas?” The answer is that, yes, small changes in complex systems can snowball and magnify the initial deviation, the so-called “butterfly effect.”
Thanks to this discovery, Lorenz has assumed an honored place in science as the father of chaos theory. He was awarded the 1991 Kyoto Prize for discovering “deterministic chaos,” insight that shook the foundations of science. If Lorenz hadn’t put in the truncated numbers, he never would have been faced with an unexpected result. If he hadn’t recognized the implications of this result, he wouldn’t have developed chaos theory.
This example illustrates the two prime ingredients of a brilliant mistake:
1. Something goes wrong far beyond the range of prior expectation.
2. New insights emerge whose benefits greatly exceed the mistake’s cost.
Our actions should aim to increase the chance of both occurring together.
All mistakes are not created equal. Some have high cost and offer little learning value, while others cost little and produce deep, valuable insight. Lorenz’s story reveals the type of mistake to embrace rather than avoid, and that mistakes can be beneficial to discovery and may be necessary to arrive at deep, new insights.
Editor’s Note: This post is adapted from Brilliant Mistakes: Finding Success on the Far Side of Failure, from Wharton Digital Press.