Edison called failure data

Decision-Making

Thomas Edison’s search for a filament for the lightbulb is quite inspiring. After trying two thousand different materials, his assistant complained, “All our work is in vain. We have learned nothing. Not sure if we can use electricity properly.” Edison’s reply suggests that their failures had not gone to waste: “Oh, we have come a long way and we have learned a lot. We know that there are two thousand elements which we cannot use to make a good light bulb.” We know the ending of this story. According to Edison’s account, “Before I got through I tested no fewer than 6,000 vegetable growths, and ransacked the world for the most suitable filament material. . . . The electric light has caused me the greatest amount of study and has required the most elaborate experiments. I was never myself discouraged, or inclined to be hopeless of success. I cannot say the same for all my associates.”

Edison’s assistant saw a heap of useless filament materials. Edison saw fewer places left to look. The difference was not optimism. It was whether each disappointment removed an option.

That distinction gets butchered whenever failure is automatically renamed “learning”. Failure produces knowledge only when somebody can state what is now known and alter the next attempt. Repeat the same conditions and the same action, then receive the same outcome: nothing has narrowed. Calling the repetition experimental does not make it scientific.

Edison’s search was brutally physical. A candidate fibre had to be found, prepared, fitted and subjected to electricity. It burned, broke or behaved badly in a specific way. The bench accumulated rejected materials; the remaining search narrowed. Midway through, nobody had the comfort of knowing which plant would eventually work. There was only another piece of material and another possible reason it might fail.

People often preserve the motivational half of this story and throw away the discipline. They celebrate resilience and refuse to close an avenue that has produced the same result repeatedly. The useful response is less cheerful. It names the assumption that failed, the condition under which it failed and the option that will no longer consume time or money. Then that option is removed. The discarded fibre remains on the bench.