Failure as infrastructure

The PatVal-EU survey asked inventors directly: what became of your invention? Some 9,216 people from six European countries answered, about patents with priority dates between 1993 and 1997. Around 36% of the European patents in the sample never reached any industrial or commercial use. About 13% were licensed out, and some 5% led to the founding of a new firm.

The first reaction to that 36% is a sense of waste. An earlier article in this Journal called an unexploited invention “a complete loss”. The other angle deserves a look too.

Variation and selection

Any system producing novelty works in two beats: it generates many variants, then discards most of them. Biological evolution has worked this way for four billion years. Markets work the same way with firms. Technical creation follows the same pattern.

A large number of variants remains the mandatory condition. A country with ten inventions a year and a 90% exploitation rate stands worse than one with a thousand inventions and a rate of 15%. The arithmetic is simple, and our intuition misses it often, since we judge percentages and lose sight of the numerator.

The price of a single success

Edison tried thousands of materials for the light bulb filament. Every failed attempt narrowed the search space for the next one. The cost of those thousands of failures is the price paid for the one result still in use.

An industrial laboratory’s patent portfolio works the same way. A few solutions reach production. The rest describe what proved impracticable, and that information holds value for anyone taking up the road again ten years later.

A publicly documented failure is worth more than one buried in a drawer. Here the patent shows its original function: publication of the solution, including solutions left unexploited.

The distinction to make

This argument has a limit, and acknowledging it matters. Failures fall into two families.

The first belongs to the nature of the thing: the idea proved wrong, the material gave way, the market was absent. Such failures are part of the mechanism and deserve treating as investment.

The second belongs to infrastructure: the solution worked, and the road to market ran into unclear ownership, a protection fee beyond reach, a missing intermediary. These remain true losses, since selection happened on a criterion other than technical worth.

Public policy holds instruments only for the second family. The first calls for something else: enough people willing to try, plus a professional culture treating the unsuccessful attempt as an ordinary stage of the craft.

We therefore support a double measure of an innovation system’s health: the number of attempts, alongside the exploitation rate. A system with few failures signals, most often, too little courage.