The resources that provide robustness in times of exigency appear excessive in times of ease.

Efficient means without excess. Efficient cars use no more fuel than the transit requires. Efficient construction uses as little material as the structure needs. Efficient corporations have employees who are never idle. Efficient crops produce the minimal amount of unharvested material.

Efficient means operating at the edge of feasibility. Efficient cars carry around no excess material in their chassis and so are too weak to sustain an unexpected load. Efficient construction uses no excess support and is too fragile to survive earthquakes or weather outside those anticipated by the designer. Efficient corporations have no excess capacity and are unable to cope with significant changes in demand, employee base, or supply. Efficient crops waste no energy in facilities to handle unexpected disease, pests, or weather.

Efficient means brittle. If something is strong enough to handle a shock that never comes, it would be more efficient if the resources creating that strength were reinvested.

In statistics, this principle has a name: “‍overfitting‍”. The function that best fits a set of datapoints will almost inevitably fail to fit any datapoint outside that set, and to prevent this overfitting requires creating a less-precise fit to the original data.

The current supply chain problems are a large-scale example of this problem. The US economy had spent decades becoming more efficient, and along the way also becoming more brittle. When a large shock was administered this brittleness was manifest by some systems failing, resulting in more shocks and more failures. This was not inevitable: if the shock had been smaller than the net surplus capacity of the system then the failures of some companies would have been covered by robustness of others, and some of that has occurred, but the overall size of the current shock is beyond what the system can handle well.

This is hardly a new problem. All systems everywhere, human and nonhuman, compete, meaning the most efficient wins—until a shock to the system gives the most efficient a major setback, letting the more robust reassert dominance.

I don’t have a solution to offer with this observation, but I still felt to make the observation because I’ve been repeatedly impressed by how many people I hear who speak as if they’d never considered this law. And law it is: it can be found in the very definitions of efficiency and robustness.

Speaking of law, one of the functions of corporate law is to create regulations intended to tame the worst parts of competition and require robustness. I saw this when working in a nuclear power plant many years ago where regulations meant many of the smaller systems had ample redundancy and could withstand huge disruption, but the largest systems were unregulated; I saw management make them competitive, rather than robust, decisions on the integrity of a turbine blade that resulted in a catastrophic failure and more than a year of repairs before the plant could operate again. We see this in banking, where after each crisis regulation increases the minimum asserts banks must retain and then gradually that is lessened as memory of the last shock fades. We see this in traffic laws, where the most efficient actions for one motorist are unsafe, meaning unable to cope with unlikely road events.

I believe—indeed, I belong to an entire religion that believesThis church-wide belief is announced publicly at least once a year; see the last sentence of this year’s report.—that robustness is more important than efficiency. I hope the current difficulties are resolved soon. I also dream that we might, by some miracle, find in the current difficulties a change of values, putting robustness above efficiency a little more in the future.