In an 1858 humorous poem The Deacon’s Masterpiece, or the Wondeful One Hoss Shay, Oliver Wendell Holmes Sr. described a carriage so artfully constructed as to have no weakest link. The carriage ran smoothly for exactly a hundred years, and then one day

it went to pieces all at once, –
All at once, and nothing first, –
Just as bubbles do when they burst,

leaving its driver sitting atop a pile of rubble and dust.

In a remark accompanying the poem, Holmes discussed how one might craft such a cab:

Observation shows us in what point any particular mechanism is most likely to give way. In a wagon, for instance, the weak point is where the axle enters the hub or nave. When the wagon breaks down, three times out of four, I think, it is at this point that the accident occurs. The workman should see to it that this part should never give way; then find the next vulnerable place, and so on, until he arrives logically at the perfect result…

In some ways, the human body may appear to function like the One Hoss Shay, running smoothly for years before multiple organ systems veer simultaneously toward failure (Nesse 1988). An evolutionary explanation might seem straightfoward: natural selection would play the role of the workman, favoring increases in the durability only of those parts that are first to fail.

However, not all of the body’s subsystems are equally likely to fail. Failure of the heart, or of the tumor suppressor systems that prevent the growth and spread of cancers, cause vastly more human fatalities than do failures of comparably complex and equally critical systems such as the liver or kidneys.

Why do some systems appear to be so vulnerable, while others appear to be “overbuilt” with respect to the stresses placed upon them during the human lifetime? At first glance, one might conclude that this pattern is a classic indication of mismatch between present environmental conditions and those conditions under which our organ systems evolved. Perhaps human hearts today face greater resistance from atherosclerotic arteries; perhaps human lifespans today exceed the duration of protection provided by our evolved tumor supressor genes.

But it would be a mistake to conclude that we require a mismatch explanation to account for these differences in failure rates. In a recent paper published in Oikos, Laird and Sherratt explain why. They lead with a story about Henry Ford – perhaps a recasting of Holmes’ poem – that ought to be true even if it isn’t:

Henry Ford, it is said, commissioned a survey of the car scrap-yards of America to find out if there were parts of the Model T Ford which never failed. His inspectors came back with reports of almost every kind of failure: axles, brakes, pistons – all were liable to go wrong. But they drew attention to one notable exception, the kingpins of the scrapped cars invariably had years of life left in them. With ruthless logic Ford concluded that the kingpins on the Model T were too good for their job and ordered that in future they should be made to an inferior specification.

Humphrey (1983)  as quoted in Laird and Sherratt (2010)

This story about Ford story has been used repeatedly as a metaphor for how natural selection should not “overbuild” some systems while leaving others vulnerable. Indeed it is the case that natural selection cannot act to improve further the reliability of a system that never fails in the first place. It is also the case that if bodily systems failed in deterministic fashion, selection would act only to improve the longevity of the weakest link. But it is simply false that in the presence of stochastic component failures, optimal design will equalize the failure probabilities of various components.

Rather, optimal design should balance the marginal return of increased investment into each system. For example, suppose it were vastly more expensive to incrementally reduce the failure probability of the heart than to incrementally reduce in the failure probability of the kidneys.  In this case, the optimal design – and the one that would be favored by natural selection – would be one in which the marginal return on investment into kidneys and heart are equal. In such a design, the kidneys would fail with lower probability than does the heart. Laird and Sherratt provide both a straightforward graphical explanation of this logic, and a slightly more complicated analysis based on reliability engineering.

As suggested above, the moral of the story for evolution and medicine is that we cannot leap quickly from the observation that failures in a small number of bodily systems results in the majority of human mortality, to the conclusion that these failures arise from mismatch between the conditions of human evolution and the current human environment.

R. M. Nesse (1988) Life table tests of evolutionary theories of senescence. Experimental Gerontology 23:445-453

N. Humphrey (1983) Consciousness Regained: Chapters in the Development of Mind. Oxford University Press.

R. A. Laird and T. N. Sherratt (2010) The economics of evolution: Henry Ford and the Model T. Oikos 119: 3-9