The record that was a shoe

Between 2017 and 2020, marathon times at the top of the sport improved faster than in any previous three-year period. The athletes had not changed much. The shoes had.

A thick, light foam midsole with a stiff plate inside returns more energy per stride than its predecessors. Laboratory measurements put the saving at around four per cent of the energy cost of running, which over two hours is enormous. The effect is not uniform: it is larger for some gaits than others, which means the technology did not simply lift everyone, it reordered them.

That reordering is what makes the governing body's problem hard. A rule banning the shoes would disadvantage athletes who had built a training block around them. A rule permitting anything would make the record a measure of equipment. The eventual compromise set a maximum sole thickness and required that a shoe be available for purchase by anyone for four months before competition, which addresses fairness between athletes rather than comparability across years.

Comparability across years may simply be gone. Historians of the sport point out that it was never as solid as it looked: tracks, timing, pacing, nutrition and course design have all shifted, and each shift produced a smaller version of this argument. What is new is the size of the step and the fact that it is purchasable.

Several statisticians now publish times with an equipment era attached, in the way that swimming times are read differently before and after the suit rules of 2010. It is a modest solution. It admits that a record is a measurement taken under conditions, and that the conditions belong in the record.