The Club's Results

On the topic of variable crew performance, it is perhaps worthwhile to note that some amount of variability day-to-day might actually make it easier for the faster crews to bump up. Bumps is unusual among races in that it has a non-zero threshold for success, and certain pathological possibilities arise from that fact: one might, for example, imagine a division in which each crew was slightly faster than the crew above, though not quite enough to bump, and thus the headship crew could be 20 or so splits slower than the sandwich boat. In a world where crew performance was perfectly consistent, such pathological rankings would in fact be stable over the week. But in reality, where crew performance is stochastic, that very stochasticity helps to ensure that smallish differences in speed rise to the threshold needed for a bump.

It may be counter-intuitive that a random element can help to give more accurate rankings. Certainly it doesn't help when the random element is very large; in the lower divisions for example the rather high frequency of reciprocal bumps might suggest an excess of randomness. But where we are, in the upper half of Div II, reciprocal bumps are quite rare, while bumps in general are more frequent than in the divisions above, and row-overs more rare. Insofar as one expects day-to-day consistency to increase with skill in the higher positions, one wonders if we are near the "sweet spot", a position where a moderate amount of consistency allows bumps to most efficiently sort crews by speed. Analytically one would expect this to be true when the random element alone would induce bumps at about the same rate as would the non-random element, as is generally the case for systems in so-called "stochastic resonance." But that would take some effort to measure. It is however nonetheless encouraging to think that, in a very real sense, we might have luck on our side.