Three years ago, I wrote Ranking University - abolute bloody nonsense: exposing internal inconsistencies and flabby reasoning in the Sunday Times List of Universities. I should have been less critical because The Institute was deemed to be Top and the administration has been crowing about it ever since. But nonsense is still nonsense especially if it agrees with your prejudice. Three decades ago, a fellow graduate student at BU asked rhetorically "When is athletics going to embrace the idea of statistical significance?" As an evolutionary biologist, he couldn't see how, if two runners are a hands-breadth apart after running 200m, one was deemed to be faster than the other and get a gold medal. Wouldn't it be more sensible, and sufficiently accurate, to say that both were extremely fast . . . way faster, for example, than my sofa at terminal velocity. Why, for example, do we arbitrarily take the first bit of anatomy over the line to define winning? Why not the last toe that might favour shorter people. The IBU specifies the first part of the first foot across the line as the defining criterion.
Here's an interesting piece about accuracy in sports, coming via kottke. Tim Burke says This Is Why There Are So Many Ties In Swimming . . . it's because of engineering tolerance in the design of swimming pools. We've just built a shed using telegraph poles as the main vertical members. These beasts are all different diameters, obviously tapered and some are slightly bowed. They are buried in a hole 90cm deep back-filled with 804 gravel aka 2-down. We wanted the 7.5m x 2.5m structure to have parallel sides and be square, but when it came to the, definitely square, corrugated iron roof sheets, we were about 5cm = 2% out of true. Bummer! but not shameful and structurally sound and sufficient. We had to trim the top supporting timbers on a scow, so that they looked correct wrt to the corrugated roof. We'd have lost our jobs and killed people if we'd allowed 2% error designing a skyscraper. The campanile at Pisa is 7% off.
It turns out the specs for Olympic swimming pools require the lanes to be 50m +/- 3cm long. Swimmers correctly reckon that it is therefore silly to measure speeds to thousandths of a second because at champion swimming speed 0.001s = 2.4mm: a distance far smaller than the potential difference in lane lengths. So Olympic swimmers are deemed to have tied if their clocked speed is equal to the nearest 1/100th of a second. Quite right too! Not so for above-ground events which are measured in 1/1000ths. Tim[e] Burke links to another site which investigates the relevance of the speed of sound: an athlete nearer the starting pistol will be off 1/100ths of a second earlier than his more distant rival.
Here's an interesting little essay from fivethirtyeight about the chaps who measure course for the more nebulous events: 50 km 'walk'; or the marathon. Did you know that these guys add an arbitrary 0.1% to their measured shortest path, to ensure that no athlete in a particular event in, say, Beijing or Tokyo is able to win because they haven't gone the full distance? It strikes me as obsessively obsessive to obsess about the variables that you can measure /control and treat the unmeasurable as invisible. The XIX Olympiad in 1968 was hosted by Mexico City which is well over 2,000m above sea-level, the air density seems to have favored sprinters and the long jump but the oxygen deficit made long-distance racers slower . . . unless they came from Ethiopia or Kenya and were used to running Up There.
You can say that it is a level playing field because the rules are written down and clearly agreed by all the competitors. But it might be more sporting if Gold invited Silver to share the top tier of the podium when they were separated by a shirts-thickness at the line.