When Torah Bright flipped and spun her way to a gold medal in snowboarding, I happened to be waiting in the Qantas lounge at LAX. I attempted to lead a round of "Aussie Aussie Aussie, Oi Oi Oi," but since the lounge was mostly filled with my fellow Americans -- 90% of whom were watching Fox News -- the cheer died out rather quickly. At the point I'm starting on this post it's 5AM Sunday in Brisbane, and I've been up since 4AM St. Louis time... on Thursday. I'm not quite sure how long ago that was, what with the dateline eating most of my Friday, but when the crash comes (probably right in the midst of the business meeting that I'm here to attend) you'll be able to hear the snoring from wherever you happen to be.
In any case, this trip has me thinking about records. Events such as those won by the smiling Ms. Bright have enough subjective factors in their scoring that it's hard to evaluate the progress from year to year, but there are plenty of Olympic events that have scores more easily tracked over time.
Take this chart for example:
Over the last sixty years, Olympic winning times for the shortest distance in speed skating have gradually declined. That decline has not been perfectly steady. At several points there are winning times that were slower than they had been four ot even eight years previously. Anyone who has watched speed skating events knows that these times are affected by local conditions (is Vancouver "slow ice" or "fast ice"?) and by changes in rules and equipment. As with many Olympic sports, even scoring hasn't been perfectly steady over the period and more recent races use the total time for two 500m sprints rather than a single race -- I divided those values in half to produce somewhat comparable times. Despite those minor fluctuations, anyone looking at this chart would have to agree that speed skating times have been on a fairly steady downward trend, and if betting on future results you'd be well advised to assume the trend will continue.
Not all Olympic events follow such a consistent course. Let's swap seasons, genders, and equipment to look at another set of data.
For the first half of this chart, Women's shot put seems to be tracking along an equally nice progression with distances moving up each season. Then, in 1984, distances take a reverse and since then the distances thrown have barely passed those of the early 1970s. Shot put is not so affected by local conditions as many events, so what happened? 1982 happened. Specifically, it was in 1982 that the IOC developed a reliable test for increased testosterone levels and began a regular and expanding program of checking athletes for abnormal levels of anabolic steroids. From 1988, distances were as best flat as the IOC improved their ability to test for various substances. That dip in 2004 represents a Cuban athlete taking the gold after the original winner of the event was stripped of her medal. Anyone predicting the outcome of the next round of women's shot put wouldn't be wise to expect a record that passes those of the late 1970s.
No matter what you think of these Olympians, there's little doubt that training and dedication makes them exceptions to the abilities of "average" people. After all, throughout the last fifty years my time on the 500m speed skate hasn't changed one bit -- a value best represented by an '8' lying on its side -- and the thought of my trying to put a shot makes my toes curl back in anticipation of disaster. However, even if Olympic records represent the extremes of human achievement, there is information to be learned from looking at the extremes. In fact, there are times when looking at the extremes may be the best information you have.
Take for example this study of record high and low temperatures. Researchers found that while we are still seeing records of both sorts, the ratio between the two had shifted over time.
This study looked at 1,800 weather stations in the United States, but one of the nice things about it is that you can, with a few minutes labor, produce a similar chart for your area. We work hard to differentiate the local, temporary events of "weather" from the global, long-term "climate" and there's no doubt that an individual high or low temperature, set on some given day in the past, is a record of weather. But by looking at a number of these measures over a longer term, we may get a sense of where the climate is headed. The National Weather Service make available the min/max date on temperatures across the nation, so it's not difficult to find your area and determines how it matches up to the national results.
Naturally, any local information is going to be more variable than that at national levels. For example, here's a chart of the DC area.
Looking at these values you can see that in Washington there were as many record high days in the 1950s as there were in the 2000s, and anyone still humming "Summer in the City" is right in thinking that the 1970s produced some particularly toasty days. But what's missing from the 1990s and 2000s is the record lows. The decade of the 2000s produced an astounding 15x as many record highs as record lows, and the 1990s weren't far behind with a ratio of over 11. The number of record low temperatures set during "Snomageddon" was exactly none.
There's something else interesting in the values found for the DC area. The station there has been recording for 138 years. Given that there are 365 (and a quarter) days in the record book, any given year would be expected to have, on average, somewhere around 2.6 record highs and just as many record lows. The 60 year period from 1950 to 2010 would, all things being equal, carry 158 of the records on both sides of the warm/chilly line. That's not quite how it worked out. The most recent period actually carries about 30 more record highs than it should based on average -- not too astounding. What's amazing is that the whole period contains only 48 record lows, 90 short of what you might have predicted. The whole decade of the 2000s scored only 2 record lows -- less than 10% of what might be expected.
It's almost as if something has changed that might be affecting the results...