A lot has been said about Usain Bolt – his brilliance, his stature and his music tastes. But the suggestion that he is the limit of human performance has been soundly rebuffed by sports scientists and statisticians alike. We did the same as many others and looked at previous performances to see if there was anything in the statistics that would allow us to make predictions.
A closer look at the numbers reveals some surprising effects. Figure 1 shows the progression in the 100 m sprint since the 1890s using world records, the Olympic Gold medal and the top 25 performances by individual athletes in each year (i.e. the best performance by the top 25 athletes). What does it reveal?
Average times (the red circles) have dropped from around 11 s in 1890 to just less than 10 s today and are clearly levelling out. There are a few periods in the 20th century when performance worsened appreciably, notably during the 1st, 2nd and Korean wars. There is also a sharp jump around 1975 when fully automated timing was introduced and the transition was made from hand-held stop watches to starting blocks with accurate electronics. How did technology cause performance to worsen by over 2 tenths of a second (around 2 m at the finish)?
The answer is reaction time; not of the athlete but of the time-keeper. It takes time to press a button once the gun is fired, which gives an apparently shorter time for the athlete. Fully automated timing removes this and times get longer and more accurate (the scatter post-1975 is smaller than that pre-1975).
So our first lesson is that we shouldn’t use data prior to 1975 unless we are really careful about the adjustment between hand and automated timing. Another lesson is that world record or Olympic medal data doesn’t tell the whole story and it’s dangerous to make big assumptions using only this.
The Usain Bolt effect
So what is Usain Bolt’s effect? The top 25 average performance in Fig. 2 shows a visible dip in 2008 and 2009 – but how much is just Bolt and how much is it his peers? (It’s important to point out that Usain Bolt, and indeed any runner is only included once in this statistic) Fig. 3 shows the top 25 performances since electronic timing was introduced in 1975. The whole plane of performances moves down over time indicating that all performances are getting better; the red spike in 2009 is Usain Bolt and shows how good he is compared to the other 850 top 25 performances over the last 35 years. It seems that the ability of Bolt, and others before him, produces such intense competition that his peers improve too.
Predicting the future
Anyone with a passing knowledge of the stock market will tell you that the past cannot be used to predict future performance. Any parameter can be modelled and we’ve even showed that global population can be used to predict performance in the 100 m sprint, the argument being that the larger the population we take our elite athletes from, the more likely that we’ll find someone outstanding (like Usain Bolt). This stops working when breeding fails to create offspring better than their parents (as in greyhounds or horse racing).
What the data definitely tells us is that improvements are still continuing and limits haven’t been reached; modelling tells us that the limit of average performance of the top 25 might be around 9.73 s (it is currently 9.95 s). As for world records, I won’t make predictions except to repeat that of an expert who knows that it will probably reach 9.4 s relatively soon. The expert? Usain Bolt.