Economics professor Daniel K.N. Johnson thinks that the world is much more predictable than you would expect. It’s no surprise that he can use economic models to predict unemployment rates. But one day, he decided to see if he could predict Olympic medal counts the same way.

He didn’t expect it to work, but it did. From the 2000 through 2008 Olympic games, he predicted total medal counts by country with 94% accuracy, and gold medal counts by country with 87% accuracy.

It’s amazing how little information he need to give the computer, and in fact, he doesn’t look at individual athletes at all. He only considers five factors: per capita income, population, climate, political structure, and home court advantage. His track record speaks for itself:

In Beijing, he predicted 103 medals for the U.S., with 33 gold. The actual count: 110 and 36. In Athens, he predicted 103 medals for the Americans, with 37 gold. The final results: 102 and 36.

He admits that China causes problems for his model, because there’s no one to compare them to. But other than that, the key to Olympic victory is clear: a successful country should be rich, big, and cold. It should also have a single-party government, and most importantly, it should host the games itself.

The model has held up remarkably well, and not even the anomaly of Michael Phelps threw it off much. But while it’s interesting that a handful of variables can tell you so much, it’s also a bit depressing that the world is so predictable. You could say they might as well just hand out the medals and skip the formality of actually competing.

When I heard about Daniel K.N. Johnson a few weeks ago, I made a note to see how well he predicted the 2010 Winter Games in Vancouver. The results are in, and surprisingly, he did a really awful job:

Not sure what happened here, but it looks like the world isn’t so predictable after all. I guess computers don’t believe in miracles.

Photo by Duncan Rawlinson