For two years in a row the baseball World Series was won in a clean sweep - four games to none. "What are the odds of that?" I wondered. So I tried to work it out and got a certain answer. Then I tried another method and came up with a different number. And then a third answer.  

Finally, I wrote a small computer program to simulate the contest, and with the computer results in hand I eventually figured out which of my analyses was correct and where the others went wrong. It was a cautionary and humbling experience.

That computer simulation could be considered a bit of experimental mathematics, although the term covers more territory than that, including methods that long predate the arrival of the computer. Gauss was a master of the art; some of his deep insights arose from just playing with numbers. 

Personally, I find experimental approaches suit my habits of thought and help cover up some of my deficiencies. But no one in mathematics believes they will ever push proof off the pedestal.

Brian Hayes
Former editor for American Scientist