We had a good laugh this past weekend while briefly celebrating the eleven year anniversary of one of the most infamous plays in baseball history. Of course, I’m referring to the events of March 24, 2001, and more specifically, the seventh inning of a Spring Training game between the Arizona Diamondbacks and San Francisco Giants, when Randy Johnson threw a fastball that struck and killed a dove.
According to witnesses and anyone who has taken the time to view the YouTube clip, a bird swooped between the pitcher’s mound and home plate just as Johnson, one of the hardest throwing pitchers to ever play the game, was releasing the ball. His fastball hit the bird at such a velocity as to make it appear as though it had vapourized into thin air, leaving only feathers as evidence of its previous existence.
For the record, the official call from the home plate umpire was “no pitch.”
While most of the humour that’s derived from this event is found in the incredible odds against it ever happening, there’s also an element of importance to it that’s overlooked.
Among baseball fans who tend to rely on the more advanced metrics, you’re likely to hear the phrase “small sample” uttered as an effective two word argument against relying on too few examples to come to a conclusion. For example: The Dustin McGowan contract extension is a good one because he struck out 11 batters in the last nine innings he pitched in 2011.
Randy Johnson hitting a bird with a fastball is such a wonderful example of “anything can happen” because it acts as a visual exhibition of why we can’t rely on small samples. Baseball is full of random outcomes that aren’t always as dramatic as the “fowl ball” that Johnson threw, but are nonetheless examples of the most likely result not actually occurring.
Somehow, things happening beyond a player’s control or other unanticipated outcomes on a baseball diamond have garnered a reputation for drawing the ire of statistics loving baseball fans. Nothing could be farther from the truth. Not only are “short hops” enjoyable from an anticipatory stand point, they’re also accounted for by the more advanced metrics.
When good baseball writers criticize managers or front offices for poor decision making, they will most likely bring up numbers to support their point of view. By doing so, they’re not suggesting that their proposed use of the bullpen or that the contract that they would have offered is certain to work out, but rather that, based on the information that they have, one outcome is more likely than another. It’s all about likelihood.
And when something happens that even scientists have a different time recreating, it’s not disappointing. It’s an aspect of baseball that makes it appealing. That is, until someone starts using one instance out of 18 billion as evidence to support the likelihood of something happening again.