Scientific American presents Everyday Einstein by Quick & Dirty Tips. Scientific American and Quick & Dirty Tips are both Macmillan companies.
Last week, I went on a three-mile bike ride with my kids. We were on our way home when the skies started to look dark. We weren't worried however, because it was only 3pm and the weatherman said it wouldn't start raining until after 7pm and even then we could expect just some light showers. Unfortunately, he must have forgotten to mention this to the clouds because a slow drizzle started to fall as we rode. Soon the drizzle turned into a downpour, complete with hail and lightning. We eventually made it back to our warm dry house, but the soaked children voiced several complaints about the forecast and vowed never to trust the weatherman again. So just how do scientists predict something as complicated as the weather, and why are they so often wrong about it?
I’m a Model, You Know What I Mean
In order to predict things like the weather, climate change, or even election results, scientists use a tool called a mathematical model. A mathematical model is a set of equations that can predict an outcome based on a set of inputs. Let's look at an example to see just what that means.
Imagine that you have 5 kids and each morning they all want juice for breakfast. Everyday from Monday through Thursday, you get 3 requests for apple juice and 2 requests for orange juice. Knowing this, you could write a set of mathematical equations to tell you how much juice you need of each type on a given day. The equations might say:
Orange Juice = 2/5 x Number of Kids
and
Apple Juice = 3/5 x Number of Kids
So now imagine that Friday morning roles around and I'm making breakfast. Since I have my model in hand, I can take my inputs (the number of kids) and figure out how much of each juice I need. Since 2/5 x 5 equals 2, I pour 2 cups of orange juice. Likewise, since 3/5 x 5 equals 3, I pour 3 cups of apple juice. When my kids come in, I gesture towards the cups with a smug smile, confident that my model has predicted the correct outcome.
Then, the unthinkable happens. One of my kids looks at me and says "I don't want apple juice today. May I please have some orange juice?" In despair I pour another cup of orange juice, muttering about the fickleness of children and the failings of science. But is my model wrong? Should I scrap it and give up on this uncertain business of predicting childhood juice consumption?
