program. I build it to include key variables, like type of of food eaten, number of calories consumed, duration of exercise, type of exercise, and age. I then use the model to model several scenarios — one where someone eats well and exercises a lot, another where they eat poorly but exercise a lot, another where they never exercise, and so on. The results are dependent on the input variables, as they are with all models.

I then go to my friend who eats like a pig and never exercises, and I warn him that my model predicts that he will weigh 300 lbs within a year if he keeps up his current lifestyle.

A year later I see my friend again, and he is a svelte 190 lbs. Turns out my friend started dieting and took up long-distance running. Ha! he says, your model was wrong! You said I would weigh 300 lbs, but I weigh less than 200lbs and just ran a sub-4 hour marathon. Your model sucks!

Is my friend right? Was my model wrong?