We usually think of description and prediction as two different things, but are they really so disconnected? Suppose I report that the temperature now is 65 degrees—that is description without prediction. But if I say that it is noon on a Spring day and the temperature is 65 degrees—that is still description, but haven't I really also made some predictions? Aren't I, in effect, predicting that during the next 12 hours the temperature will fall, but that three months from now the temperature will be higher?

When I predict that the temperature will fall during the next 12 hours it is not based on a trend, or a statistical analysis, it is because I understand that the sun causes the temperature, and that because the Earth revolves on its axis every day, 12 hours from now we will be in the shade and therefore cooler. Because I know what is happening, I know how the present is related to the past and I can predict the future. If you can't predict anything of the future it is because you don't understand the present.

Take another example: Suppose we have a group of four animals whose average weight is 85 pounds and average age is 22 years—is any prediction possible? Not much. But suppose that the description includes that the four animals are people, two children, age 9 and 11, and their mother and father, age 31 and 33.