Vacationing this week in the English Lake District, I am reminded that the easiest weather forecast is to assume that tomorrow will be just like today. In a statistical sense, that’s not a bad guess. In terms of being able to plan for when the rain will stop and I can go for a hike, it’s completely useless.

Science is based on such extrapolation. You try to generalise from what you see going on around you, and then you assume that this generalisation will work as you venture further and further from where you drew your inferences. You assume; but then, of course, the rest of doing science is to test those assumptions.