To Hume it may Concern

One can only learn from experience when there is something to learn. As a general method, empiricism is simply a recipe for superstition. Indeed, the history of religion is in large part the history of an interminable research project, an attempt to figure out the wishes of the gods. The ancients were quite methodical about it: the Mesopotamians carefully correlated the configurations of the stars and the shapes of the livers of sacrificed animals with what happened later, and the Roman priesthoods and the Senate itself noted the political results of prayers and supplications to the various gods and tried to learn from experience as best they could. One thinks of theology as a largely deductive operation, but a great deal of scholastic logic chopping is devoted to explaining away the apparent failures of an underlying inductive methodology.

You often hear that science is made possible by the faith that the universe operates according to regular and comprehensible laws, but that thesis can only be valid to the extent that a certain amount of hopefulness is indeed a psychological precondition for persistent inquiry. Nevertheless, the fact that the fisherman who goes on fishing is the one who may actually catch something doesn’t mean that the optimism that motivated his patience is really warranted. After all, as we all know, lots of the time it isn’t, just as for the most part the things in the universe don’t make any damned sense at all. Induction works, when it works, which isn’t often, not because of some theological or metaphysical principle but because detectable regularities do govern a tiny proportion of possible cases. It looks like it works in general only because in general we focus on the exceptions, the relationship between the temperature and volume of a gas, for example, instead of the relationship between a person’s temperament and the position of the planets at her birth or the likelihood of my coming down with a cold and the color of my shirt last Thursday. The much-mooted problem of induction, like the unreasonable utility of mathematics, is a chimera, an accident of sampling. I thought you should know.