Richard Dawkins has a wonderful column in today's Washington Post (here). I especially like his suggestion that the strength of a scientific theory can be measured as a ratio of the number of facts about the world it explains divided by the number (and, I would add, size and scope*) of assumptions required to explain those facts. "A theory that assumes most of what it is trying to explain is a bad theory." Notable examples of such "bad theories" are creationism and intelligent design.
By the way, Dawkins' definition of "bad theory" should also have application in the social sciences. For example, the "findings" of economic studies, especially those based on formal models, too often depend entirely on initial assumptions.
The entire column is both instructive and entertaining.
*Merely assessing the ratio of number of assumptions per number of facts explained would actually seem to favor the God hypothesis. After all, it requires only a single assumption: a self-motivating, omnipotent God. And it can explain (if only poorly) every fact in existence. Dawkins makes clear in his op-ed, however, that it is not just the number of assumptions that matters, but their size, scope and plausibility, given facts we already know.