Your faithful correspondent recently questioned the "science" of computer modeling as it relates to climate change. Arnold Kling weighs in on modeling as it relates to economics. This one is worth quoting directly:
"A model is not all-or-nothing, right-or-wrong. A model is like a pair of binoculars. It helps you see some things more clearly, at the expense of not seeing other things at all. If you do not have that understanding of the modeling process, then you are missing what I see as a fundamental methodological truth."