Reply #3: Faith over reason [View All]

[zipplewrath]

It is a difficult job, as someone who makes analytical models, to accept that a model is "correct" when it is contrary to your "belief" or preconcieved notion of what is "right". The problem with the right wingers is that their faith is strong, even in the face of information to the contrary. Their usual response to such information is that the information itself is wrong, never their strongly held convictions. What Krugman is describing is some model makers how are unhappy with their own models, not because of known flaws in them, but merely because they are predicting things that are contrary to their desires.

What Krugman is pointing out is that a common model verification methodology is being ignored. If you have a model, present it with the inputs of past events, where you know what the outcomes were. If they acurately predict the known outcomes, that indicates the models are probably working correctly. If they can accurately reproduce the past, they have little hope of predicting the future.