
The whole global warming schtick is based on ad hoc climate models. The problem with all computer modeling is that you get out what you put in. I used to call it Garbage In Gospel Out. It's fascinating how people will salute when you say well the XYZ Model says ... whatever. If they know that the model says just what the programmers made it say, they might seek a little firmer ground for believing it. Generating the grounds for believing a model is called model validation. Backward validation shows that the model accurately predicts what you have seen before. If a model fails here it is really really bad because you know all this data. Forward validation is does it predict the future. This is tougher because you have to either wait of the future or hold back some information to use for forward validation from an earlier data set. Climate models don't validate very well.
No comments:
Post a Comment