Flag Content?

A question for Robert: Why are computer models generally poor at prediction? With the exception of weather models that have become pretty good at 7-10 day prediction (but not further into the future than that), computer models are generally no better than historical extrapolations when it comes to predicting the future. It is hard to believe that economists and investors have not been more successful at modeling - heck, maybe they have, and they are not telling... Highly regarded models present contradictory evidence for global warming, and if you believe the pessimistic models, there is poor agreement from them on what steps are likely to be the most impacting to reverse the warming trends they predict. So, what gives with our inability to predict the future with a computer? What have we not modeled well? Or, what do we completely misunderstand about modeling as a process or methodology? It must be a matter of the software, because CPU cycles are "grandly" abundant - even on my desktop. It is time for someone to express a corollary to Heisenberg's uncertainty principle regarding prediction. Or is this more a matter for Schrödinger's cat?