A worthwhile essay...

at The Scientist: Promises, Promises. "Ill-judged predictions and projections can be embarrassing at best and, at worst, damaging to the authority of science and science policy."

And know this: computer models are not facts.