Replies

But Pielke's skepticism surfaces when the conversation turns to the computerized climate models used to forecast future warming. The models are incomplete and unreliable, he says, especially when used to predict climate change at the regional level.

"My feeling is that the climate system is so complex that we can't predict, with skill, what will happen in the future," as levels of carbon dioxide and other heat-trapping gases continue to rise, Pielke told the Rocky Mountain News earlier this year.

"I think we should probably control CO2 (carbon dioxide)," he said. "But to try to base it on these models is not solid. It's not good science."

I have said this, and said this, and said this.

Computer modeling is hard. When you get everything right, by developing a model and testing it iteratively against reality, then you have something really valuable. But that's hard, even for simple systems. I have a fair amount of experience in this.

For something as complicated as the Earth's climate over decades, it would be a Herculean task to make such a model, but it would be doable if you had enough decades-long stretches of data to check against. But we don't, and we won't.

At best, the models tell us what could potentially happen, and that's valuable. But their predictions must ultimately be treated as guesses, and not as prophecies. That makes a big difference in public policy decisions.

What bothers me is that computer models are almost automatically regarded with derision by climate-change skeptics, even though they are just extensions of non-computer theoretical models the same way computers themselves are extensions of our own brains. If anything, they have the potential to help us to understand more precisely, and provide a greater integration of contingent factors to climate change or any other long-term phenomena which can be predicted scientifically. Just as radio telescopes and electron microscopes allow us to respectively see in a wider spectrum and in finer detail than their optical counterparts, computer models have the potential to both broaden the scope and reduce uncertainty (within limits, of course) in our predictions. I hope you do not feel insulted by may saying this, Phys, or if my comments have come off as ignorant or unrealistically optimistic to you. After all you've spent most of your career doing computer modeling in a wide variety of experimental and industrial applications, and are more qualified than anyone else here to comment on the issue. But I still feel that computer models are important and even crucial in analyzing and evaluating global climate change. What they can't tell us is what are the necessary economic policies we must take to prevent its adverse effects.