Tuesday, December 8, 2009

More Global Warming

As I mentioned, global warming is heating up in the news (hah!)

Now Science Daily is adding fuel to the fire (hehe! oh, it's too easy!)
"the Earth's temperature may be 30-50% more sensitive to atmospheric carbon dioxide than has previously been estimated"
This sort of thing would be an embarrassment in microarchitecture models (oops, yea our old performance model was off 30-50%, but the new one is good, trust us!). Somehow, in climate models, being less hugely off is a big improvement.
"We found that, given the concentrations of carbon dioxide prevailing three million years ago, the model originally predicted a significantly smaller temperature increase than that indicated by the reconstructions"
Now we come to one of the big points that causes my skepto-meter to spike. If you are correlating data in your model against events from millions of years ago, you are feeding data that is almost certainly wrong into it (since the earth is, very likely, only thousands of years old).

Bringing bad assumptions (which makes bad models), and feeding in bad data is going to drive you into circles of bad decisions.

Making wild predictions ("the world as we know it will end unless we shut down our economies!") doesn't help.

No comments: