Nate Silver, New York Times writer and rising-star statistician, has written a book that clears the air in many sectors, not the least of which is climate science. The Signal and the Noise is not exclusively about global warming, but the one chapter devoted to the statistical challenges facing climate scientists does a great deal to clarify the facts of climate change.
"In the scientific argument over global warming, the truth seems to be mostly on one side: the greenhouse effect almost certainly exists and will be exacerbated by manmade CO2 emissions. This is very likely to make the planet warmer," writes Silver. "The impacts of this are uncertain, but are weighted toward unfavorable outcomes." Spoken like a true scientist, using the requisite probabilistic language in which scientific theory is cast. But even as the Bayes' theorem demands that scientists couch theory in conditional language, politics eschews uncertain language; the two worlds do not mix, Silver argues.
The problem is not, as Silver sees it, one of determining whether or not the greenhouse effect exists now and can be expected to elevate temperatures on earth. That much is known. The real problem lies in the statistical models that predict what happens next—and in our political system, which is making the situation worse.
Many climate predictions are inaccurate, and bad prediction models have been exploited by partisan politics. A bad prediction model becomes a weapon in partisan hands, harming the progress of climate science and misinforming the public. The outcome has been that more Americans now disregard global warming than did a few years ago. This is a setback, since future preventative actions may require public support.
Climate, Silver says, is a dynamic and extremely complex system, and such systems cannot be predicted with precision. In fact, Silver offers the tip to readers that whenever someone makes a confident, precise-sounding prediction about a dynamic system—such as exactly what the economy, the climate or California's earthquake faults will do next year—confidence is proof of inaccuracy. Predictions that include uncertainty are typically more reliable.
Although there is good modeling, Silver claims, for general temperature increases and sea-level rise, no models can accurately predict much more. Not a lot can be known for certain about changes that may come as increasing carbon levels exacerbate the greenhouse effect. But Silver advises science to remove itself from our dysfunctional political system, which is too polarized to seek honest outcomes. Only scientists working together without political interference, he says, can advance climate science by using the Bayesian process—thinking and modeling in a probabilistic fashion.