Roger Pielke Sr., in a post about the climate models’ difficulties in getting clouds right, poses this question:

How can international climate protocols, such as Kyoto, be established when the climate response to human climate forcings (and even natural forcings) is not adequately understood?

The short answer is that we make decisions all the time, large and small, in the face of uncertainty. We do it based on our understanding of the uncertainties, which we hope scientists have done a good job explaining to us. And we do it based on other values and considerations. The science can help us here, but it’s not the only input into the process.

For the slightly longer answer, I’ll borrow a bit from Sarewitz et al. (one of the “al.” being Roger Pielke Jr.) from their introduction to Prediction: Decision-Making and the Future of Nature:

The theoretical and technical difficulties of predicting complex natural systems are immense, and the magnitudes of the uncertainties associated with such predictions may be not only larege, but also themselves highly uncertain…. The idea that predictive science can simplify the decision-making process by creating a clearer picture of the future is deeply appealing in principle, but deeply problematic in practice.

The expectation that we need merely banish scientific uncertainty is an appealing one, but experience suggests it is wrong. In decision-making outside the scientific arena, we make decisions of vast import with far less of a clear picture of the expected outcomes that Roger seems to be demanding – going to war, for example, or implementing vast federal tax and spending policies. But even the picture of certainty Roger seems to want from science-driven decision-making is an illusion, as Daniel Sarewitz has persuasively argued:

(S)cientific uncertainty, which so often occupies a central place in environmental controversies, can be understood not as a lack of scientific understanding but as the lack of coherence among competing scientific understandings, amplified by the various political, cultural, and institutional contexts within which science is carried out.

In the climate case, Sarewitz’s argument is on point: writing two years ago, he noted the ongoing argument over what at the time looked like discrepancies between surface and satellite-derived temperature trends. Opponents of action to reduce greenhouse gases frequently cited the uncertainties caused by the discrepancies to support their position. Sarewitz presciently wrote: “resolving the discrepancy to everyone’s satisfaction would not really solve anything.”

Well, the discrepancy’s now been largely resolved, but, to no one’s surprise, the argument over greenhouse gas reductions has not been settled as a result. It obviously takes more than merely sorting out the science.

As Steve Rayner points out in an essay in the aforementioned volume, climate models are but one of many factors driving decision-making:

While scientists may reasonably strive for a break-through improvement in the quality of their long-term modeling of the climate system, the broader field of societal risk management or practical politics suggests that policy will likely be driven by negotiation among competing values rather than probabilistic forecasting.