approach that climate science must take advantage of, it also comes with its own set of novel pitfalls, which must be carefully marked if they are to be avoided. More specifically, I argue that careful attention to the nature of chaos should force us to attend to the limitations of science by simulation, even in ideal conditions. It is worth emphasizing that these limitations are just that, though: limitations, and not absolute barriers. Popular dissatisfaction with the role that computational models play in climate sciences is largely a result of conflating these two notions, and even some people who ought to know better sometimes confuse the existence of chaos with the impossibility of any significant forecasting. We’ll think about the nature of the limitations imposed by chaos (especially in light of the method of computational model building), and see how those general limitations apply to climate science. Finally, I’ll argue that even with these limitations taken into account, the legitimate predictions made by climate science have serious implications for life on Earth.
5.1 The Challenges of Modeling Complexity
Individual special sciences have been increasingly adopting the concepts and methods of complexity theory, but this adoption has been a piecemeal response to the failures of the decompositionalist method in individual domains. So far, there exists little in the way of an integrative understanding of the methods, problems, or even central concepts underlying the individual approaches. Given the highly practical nature of science, this should not be terribly surprising: science does the best with the tools it has, and creates new tools only in response to new problems. The business of science is to figure out patterns in how the world changes over time, and this business requires a degree of specialized knowledge that makes it natural to focus on the trees rather than the forest (unless you happen to be working in forestry science). As a
149
