A Vast Machine: Computer Models, Climate Data, and the Politics of Global Warming (review)
Reviewed by: A Vast Machine: Computer Models, Climate Data, and the Politics of Global Warming Fernando Elichirigoity (bio) A Vast Machine: Computer Models, Climate Data, and the Politics of Global Warming. By Paul N. Edwards. Cambridge, Mass.: MIT Press, 2010. Pp. xxviii+518. $32.95. This history of climate science modeling is an important book. Its narrative covers roughly 180 years, chronicling the emergence of a planetary monitoring infrastructure as it developed from the increasingly intertwined trajectories of meteorology, weather forecasting, and climatology. Meticulously researched and lucidly written, it succeeds in giving the reader a nuanced and rich historical understanding of the relation between climate science and computer modeling as well as of the complex data problems involved in that relation. There are fifteen chapters (more or less chronological) plus an introduction [End Page 243] and a conclusion. Chapters 2 through 5 look at the history of weather research and climatology prior to 1954. The last three chapters provide a fascinating discussion of climate science, knowledge production, and international politics. Chapters 6 through 12, the heart of the book, discuss the longue durée of data collection about Earth’s climate, a project that over time became international and ultimately global. These data, in turn, are analyzed, remade, and also often produced in a mutually constitutive manner by computer models. Edwards discusses in great detail the real problems of collecting data needed to build the long, continuous datasets essential to creating realistic and useful computer models of Earth’s climate over time. For example, meteorological data collected in the Alps in the nineteenth century were marred by changes in the placement of precipitation gauges and thermometers at different times and in different places, thus thwarting the identification of long-term trends. Meteorological data have been collected largely within national frames of reference, with countries evolving distinct methodologies and idiosyncratic practices related to the placement and calibration of instruments and formats of the data collected. Thus national datasets often were incompatible with one another, a complication amplified by the international politics of data sharing, as, for example, during the cold war. International frameworks eventually were developed to deal with these problems, such as those within the World Meteorological Organization, which is discussed in chapter 8. In the end, a fundamental aspect of data collection is that “we are stuck with whatever data we have already collected, in whatever form, from whatever sources with whatever limitations they might have” (p. 15). Thus climate models will never provide the definitive answer about past or future climate states. They can only do so within a range of probabilities that scientists strive to reduce, in part, with better algorithms and modeling. Even those solutions are imperfect, for additional problems arise due to changes in the algorithms used to process satellite data. The development of climate modeling started with the numerical weather models of the 1950s—models that eventually evolved into simulation models that allowed scientists to create “what if” scenarios covering long periods of time. Reanalysis models, which emerged in the early 1990s, are a mix of simulations plus real-world weather datasets. Reanalysis modeling is perhaps the most controversial form of modeling, as it creates new data out of the mix of collected data and simulated data, which, in turn, is used to further refine the models. Edwards also discusses what he calls “data analysis models,” an umbrella term referring to “mathematical techniques, algorithms, and empirically derived adjustments to instrument readings” (p. xv). While Edwards is careful to state that he is not attempting to cover [End Page 244] every stream of science and technology that has gone into the emergence of a global knowledge infrastructure, he comes close to doing so in practice. As a result, his too-brief discussions of certain important programs feel like gaps: for example, NASA’s Earth Observatory System and its attendant satellite infrastructure and data should be discussed at greater length. Similarly, the “Limits to Growth” Project, the first computer model attempting to model anthropogenic impact on Earth, is subsumed within the larger narrative, thus obscuring its importance in developing a global perspective focused not on the planetary climate per se, but on the potential exhaustion of...
Read more
2