Qualitative modeling
Prior to World War I, climatology was dominated by qualitative approach and regional studies. Until the 1930s, almost all the weather and climate data were collected at the surface – there was little atmospheric data. Furthermore, scientists at that time did not really understand large-scale atmospheric movements. Edwards in this chapter goes into great detail to delineate the split between forecasters, theoretical meteorologists, and empiricists, to show the forces behind the development of the weather and climatology infrastructure:
“Scientific discussions of climate first appeared mainly in the context of natural history and geography. Descriptions of climate, topography, and the other physical features of regional environments accompanied narratives and catalogs of flora and fauna. Though climatic description sometimes included data analysis, more often it took the form of experience-based, qualitative narrative, perhaps with a few measurements thrown in for support” (2010:62-63).
There was a divergence of goals and interests among these groups of forecasters, theoretical meteorologists, and empiricists that shaped the way they interacted with each other. Early studies of the climate followed the natural history tradition, with its emphasis on classification, ordering, and holistic description. Moreover, in the 20th century, climatologists were not worried about global warming, but global cooling – they were trying to figure out how to avoid the next ice age, (discussed in Arrhenius 1896 publication section). In Edwards’ discussion of Callendar and CO2 : 1930s climatologists were skeptical about anthropogenic causes of global warming due to weaknesses in atmospheric data and understanding of atmospheric physical processes.
Friction
“In physical systems, friction means resistance. It occurs at the interfaces between objects or surfaces. It consumes kinetic energy and produces heat. Friction between moving parts consumes substantial amounts of the energy required to operate any mechanical device. Machines transform energy into work; friction reduces the amount of work they can do with a given input” (Edwards 2010:83-84).
For Edwards, friction is a metaphor that captures the difficulty in collecting, processing, distributing, and using the large datasets necessary for studying weather and climate change. On the one hand, there is computational friction: the difficulty in performing calculations, or as Edwards describes it: “the sociotechnical struggle with numbers that always precedes the reward” (2010:84). On the other hand, data friction is “the costs in time, energy, and attention required simply to collect, check, store, move, receive, and access data” (2010:84).
Friction can cause turbulence: social conflict and disagreement, in both the physical and human energy expended. People have to work to overcome friction. For example, prior to the widespread use of computers, people had to find analog ways to get around computational friction: “ways of substituting measurement for analysis or calculation by modeling, or simulating, the system in question. One analog technique involves building a physical scale model” (Edwards 2010:86). Another way was to use more people, which where the term computer came from – the people who did the math. These computers tended to be women, as seen in these films:
Earlier, we discussed the tension between scientific internationalism and scientific nationalism. While World War I interrupted the international flow of data, it also helped to spawn the creation of denser national observational networks. This was necessary for the professionalization of meteorology, which had been largely descriptive and not theoretical: “the primary forecasting techniques of frontal analysis were in fact cartographical rather than mathematical” (Edwards 2010:92). But there was a great need for the theoretical development of meteorology with the growth of air travel. This was a need that was expressed by leaders in both military and business interest groups, a foreshadowing of what Eisenhower would call the military industrial complex, and what Edwards refers to as ‘mutual orientation will call the scientific military industrial complex. The professionalization of meteorology and the needs of military and business sectors also increased the divergence between weather forecasters and climatologists, along with data friction:
“By the early twentieth century, data friction helped to create, and then to widen, a split between the data used by forecasters and those required by climatologists” (2010:97) “consequence of the split between forecasting and climatology was the rise of parallel overlapping networks for collecting and handling weather data” (2010:98).
One of the main points that Edwards is trying to make, however, is that all of this in the end is a social and cultural practices, as is also emphasized in the two movie trailers above:
“In the age of the World Wide Web, it is easy to forget that data are never an abstraction, never just “out there.” We speak of “collecting” data, as if they were apples or clams, but in fact we literally make data: marks on paper, microscopic pits on an optical disk, electrical charges in a silicon chip. With instrument design and automation, we put the production of data beyond subjective influences. But data remain a human creation, and they are always material; they always exist in a medium. Every interface between one data process and another-collecting, recording, transmitting, receiving, correcting, storing-has a cost in time, effort, and potential error: data friction” (2010:109).
Forecasting goes quantitative
Edwards writes:
“Computer models for weather forecasting rapidly came to require hemispheric data, and later global data. Acquiring these data with sufficient speed demanded automatic techniques for data input, quality control, interpolation, and “bogusing” of missing data points in sparsely covered regions. The computer itself helped solve these problems, but their full resolution required substantial changes to the global data network. Like all infrastructure projects, these changes involved not only scientific and technological innovation, but also institutional transformation” (Edwards 2010:111).
Institutional transformation was also made possible by the linking of commercial and national security interests. There were extensive civilian uses for military expenditures that otherwise could not be funded by business corporations. This is the ‘mutual orientation’ that Edwards describes, or what I refer to as the scientific military industrial complex: scientists and engineers oriented their military sponsors toward new techniques and technologies, while the agencies oriented their grantees toward military applications (Edwards 2010:112).
The development and climbing of the hierarchy of models did not lead instantaneously to better forecasts; in fact, initially the opposite was true. Simple barotropic models remained extremely popular with working forecasters well into the 1970s, long after baroclinic and primitive-equation model forecasts had become routinely available (Edwards 2010:132). While Edwards doesn’t add this as a type of friction, I would suggest that this can be understood as ‘cultural friction,’ in that new ideas and practices take time to be absorbed by specialists in a particular field; quantitative methodologies took time to be accepted by sports professionals.
As an added bonus, a little more on the Eniac women computers: