Atmospheric science may not seem like a particularly subversive job, but from an information science perspective, it involves continually dismantling the infrastructure that it requires to survive. At least that’s the way Paul Edwards, Associate Professor of Information at the University of Michigan described climatology, and one other sister science, in an interesting hour-long interview on the radio show, “Against the Grain” last week. (Full audio is also available for download.)
In the interview Edwards describes how the weather observing and forecasting infrastructure works (skip to about the 29 minute mark if that’s familiar), then notes that climatology is the art of undoing all that:
To know anything about the climate of the world as a whole we have to look back at all those old [weather] records. …But then you need to know about how reliable those are. [Climate scientists] unpack all those old records and study them, scrutinize them and find out how they were made and what might be wrong with them–how they compare with each other, how they need to be adjusted, and all kinds of other things–in order to try to get a more precise and definitive record of the history of weather since records have been kept. That’s what I call infrastructural inversion. They take the weather infrastructure and they flip it on its head. They look at its guts.
In his book, The Vast Machine: Computer, Models, Climate Data, and the Politics of Global Warming, Edwards points out that people don’t realize how much of this unpacking—and with it multiple layers of numerical modeling–is necessary to turn observations into usable, consistent data for analysis and (ultimately) numerical weather and climate predictions. The relationship between data and models is complicated:
In all data there are modeled aspects, and in all models, there are lots of data. Now that sounds like it might be something specific to [climate] science, but …in any study of anything large in scope, you’ll find the same thing.
In part because of this “complicated relationship” between observations and models, there’s a lot of misunderstanding about what scientists mean when they talk about “uncertainty” in technical terms rather than in the colloquial sense of “not knowing”. Says Edwards,
We will always keep on finding out more about how people learned about the weather in the past and will always find ways to fix it a little bit. It doesn’t mean [the climate record] will change randomly or in some really unexpected way. That’s very unlikely at this point. It means that it will bounce around within a range…and that range gets narrower and narrower. Our knowledge is getting better. It’s just that we’ll never fix on a single exact set of numbers that describes the history of weather.
Climatology is not alone in this perpetual unpacking of infrastructure. Economists seem like they know all about what’s going on today with their indexes, Gross Domestic Products, inflation rates, and money supply numbers. That’s like meteorology. But to put together an accurate history of the economy, they have to do a huge amount of modeling and historical research to piece together incongruous sources from different countries.
There is a thing called national income accounting that has been standardized by the United Nations. It wasn’t really applied very universally until after the Cold War….Just to find out the GDP of nations you have to compare apples and oranges and find out what the differences are.
And to go back as recently as the 1930s?
You would have to do the same things the climate scientists have to do…invert the infrastructure.