Sunday, October 21, 2012

Field Experiments as "UnNatural Experiments" the Case of Geo-Engineering

The NY Times reports that Russ George of California launched his own geoengineering field experiment.  His "treatment" was to dump 100 tons of iron dust in the Pacific waters off western Canada.  His goal was to regenerate plankton that would give salmon something to eat (and hence help them to prosper in our hotter future) and to sequester carbon dioxide.

"The iron spawned the growth of enormous amounts of plankton, which Mr. George, a former fisheries and forestry worker, said might allow the project to meet one of its goals: aiding the recovery of the local salmon fishery for the native Haida.
Plankton absorbs carbon dioxide, the predominant greenhouse gas, and settles deep in the ocean when it dies, sequestering carbon. The Haida had hoped that by burying carbon, they could also sell so-called carbon offset credits to companies and make money."
So this is the new free market environmentalism in the age of climate change.  This entrepreneur thought he could kill two birds with 1 stone as he generated more salmon and less global pollution.
His efforts have outraged the nerds and the NY Times.
I have more mixed feelings.  I know that we don't know much about geo-engineering.  In any case where learning needs to take place, we need to experiment.  In the 1990s, economists passively waited for "natural experiments" to take place such as a change in a law such or a volcano erupting.  In the 2000s, economists now actively run field experiments.  The climate scientists appear to be resisting the urge to run field experiments or are not in agreement about the protocols for how to have an orderly set of such experiments run.   
Academics such as UCLA's Ted Parsons and Columbia's Scott Barrett are writing about geo-engineering.  If we agree that "experimentation is good" and that we are doing "too little" geo-engineering experimentation then what do we do next? Do you trust the United Nations to figure out the optimal protocols?   When could individual efforts really backfire?  
Now, I'm not a scientist and  I can't judge whether Russ George's methodology will allow real scientists to use his evidence to test hypotheses.   
Consider the challenges that we face in our macro economy right now.  The fundamental problem that macroeconomists face is that we can't run enough controlled experiments to establish cause and effect.  The macro guys have very few data points to study.  There aren't enough post-War years and in a globalized world it is difficult to believe that your can study nation/years in isolation of one another.    How do we do science when we can't experiment?   One's "prior" world view tends to dominate discussions when little new data is arriving.  Are you a dogmatic Bayesian?