I am currently attending the Southern Economic Association meetings in Fort Lauderdale, where the street facing the hotel was underwater during the most recent hurricane to pass through.
Anyway, I saw a talk today that took me back to when I first learned about chaos theory, actuallly in the early 1970s before the word "chaos" was used for it. I learned about it and the butterfly effect, aka sensitive dependence on initial conditions, while working on a combined model of global climate change and food production. It was called "irregular dynamics" back then, and the model showing it was climatologist Edward Lorenz, published in 1963. Blew my mind then. Anyway, it is widely accepted that the global climate system is chaotic, which is why one can only make weather forecasts for fairly short periods of time into the future, although one can forecast longer run average changes of averages such as average global temperature.
Anyway, I saw a talk by Emmanuele Masssetti of Georgia Tech that reminded me of all that, a talk thet explicitly drew on this chaotic effect. So he has been simulating future climate using different assumptions for the various climate models the UN has been using for its IPCC reports. What he found was that indeed the overall average temperature change projected did not vary as he varied initial conditions by small amounts. But what the projection for particular regions of the world did vary, indeed very much so as in the butterfly effect. So, for exmple, the Great Plains of the US would warm a lot under one simulation, but then actually cool for a simulation following a slightly changed initial conditions. This is atunning, but not really surprising given the underlying chaotic nature of th global climate system.
Another talk was a keynote by Richard Zeckhauser of Harvard, who was pushing for us to study geoengineering. He made a strong case for it.