Erik Larson

Dec 14, 2009

Philosophy of GW

From a recent email I wrote:

I don’t think there’s a “Godelian” argument that the models are necessarily poor, but as a horse sense rule, when people tell me they can predict the future, I start wondering. That’s an old trick.

The difference between local weather systems— which are chaotic in the sense that they have a sensitive dependence on initial conditions — and global temp is that, in the former, we have causes, while in the latter, the causal mechanisms are still unclear.

I mean, in a chaotic system we all know that laws like Boyles etc. etc. apply, it’s just that the actual complexity of the system means that there isn’t any way to set up some differential equations and get the future state from some specification of the initial state. It’s hopeless, actually, as scientists like Lorenz showed as early as the 1960s. (This is why, when we get models of likely trajectories for hurricanes making landfall, they essentially draw out a set of paths that are so broad that Grandma could do the same with a crayon).

With mean global temp, we have to have a causal mechanism that we know reliably pushes the temperatures up, given the presence of the cause. This is the “sufficiency” condition that philosophers speak of: many factors may be necessary, but we need to know what subset of them are sufficient to make the entire temperature of the Earth rise. This is not a chaotic system, but more like the modeling of social science systems, or the economy: too hard to tell what subset gives us the sufficiency condition. So, when we see the mean global temp trend down between 1940 - 1960, even while massive amounts of C02 were getting released into the atmosphere, it’s puzzling. Fair minded people ask: if C02 is the sufficient condition, why then doesn’t it graph directly? In other historical periods, also, when C02 was much higher than now, why was temperature comparably low, and what have you? This is the sort of stuff that Lindzen at MIT asks, and so on.

For my part, the climatologists can work out the details of ice core records and so on, but I think it’s fairly obvious that there’s more going on than C02 (it could of course be a necessary condition, or it could be sufficient given background factors N, in which case we should be asking a lot more questions about N, like if N = {water vapor,…}, and so on.)

So this is kind of Popperian philosophy of science, where you have to figure out what you’re really claiming, and what falsifies it, etc.

Anyway, as I said, others can work out the details (assuming they’re honest about what they’re measuring), but I’d like to see the sufficiency case made more obvious, so that we can stop having the debate. (Put it another way, if the sufficiency case is that damn obvious, someone should let us in on it, instead of airing guilt-trip videos of children drowning .)