Prophecy and Global Warming

Prophecy in the Modern Age: Economics and Global Warming

Nassim Taleb speaks of a kind of statistical modeling that cannot work. It is what he calls "the Fourth Quadrant." There are three others; they are arranged on a square grid. In the first and third quadrants, decisions are binary: yes or no. In the first and second quadrants, there is very little difference from the mean. That is to say, if you say 'yes' and you are wrong, or if you say 'no' and are wrong, the consequences are not that very different from if you were right.

So, in the third quadrant, there is a great difference between being right and being wrong: but you have a simple, binary decision. Yes, or no.

In the fourth quadrant, the decisions to be made are complex, and also the difference in being right or wrong is great. Mankind does not know how to build models that work in this place, Taleb says. There are two reasons why. The first is that we cannot imagine everything that might affect the model: if we are building a model of the economy, what if there are massive snows that year? How does the model compute the possibility, and account for the costs to production of having various roads closed? Did you even think to include that in your model at all?

The second is that we haven't lived long enough to have reasonable ideas about what probabilities are. If we say that something should happen 'once in a century,' how do we know? By looking at how often that thing has happened before, of course. But how many centuries have we had an industrial society? Less than two? Six, if you are very generous? We have nothing like the data set we'd need to build truly accurate models.

As a result, he says, when you are in the Fourth Quadrant you must stop making models. They don't work, and they can't work. We don't have adequate imaginations to work in every possibility; and even when we successfully imagine the possibilities, we don't have adequate experience to know how to compute the real odds of the thing occurring.

So, we must stop making models in this area: wherever there are complex decisions, and the range of possible outcomes is large.

Taleb has been writing about economics, but it works very well for climate change.

All those models of how the climate works? If Taleb is right, every single one of them is necessarily wrong. His theory, if correct, is a sufficient condition for discarding every single one of them.

People hate this. "All you're telling us is that our models are bad, but we still have to make decisions," they say. "Tell us how to make better models." But he can't do that; the thing to be learned is that no model can work at all here. Prophecy is not to be trusted.

At the least, we must say that we have absolutely no idea whether the models are correct -- nor even a capacity for guessing how likely it is that they are correct.

The whole environmentalist movement is based on Fourth Quadrant models. Rationality says that, unless someone can demonstrate that Taleb is wrong -- and I cannot see any way in which that could be demonstrated even in theory -- we must set every one of those models aside. They are necessarily unreliable, and cannot guide us.

No comments: