Monday, November 1, 2010

Is stupidity Dangerous? - Part II

Here is an example of a horrible argument made by someone who I think wants to be taken seriously.

Global warming is going to make the earth warmer. Let's take that as a fact. The result is that there will be less arable cropland in 100 years. This will increase the probability that poor countries reliant on farming fall into chaos. The U.S. will be less secure because these failed states harbor terrorists and commit genocide and other actions that draw U.S. involvement. 
So the U.S. needs to do something about global warming.

There is something that jumps off the page about that argument that is a real sin. But I don't think enough people are trained to see what it is. Thankfully, I've beaten it like a dead horse on this blog so I'm not even going to state it.

But beyond the main problem, the argument also fails to correctly account for benefits.

If your argument about policy is coherent you should be able to turn it into a list of costs and benefits and assign each benefit a distribution of value and a distribution of probability that seems reasonable. That is your argument and you are just using words to dress it up to make it readable. But the two forms should be equivalent and you should always be able to turn one into the other.

Here I don't think the person who made that argument really thought much about the values of the probabilities of the benefits they are talking about (i.e. the benefits from averting a terrorist attack). Why do I think that? They imply that agricultural-based economies will be common in 100 years. But no one thinks that considering the rate of global economic growth.  They never mention their assumptions about the discount rate, which is critical over a 100 year horizon. And they suggest that P(supports terrorists| failed state) is non-trivial, which I don't think is reasonable.

No comments:

Post a Comment