Wednesday, November 21, 2012

Guys and Polls

What do Damon Runyon and David Brooks have in common ?

"nothing between human beings is more than three to one." Attributed to Damon Runyon by Robert Solow, American Economic Review, 1985

 If there is one thing we know, is that even experts with fancy computer models are terrible at predicting human behavior."  David Brooks

The problem, the reason that Runyon was joking and Brooks is a joke, are the the words "nothing" and "terrible."  It can safely be predicted that Damon Runyon will not correct David Brooks, first because he is dead and second because he has better things to do with this time.

Brooks's argument is an example of an fallacy which is more common than any other fallacy or any valid method of reasoning -- the false dichotomy.  He notes that there are some things which we can't predict and concludes that we can't predict anything. 

 I think there is a not quite so general errors made by those who weren't totally delusional,  but were determined to argue that Romney's chances were well over those calculated by Nate Silver. First the natural guess that if we can't make predictions about something which is relatively simple, then we can't make predictions about something more complicated.  So since we can't place meaningful probabilities on many statements about what a single person might do, we certainly can't predict what an electorate will do.   This seems to sound sensible to many people, but it is totally false.  We can't place meaningful probabilities what an electorate will do in the sense of what each and every elector will do (including sneeze).  We can place meaningful probabilities on how people ill vote on average.  Note that Brooks's argument is not specific to 2012.  He also claims it would have been nonsense on October 31 1972 to say Nixon's chances were definitely over 90%.  Brooks is rejecting thermodynamics as well as critiquing Silver.  The logic of his argument applies to molecules too.

To be honest, I think that there is something else Runyon and Brooks have in common.  Neither really believed what he wrote.  Brooks is sane and not stupid.  He can understand as well as I can that his conclusion is much too strong.  I am very confident that the problem is that he considers his job to be to make reasonable sounding arguments, in which he gives the naive the impression that he is just trying to understand a topic, and then conclude that what Republicans say is true.   

As noted above individual humans can be unpredictable.  The false dichotomy fallacy might tempt us to conclude that all individual humans are unpredictable.  However, David Brooks is predictable.

No comments: