Wednesday, November 27, 2013
the perils of false pattern recognition bias
I like to think that I understand Nate Silver, but my tiny primate brain still wants to believe certain things that are probably not true. For example, the graph above proves that it is impossible to claim that there is any sort of solid pattern to homebuilding in the U.S. In general, housing starts match population growth, but the complex interaction of interest rates, inflation, politics, and demographics create a natural volatility. No single factor can explain what can or should happen over the next decade. I'm tempted to say that there was less volatility prior to 1970, but if we were to extend this graph back to the 1930's we would see some real wildness.
Could there be a way to smooth this out? Could there be a way to make things more stable? Is that the dark cry of the Marxist? More on this later, maybe.