Chomsky, Prediction, and Polls

An interesting item from Dan Foster. Here.

2 Comments

  1. Dr. Strangelove says

    I’m in the middle of reading Nate Silver’s “The Signal and The Noise.” I think that Chomsky and Foster have Silver’s belief in the art of forecasting wrong. When Silver talks about the best qualities of Weather forecasting, or baseball prospect predictions, or poker players, or chess grandmasters it is the use of Bayesian models along with human expertise in the topic/a firm understanding of the dynamics of the system being modelled.

    Silver predicted Santorum’s victory in Iowa even though the polling data predicted otherwise because of particular aspects of the Iowa race that the model wasn’t accounting for. Nate Silver is constantly describing the limitations and failings of forecasts based solely on contextless data mining. Similar to how Jay Cost’s approach is described as holistic, a key component of good forecasting is constantly calibrating our theoretical understanding of a system with the success of it’s predictions.

    http://fivethirtyeight.blogs.nytimes.com/2012/01/03/why-id-bet-on-santorum-and-against-my-model/

    The attack on seems less about an actually disagreement about the methodology of forecasting but about disagreeing with who Nate Silver’s blog is forecasting to win the election.

    http://fivethirtyeight.blogs.nytimes.com/2012/11/06/nov-5-late-poll-gains-for-obama-leave-romney-with-longer-odds/#more-37295

    Posted November 6, 2012 at 3:22 am | Permalink
  2. Dr. Strangelove says

    http://www.fangraphs.com/blogs/index.php/nate-silver-and-imperfect-modeling/

    Posted November 7, 2012 at 4:03 pm | Permalink

Post a Comment

Your email is never shared. Required fields are marked *

*
*

You can add images to your comment by clicking here.