Wednesday, November 7, 2012

Bayes and Silver

Dear Readers,

This is a teaser and a plug for Nate Silver's book "The Signal and the Noise: Why So Many Predictions Fail -- but Some Don't." As her Dear and Faithful Readers will recall, Madame L wrote a preliminary review of the book last Sunday, lamenting that she hadn't been able to finish reading it and promising to provide a more complete review when she has read the book carefully.

Madame L was reading this morning about how the election results did and did not conform to the polls and punditry of the past months. Not surprisingly, the convergence of the predictions with the actual results depended on the prediction model the pundit used. Not surprisingly, because of his Bayesian model and careful use of facts, Nate Silver came close.

Madame L kept reading Nate Silver's fivethirtyeight blog at the New York Times website as the election went along, as did many others. Salon.com's Jacob Sugarman this morning provides a side-by-side comparison of Nate Silver’s predictions with the presidential election’s results as of last night:











Madame L is waiting for the book's price to drop into her affordable range, at which time she will snap it up and underline and scribble, turn page corners down, and go through all her other rituals for figuring out and understanding difficult material; and then share what she learns from it with You, Dear Readers.

Sincerely,

Madame L 

Post-Script: Just after posting this, Madame L read that sales of "The Signal and the Noise" shot  up by 800 percent on Amazon.com as election day drew into night, which means Madame L will soon be buying the book at the "best-selling" price discount at Amazon or her local bookstore.

Post-Post-Script: Madame L has also found an article "explaining" why Nate Silver isn't really "a towering electoral genius." Hmmm. Madame L thinks it's pretty easy to write something like that now and a lot harder to actually do what Nate Silver does. (And the author does write, "To be fair, the art of averaging isn't simple....") No duh. However, Madame L sees, upon actually reading the article, that the writer is not denigrating Mr. Silver's accomplishments but rather warning the newly converted to his cult that he's just a numbers manipulator, not some new messiah of the universe. And, Madame L sees, after looking up some other articles by that author, that he's mostly just doing his job of writing provoking and clickable (and therefore revenue-producing) articles about science.



No comments: