# More on Nate Silver

I originally posted this on Facebook. But why not here as well?

**tl;dr:** I complain about the use of 'Bayesian' in popular culture, and in the process rant (even more) about Nate Silver.

More people (a psychology professor and a computer science professors at NYU, none the less!) get the definition of 'Bayesian' wrong, this time in the New Yorker.

Marcus and Davis start off strong^{1}. And I understand that they're writing for New Yorker readers (presumably more mathematically sophisticated than your average American, but not amazingly so). But 'Bayesian' != 'Someone who uses Bayes's theorem'. Bayes's Theorem isn't even a theorem! It's a corollary of basic results about conditional probabilities.

A Bayesian is, in the loosest sense, someone who treats the parameters of a statistical model as if they are themselves random variables. A frequentist treats them as fixed (or, if you'd like, a trivial random variable where all of the probability mass lies on a single value). In practice, that's it.

Larry Wasserman (a card-carrying statistician, unlike Silver, Marcus, Davis, or, well, myself) explains.

*The Signal and the Noise* is a fantastic book. If anything, it shows how far you can get using statistics without a firm understanding of its foundations (or lack there of).