Sunday, March 22, 2015

Book Review: The Signal and the Noise

The Signal and the Noise: Why So Many Predictions Fail--But Some Don't
Nate Silver
Statistics, cognitive psychology

The Signal and the Noise is a user's manual about how to make better decisions with imperfect information. As is often the case, the people who would benefit most from the book are the ones who are least likely to read it, or to believe it if they do.

It turns out that foxes make better predictions than hedgehogs, that thinking in terms of probability is really important, that people who refuse to apply mathematics often talk hooey, that self-proclaimed "experts" often aren't, and that it pays never to be too certain of your initial assumptions. This book, in fact, confirmed all of my prejudices.

Therefore, it's obviously wonderful.

Put that way, it sounds stupid. And yet this is one of Nate Silver's (and others') key findings: we all think that way. The difference is that truly rational people know that they're not rational--and factor that into their thinking. Put another way, one of the most important sources of human error is to look at the data with an eye towards confirming your own biases: in real-world data, there's usually enough noise clouding the signal that a determined eye can pick out a false pattern from it.

The book doesn't put great mathematical demands on the reader, although in fact basic probability and statistics don't require much beyond high-school algebra. There are a good many charts and graphs, which are clear and easily understandable. The book is written in straightforward, entertaining, clear, and fluent prose. It perhaps loses a little steam towards the end, where Silver chooses to multiply examples rather than honing his conclusions. Aside from that nitpick, I'd recommend The Signal and the Noise to anyone who wants to be a better thinker, which I assume includes anyone who's read this far.

Pairs well with Daniel Kahneman's Thinking Fast and Slow or Dubner and Leavitt's Freakonomics.


  1. This is in my list to read, as is Freakonomics. I assume you listen to the podcast?

    Thinking Fast and Slow was great. I really enjoyed it. Makes you wonder how the human mind works. When should you use the frontal lobes, and when should you let your mind get out of the way. I forget where I heard it, but I did hear that jazz musicians have almost no frontal lobe activity when improvising. They bypass the rational thought part of the brain altogether.

  2. I'm not generally podcast-oriented. Unless I'm driving, the audio-verbal channel is too damn slow. That's one reason I skipped so many lectures in college (though not the only one, as you may well suspect).

    I'm glad you liked Thinking, Fast and Slow. It seems to me that one of the best things anyone could do, cognitively, is to develop a good "system 1" (fast) sense for when it's a good idea to invoke "system 2" (slow). I work consciously at this.

    What you say about jazz musicians doesn't surprise me at all. The "system 1" brain, it seems to me, is very heavily about patterns. (We're so good at recognizing patterns that we see them even when they're not there.) And I can think of nothing that would engage the pattern-matching brain more than jazz improv.