Thursday, June 11, 2015

Review:: The Signal and the Noise, by Nate Silver

Title: The Signal and the Noise: Why So Many Predictions Fail - But Some Don't
Author: Nate Silver
Format: Hardcover
Rating: ★★★★☆

I'd been meaning to read this book for a while. Nate Silver talking about predictions, big data, and our limitations sounded so up my alley I couldn't even. But then I forgot and read other things. And can I say how glad I am that I picked it up? The mathematics in the book was pretty easy to grok and the ideas laid out are definitely worth perusing. It's also a pretty simple read, for not feeling entirely popsci. (Although, let's be honest: it's popsci.)

Nate Silver, who came to fame mostly for predicting the 2012 elections much better than most other people, covers a variety of topics where we've been working at making better predictions as our technology improves. Some of the highlights include weather, earthquakes, politics, economics, climate change, and terror attacks. There's also a good deal about sports and gambling, which makes sense, considering Nate Silver got a lot of his start doing baseball predictions.

The book is basically a big proponent of Bayesian thinking. And while sometimes Bayesian logic makes my head hurt, I'm not about to say that it's not a powerful tool. And I definitely think more people should be aware of it. Nate Silver also looks heavily into how our biases can impact what signal we find (true or false) and how we share our prediction with others (frex, do we explain our level of uncertainty?).

I found the section on weather and earthquakes incredibly fascinating, which probably isn't surprising, considering I live in the San Francisco Bay Area. Weather is one of the few areas where more computing power and data has actually helped us create more accurate forecasts. And they're highly accurate, even if the commercial forecasts so many of us see aren't. The section on earthquakes was pretty depressing. (We're not likely to be able to accurately predict specific earthquakes ever.)

The sections on politics and economics, however, were kind of terrifying. Basically, no one, not even political scientists are any good at political prediction. And economics is almost worse. Most economists can't predict a recession, even when we're in one. But instead of just pointing out those chilling facts, Nate Silver provides good reasons why this is the case, arming the reader with better ways to view those kinds of predictions.

Perhaps my favorite thing about the book was the emphasis on how changing your mind is not weakness. It's the right thing to do when faced with evidence that contradicts your model of the world. Just sticking to your guns and going down in a blaze of glory doesn't help you... or anyone else.

Another thing I appreciated was the sheer number of footnotes. Pretty much everything was cited.[citation needed]

Oh, and the chess parts were awesome.

No comments:

Post a Comment