Rating: 4 / 5
Nate Silver, for those (like me) who only pay attention to politics when the presidential election is rolling around, shot to fame when he correctly predicted the results of the latest U.S. presidential election. In The Signal and the Noise, Silver explains the world of probability, and why some predictions are spot on, while others are completely off base. Although Silver is probably best known in the political realm, in his book, he interviews people in a variety of fields to give the reader a broad understanding of forecasting, probability, and uncertainty. Silver champions predictions that are probabilistic (coming up with a range of possibilities and assigning a probability to each one), use both quantitative and qualitative information, and are revised when new information surfaces.
I thought Silver’s explanation of the housing bubble and the recession was spot-on, easy to understand, and fascinating. It was interesting to see how and why the vast majority of political pundits, economic analysts, policy makers, and even financial institutions got the recession completely wrong, in terms of forecasting the recession as well as its duration. The Signal and the Noise also touches on probability and prediction in baseball, poker, weather, epidemiology, chess, earthquakes, and terrorism. While some spots were a bit dry or seemed far-fetched (prediction terrorism), for the most part I found the topics interesting and engaging. I honestly didn’t think I’d be interested in reading about earthquakes, but after reading that section, I gained a better understanding about the complexities surrounding earthquake forecasting (and why weather forecasting has greatly improved over the past few decades, although some might consider that debatable).
Although I have taken a course or two that touched on statistics, I am in no way a statistician, but didn’t need to be to understand Silver’s writing. This book is about how better predictions can be made, and how the use (or in some cases, the overuse) of information (signals and noise), wrongly applied, can lead to bad forecasting. Humans have an innate tendency to try to see patterns where patterns do not actually exist, and while more information is generally thought of as better, it does not always lead to better predictions. This book should be widely read, especially by news/political junkies, since he really hammers home the concept of overfitting, causation, correlation, and they are misunderstood yet make good headlines.