What do baseball, weather, politics, hurricanes, and poker all have in common? If you predicted the answer was the need for predictions you would be right. In every field, we are constantly trying to predict what will happen next. These predictions can have major consequences, as they determine what we will do now. In baseball, this means millions of dollars, in hurricanes this means hundreds of lives. Nate Silver proves worthy to speak about prediction as he most recently predicted all 50 states in the national election. The funny thing is, he will tell you that part of what makes him good at predictions is the fact that he knows that was luck. The need to think probabilistically is a major theme in the book. We should always understand that we can only know what will happen to a certain extent. Even when we are sure, this may mean 90 or 95 percent, but not all the time. I have always seen the opposite type of thinking in sports. Analysts are trying to predict games at a time or even predict the champion before the year begins. This is foolish in so many ways, but perhaps most importantly it lacks the understanding of uncertainty. Another major issue with predictions given in the media is that the people that are good for TV are hedgehogs, people with one vision and fit everything into it, instead of being foxes, who take everything into account. This leaves out valuable information, so its no wonder that political pundits are more often than not so very off the mark. Silver advocates the use of Bayesian statistics. This essentially means that as we gain more information, our predictions should constantly by moving. This enables us to consistently be making better predictions and getting closer to the real occurrences. Doing this will also help us get to see the signal from the noise, or find the causation in the correlation. When our predictions fail, our probabilities need to change, and as the foxes we should not be stubborn with them. It is these problems and many others that he outlines to be the problems with prediction today. In his many years as a prognosticator, Silver analyzes a diverse array of topics, some of which are in the first sentence of this post. At times I found a lot of the examples to be unnecessary, as he had made his point long ago, but to those interested in the specific topics they may have been enlightening.
For the past six months, I have been working for an insurance firm doing business modeling which involves making predictions. Also, graduating with a math degree and taking the first actuary exam gives me a good background in this specific field. While reading this book, I was happy to say that I was already using many of the theories in the book. The difference is that I had no language to use for these methods. Silver gave me that language, which concretized the ideas for me. Anyone not in this type of field would probably be able to get even more out of this fantastic book. It's somewhat long, but for those worried about it being too mathy, it isn't.
