I had been wanting to read this book since it was released, and I finally had the chance. I wasn't disappointed with it. In this book, the author takes a non-mathematical approach to understanding how statistics enable predictions to be made. Instead of talking about actual formulas or complex theories, he tells stories that give examples of predictions that have failed and those that have been successful. Personally, I would have liked to have been exposed to more of the math; however, I recognize that it would not be for everyone.

The book is divided into 13 chapters, and each chapter aligns with a different example of predictions that are successful and those that are not (more or less). Admittedly, some of the chapters were less interesting to me than others. Nevertheless, I feel as though I was able to learn some very important concepts in each chapter. For example, the author's background is in statistics, and he made a name for himself with baseball statistics. I never thought much about baseball statistics, yet I learned something incredibly valuable from this chapter. The author made a comparison between baseball predictions and presidential election predictions. I would have thought that presidential elections would be far richer in data (because of the magnitude of important); however, that is far, far from the truth. Presidential elections happen only once every four years--and there have been only 57 presidential elections in the entire history of the United States. In contrast, there are more than 57 games of baseball played every single year. The dataset in baseball is insanely rich, and what we learn about predictions in baseball can carry over into other data-rich fields.

Another field that is rich in data is weather prediction. I have never wanted to become a meteorologist in my life. I struggle with getting predictions wrong so often. Even so, this chapter was fascinating because the author describes why it's far easier to predict good weather than it is to predict bad weather. It is those bad weather predictions that seem to go wrong so often that make people question the skills of meteorologists, yet it is statistically less inaccurate than we might think. Another thing that I learned in this chapter is that there is a very, very big difference between meteorology and climatology. Climatology attempts to predict weather patterns over many, many years (e.g. 60 to 100); however, meteorology attempts to predict daily weather patterns. Over the longer duration, it is less challenging to predict weather patterns. It is far more difficult to predict daily fluctuations than it is to predict long-term trends. Incidentally, this heuristic holds true in other fields that the author described in other chapters. The stock market is a perfect example of this. The long-term trends in the stock market are much easier to predict than the daily fluctuations.

Far and beyond, my favorite chapter was the one that covered the game of chess. Chess is my favorite game by far. I am fascinated by the game because, unlike poker (described in a separate chapter of the book), you know everything that your opponent knows. All of the pieces are on the board in front of both players. There are no cards that are being held in your opponent’s hand that you have to guess about. Moreover, you know every possible move that is allowed by both you and your opponent. Even so, with all of that knowledge, people still lose at chess. It seems inconceivable that there were ever be anything other than a draw, yet it happens all the time. Good players win. The author talks a lot about how difficult it is to make predictions about the best play in chess because humans are only able to think about two or three moves at a time. Those players who can think out longer moves seem to do better. Enter the computer. Computers have the ability to calculate more moves in less time than humans. This provides computers with far better predictive power than humans. After Deep Blue defeated Garry Kasparov in 1997, the face of computer chess changed significantly. The author predicts that a human will never again be able to defeat a computer at chess (at that level of competition).

In the end, this last example is what seems to be the framework on which the entire book is built. Technology has changed the way in which we make predictions. Computers have the ability to process more data than ever before. And more data exist than ever before! New fields, like data analytics and big data, are pushing the boundaries of what computers can do with large datasets and their utility in prediction. Of course, some systems (like the weather) are less predictable than others (like chess); however, technology is enabling us to get closer and closer to more precise predictions. The author feels as though this ongoing advance in data and technology will ultimately be helpful in more and more fields include homeland security and the war on terror. I, for one, can't wait to see where it leads us.

Nate Silver has a fairly straight-forward style and gives lots of good information and stories illustrating his points. The narrator's tone occasionally seems terse, bordering on annoyed. This might have been a result of having to repeat some of the same language many times over.

Thought provoking as you listen to pundits and realize they don't really know what they are forecast or predicting.

The author used is money ball type experience and brings that analysis to examples range from politics to weather to hurricanes and earthquakes.

no

boring - or a documentary ;)

This was a great book, loved the discussions on baseball, odds and the many topics the author covers. Its a very hard book to describe though. I never got bored and every chapter offered me something new and fresh. Be a great book for anyone in marketing or SEO. I would recommend it. Worth the money.

There is a lot of background material that goes over prediction for climate change, stock market and terrorism which is okay but the real mechanics of bayes is really only a small portion of the book.

There are other books that cover bayes in more in depth so that's fine but it felt like 75% of the book was talking about backgrounds that really what you can do step by step to do better.

Maybe if I didnt get the unabridged version, it might have been more concise to what I was looking for.

People who need to know why making predictions is difficult.

Write more about how to make a good prediction - what works?

Too mechanical.

-

It didn't really answer the second part of the question: "and some don't".

Then you would have this book. The anecdotes are well told, but there is nothing revolutionary in this book. Nate Silver just reminds us of how bad humans are at just roughly figuring out statistics, then roughly tells you how to get better at forecasting and making predictions. My advice: Read the wikipedia page on Bayes' Theorem and read Thinking, Fast and Slow by Daniel Kahneman. You will have a better time.

bayesian statistical predictions.

I am being forced to write additional words to meet a 15 word minimum even though I was asked to sum up the book in 3 words

just read it. it's fun for all ages!

I would never sit long enough to read this book in print. It is very "dense" with information and statistics, which it should be because it is a book about statistics, but for the casual or business reader, there is just too much detailed information. It really would have been better as separate books on each topic. If you use statistics for a living, this is a must read.

The most interesting idea was that most studies report erroneous results and no one calls them on it or reports the errors, meaning that most of what we KNOW to be true, isn't. Bayesian analysis was very interesting and useful. Would have liked more information on how to apply it to everyday life and business situations.

Nate does what no one else has managed to do. Write a comprehensive book about statistics and probability that is interesting and informative. I loved listening to this book.