The Signal and the Noise: Why So Many Predictions Fail - But Some Don't

Interesting book about forecasting, statistics, and why simply having more data is not going to result in better predictions. The writing is mostly entertaining and accessible to all, but if you're interested in the details there's enough there that I was able to correctly answer 3/4 questions on a Bayesian theory test a friend coincidentally posted on Facebook while I was reading this book.

The direction seems a little scattered though, it's more like a series of case studies or vignettes without a clear and cohesive direction. The most important information in the book (in my opinion) is Bayesian theory and how we can and should use it to keep our forecasts realistic; yet it isn't mentioned till over half way into the book and then isn't consistently emphasized through till the end. The rest of the book is examples of predictions gone right or wrong and examinations why; interesting but a little disjointed seeming at times. Still, very interesting read and worth picking up.

It goes without saying that "popular statistics" book is mostly an oxymoron. On the one hand, statistics is largely a very dry field. On the other hand, those of us who do understand statistics (and even freaks, like my husband, who enjoy statistics), find any attempt at popular statistics largely too elementary to be interesting. Nate Silver doesn't just walk the fine line in the middle, he eliminates it and writes a completely novel statistic book that is appealing to both the mathematician and the math hater: this book fascinates.

Nate Silver focuses on the forecasting in areas that are difficult to predict: weather, climate, earthquakes, poker, politics, chess and sports. Each of these areas is individually interesting -- I had never spent much time considering online poker, for instance, and the chapter focusing on poker is not just mathematically-focused, but also an expose on the world of online poker and the life and times (or at least the two year subset thereof) of Silver's 6-figure gambling career. In addition, his overall thesis, which seems to be that we should use Bayesian analysis to think probabilistically about the world and continually evaluate our probabilities both builds naturally and has far-reaching applications.

I feel like I have spent years of my life trying to explain to medical students (and more advanced physicians who should really know better) why every time a paper is published with a p<0.05 we can't totally disregard all prior medical knowledge and dive after the new information. Silver's easy explanation of Bayes' theorem nicely summarizes why this is true - that alone should make this a must-read for anyone in an academic field.

Given the technical nature of what Nate Silver does, and some of the early mentions of the book, I had higher hopes for the technical portions of the book. As usual for a popular text, I was left wanting a lot more. Again, the lack of any math left a lot to desire. I wish technical writers could get away with even a handful of equations, but wishing just won't make it so.

The first few chapters were a bit more technical sounding, but eventually devolved into a more journalistic viewpoint of statistics, prediction, and forecasting in general within the areas of economics, political elections, weather forecasting, earthquakes, baseball, poker, chess, and terrorism. I have a feeling he lost a large part of his audience in the first few chapters by discussing the economic meltdown of 2008 first instead of baseball or poker and then getting into politics and economics.

While some of the discussion around each of these bigger topics are all intrinsically interesting and there were a few interesting tidbits I hadn't heard or read about previously, on the whole it wasn't really as novel as I had hoped it would be. I think it should be required reading for all politicans however as I too often get the feeling that none of them think at this level.

There was some reasonably good philosophical discussion of Bayesian statistics versus Fisherian, but it was all too short and could have been fleshed out more significantly. I still prefer David Applebaum's historical and philosophical discussion of probability in [b:Probability and Information: An Integrated Approach|3623935|Probability and Information An Integrated Approach|David Applebaum|https://d202m5krfqbpi5.cloudfront.net/books/1266767217s/3623935.jpg|3666736] though he surprisingly didn't mention R.A. Fisher directly himself in his coverage.

It was interesting to run across additional mentions of power laws in the realms of earthquakes and terrorism after reading Melanie Mitchell's [b:Complexity: A Guided Tour|5597902|Complexity A Guided Tour|Melanie Mitchell|https://d202m5krfqbpi5.cloudfront.net/books/1348191495s/5597902.jpg|5769253], but I'll have to find some texts which describe the mathematics in full detail. There was surprisingly large amount of discussion skirting around the topics within complexity without delving into it in any substantive form.

For those with a pre-existing background in science and especially probability theory, I'd recommend skipping this and simply reading Daniel Kahneman's book [b:Thinking, Fast and Slow|11468377|Thinking, Fast and Slow|Daniel Kahneman|https://d202m5krfqbpi5.cloudfront.net/books/1317793965s/11468377.jpg|16402639]. Kahneman's work is referenced several times and his book seems less intuitive than some of the material Silver presents here.

This is the kind of text which should be required reading in high school civics classes. Perhaps it might motivate more to be interested in statistics and science related pursuits as these are almost always at the root of most political and policy related questions at the end of the day.

For me, I'd personally give this three stars, but the broader public should view it with at least four stars if not five as there is some truly great stuff here. Unfortunately a lot of it is old hat or retreaded material for me.

The book is an impressive tour-de-force through a great range of topics. Nate Silver discusses forecasting in the areas of political elections, baseball, weather, earthquakes, the economy, infectuous diseases, basketball, climate change, chess, poker, Bayesian theory, war/terrorism...

The insight into each topic is fascinating, and one gem of the book is a step-by-step procedure on using Bayes' theorem to get a confidence estimate for various scenarios, and how to update it based on new observations. I felt that it was a well-done explanation for most readers.

I give the book 5 stars because I liked the topics so much, as well as his insights into them. On the other hand, I should caution that I found two problems with how the book is written:

1. the choice of a target audience is not clear: some passages introduce a topic from scratch, while others assume some advanced knowledge to make any sense at all. I am a scientist and I have played chess and poker and I have watched basketball, so almost all the book was accessible. On the other hand I know nothing about baseball, and I never understood any of the discussion because it assumed familiarity with how player statistics work. When he would give a number, I didn't even know if it had to do with batting, pitching or catching. So that one chapter had reduced impact on me, but I can see people being completely lost in the chess chapter if they don't know the game well. On the other hand, the poker chapter has a complete introduction to the game as if someone did not know it, so that tone is uneven.

2. the writing style jumps around a bit, alternating between colloquial, anecdotal, journalistic, somewhat-scholarly, ... This felt a bit weird as I read the book, although it does not hurt the teaching effectiveness.