isabellesbooks's review against another edition
3.0
This book was a mixed bag of sorts for me- I was captivated by the chapters about weather predictions, natural disasters, and terrorist attacks, yet found myself bored through the multiple long chapters about sports and politics. It’s very well written but the commentary at times was a bit much for me, when I really just wanted the data. I did learn a lot and noticed this gradually turned into my nightly audiobook that I used to help me fall asleep (interpret that as you may).
mcparks's review against another edition
4.0
I really loved this book, especially the chapters about politics, the economy, the stock market, and political pundits. I think probabilistic thinking is great and would love more people to understand how it logically applies to everything we do. I say this as a geologist who gets asked, "Will the Big One happen?" at every party I go to, and I wish people would understand that the answer is "Yes!"
While I loved 95% of the book, I had a lot of problems with the chapter on seismology. I don't understand what Silver's conclusion about seismology is, other than that he feels that they had "failed" at predicting earthquakes? I found it ironic how critical Silver is of seismologists because everything I know about statistics, probability theory, and Bayesian probability I learned in various seismology classes.
What frustrated me most of all was including L'Aquila in the seismology chapter. I think L'Aquila was a clear case of seismologists doing probabilistic forecasts the right way. The only lesson learned in L'Aquila is that more outreach needs to be done to catch up civic authorities and building codes to modern earthquake engineering. The fact that residents thought that sleeping outside of their own homes was a solution to earthquake risk makes as much sense as telling your kid not to bike too fast instead of wearing a helmet.
My other observation about the seismology chapter was a frustrating dismissal of some seismology fundamentals. For example, he suggests that there could be greater magnitude earthquakes than seismologists indicate(which is based on rupture area scaling with magnitude) and he overemphasizes the Gutenberg-Richter relationship (which is best applied to source areas whereas individual faults can have characteristic frequency-magnitude relationships). I found this to be ironic, because Silver states that a big problem is statisticians ignoring basic science, while Silver rejects some basic seismology to focus on Gutenberg-Richter plots to prove his points. However, I will give Silver major props for writing a good explanation of why the Parkfield prediction failed.
While I loved 95% of the book, I had a lot of problems with the chapter on seismology. I don't understand what Silver's conclusion about seismology is, other than that he feels that they had "failed" at predicting earthquakes? I found it ironic how critical Silver is of seismologists because everything I know about statistics, probability theory, and Bayesian probability I learned in various seismology classes.
What frustrated me most of all was including L'Aquila in the seismology chapter. I think L'Aquila was a clear case of seismologists doing probabilistic forecasts the right way. The only lesson learned in L'Aquila is that more outreach needs to be done to catch up civic authorities and building codes to modern earthquake engineering. The fact that residents thought that sleeping outside of their own homes was a solution to earthquake risk makes as much sense as telling your kid not to bike too fast instead of wearing a helmet.
My other observation about the seismology chapter was a frustrating dismissal of some seismology fundamentals. For example, he suggests that there could be greater magnitude earthquakes than seismologists indicate(which is based on rupture area scaling with magnitude) and he overemphasizes the Gutenberg-Richter relationship (which is best applied to source areas whereas individual faults can have characteristic frequency-magnitude relationships). I found this to be ironic, because Silver states that a big problem is statisticians ignoring basic science, while Silver rejects some basic seismology to focus on Gutenberg-Richter plots to prove his points. However, I will give Silver major props for writing a good explanation of why the Parkfield prediction failed.
roshk99's review against another edition
4.0
Excellent read with lots of overlap into the fields of AI, robotics, machine learning, economics, and more. The book is contains several domain examples including baseball, poker, earthquakes, etc., and I found some more compelling than others. His overall recommendation seems to be the use of Bayesian priors to inform predictions as well as more precision on what/how predictions are made. The tone is quite conversational, but it does get technical in bits (though he tries to simplify it for the lay-reader). Overall, a very informative read, and especially timely around election season.
mbrousil's review against another edition
5.0
Great book, and an appropriately timed reminder of proper expectations and practices for evaluating information, models, and predictions.
"The volume of information is increasing exponentially. But relatively little of this information is useful - the signal-to-noise ratio may be waning. We need better ways of distinguishing the two."
"The volume of information is increasing exponentially. But relatively little of this information is useful - the signal-to-noise ratio may be waning. We need better ways of distinguishing the two."
mark_lm's review against another edition
4.0
Reading this book is like going for a walk with Nate Silver while he discusses finding the signal in the noise in situations from his life and that he has researched for you. The subjects are wide-ranging. All are centered on the idea of prediction, but prediction is defined loosely enough to cover almost any analytical problem. Topics include the 2oo8 financial collapse, the inaccuracy of predictions in politics and economics, baseball statistics, weather forecasting, earthquake prediction, predicting next year’s influenza variant, professional poker, chess, Bayes theorem (explained with unusually clear charts), and the difficulty in predicting military and terrorist attacks (as the author says, Where our enemies will strike us is predictable: it’s where we least expect them to.).
crissb's review against another edition
2.0
1.5, rounded up to 2. I found a few things interesting in this book, but it seemed about 3-4 times longer than it needed to be. The comparisons of commercial vs. government weather forecasts, poker and terrorist forecasts were compelling for me. The rest I could have skipped over.
Listened to it as a book on audio, and still took me many months to power through it.
Listened to it as a book on audio, and still took me many months to power through it.
jtlars7's review against another edition
Covered many themes I’ve read elsewhere(Tetlock, Kahneman, etc.), but worth reading too for its many interesting examples and emphasis on risks of making forecasts from big data without a grounding in theory.
emgusk's review against another edition
2.0
I'm a gal who loves baseball, polling and statistics, so you'd think this would be a favorite book, but I just didn't love it. Silver's writing is accessible, the subject matter is certainly interesting and I definitely learned something from the book, but I didn't find myself saying "hmm!" all that often. It's split up into defined chapters (one on polling, one on weather forecasting, one on sports betting, one on earthquakes, one on chess, another on terrorism, another on poker, and clearly one on baseball) but call on each other for background. I liked certain chapters far more than others (I found chess a bore, but learned much from the weather chapter). Interested to hear what the rest of you think (and how you think he'll do at ESPN).
mikiher's review against another edition
3.0
This book advocates taking the Bayesian approach in dealing with hard prediction problems, and analyses a number of cases, including baseball, chess, poker, climate, and terrorism. The book is uneven in quality - while there are some highly interesting and entertaining chapters, there are also some very dull ones (for example, I just cannot relate to the baseball case - what Americans find in this dull, endless game I will never understand). But in general, I enjoyed it.