The book is quite easy to read and not uninteresting. However, if you have read a similar book before (e.g the Black Swan), you will probably not find anything quite surprising.
For some mysterious reason, the author completely ignores the contributions of physicists to the problem of getting signal out of noisy data.
For someone with a background in natural science or mathematics I recommend Jaynes "Probability Theory: The logic of science" (which exists in an unedited free version) as it tackles the Bayesian concept in much greater depth on the first few pages. Jaynes book - albeit in the later parts of the book that are mathematically quite challenging - properly discusses the problem of overfitting, whereas Nate Silver simplay (and unsatisfactory) states, that it is not easy to solve.
The author often discusses the problems of predictions in the context of gambling and sports. However, when you are not familar with Baseball, some discussions are hard to follow.