I have read quite a few chapters. So far, this is the best thing that I learned from the book (reproduced in my own words):
Suppose, I claim that no human being is longer than 100 feet. Since we don't know the exact details of how height of human beings is controlled, there is no direct proof of the statement. Naturally, we'll have to resolve to inductive reasoning. That is, each example that you see will strengthen your "belief" in the claim. You see Mr. A; ok, he is shorter than 100 feet. You see Mr. B; he is just 6 feet. He also confirms the hypothesis, and so on.
Now suppose that you see a 99 feet man (perhaps, in a circus). Theoretically, that should strengthen your belief in the hypothesis even further (as 99 < 100). But in practice, this example shakes your belief!
I have become a fan of William Poundstone. The first book of his that I came to read was How Would You Move Mount Fuji? Microsoft's Cult of the Puzzle. I bought it because I had heard that they ask weird puzzles in MS interviews. But I was pleasantly surprised to see that half of the book was on the history of the term IQ, the formation of the silicon valley (trust me, it's very interesting), and a general discussion on what puzzles show and what they can't.