"Expect the unexpected" -- an aphorism that almost completely summarises the book. Cliches exist for a reason, but 196 pages later I feel the point has been well made.
Taleb is a stock market trader. As a trader, he believes that there is no way in general to predict the stock market -- that there are so many variables that the resulting stock price is indistinguishable from pure noise. Unfortunately, his profession is filled with people who believe that they *can* predict the market. In fact, some of these people trade very successfully for a while. Taleb belives that no trader can stay ahead of the market forever. The success of a small number of traders for a short amount of time can simply be explained by statistics: after all, given a large enough group of traders, some are bound to be successful some of the time even if their trades are entirely random.
This point is well illustrated with many examples, and this is where the book falters for me. Some of the examples are great (he goes into a lot of detail about various Monte Carlo simulations he has constructed), but some just seem like strawmen (he goes to a lot of trouble to describe the fancy cars, extravagant house, impulsive trading strategy, and hubris of a hypothetical trader, setting him up for his inevitable fall at the hands of the "unexpected rare event").
Some examples are just plain wrong, or at least misleading. For example, he goes into detail describing the work of Karl Popper, a philosopher who was interested in the question "how do you tell the difference between science and non-science?" Popper proposed several theories of science to answer this question.
The first, naive falsificationism, essentially states two things: firstly, that a theory is scientific if it can be falsified; and secondly, that a theory which has been falsified should no longer be used.
The second, sophisticated falsificationism, was a response to both the negativity of naive falsificationism and also a response to the criticism that science doesn't work that way (by immediately discarding theories the moment they contradict any evidence). Sophisticated falsificationism again states two things: that a theory is scientific if it leads to the discovery of novel (ie, new) facts; and that a theory is falsified by a new theory if the new theory can accurately predict everything the old theory could but *also* can accurately predict novel facts. General relativity is the canonical example of this: it incorporates Newtonian mechanics at low speeds but also explains things that Newtonian mechanics cannot at near-light speeds.
Anyway, Taleb's summary of all of this was "Popper's detractors called him a naive falsificationist." They did not, or at least not disparagingly, because Popper himself proposed the term, and labeled himself accordingly! I can understand why Taleb might not have wanted to include a huge explicative paragraph on Popper, but the book would have been just fine without the dubious example.
Another example of sloppiness occurs later, when Taleb talks about the QWERTY keyboard layout as an example of "the vicious dynamics of winning and losing in an economy" and repeats the old saw that the QWERTY layout was designed to slow typists down, and that it succeeded anyway because of a "snowball effect" of popularity. This claim is highly contentious (in fact one way to both speed typists up and prevent mechanical key jams is to put commonly-used letter combinations on opposite sides of the keyboard, something that QWERTY does to some extent but that Dvorak does much better), and once again the book would have been just fine without it.
Taleb is best when he is covering his area of expertise -- probability -- and the chapters dealing with probability are great. There are also several enjoyable and illustrative personal anecdotes which are a fun and memorable read.
Overall I enjoyed the book. I just wish he'd put more thought into his examples.