I had picked up Nassim Nicholas Taleb's book The Black Swan when I was in New York in the spring. It had been on my list of current social sciences literature classics to read for quite some time. Taleb quite obviously is an impressive erudite, scholar and practitioner. He is fluent in multiple languages and can (according to Wikipedia) read a number of classical ones in addition. His literary and scholarly references abound in The Black Swan.
And yet, he is not a great writer. His style felt too much like someone highly intelligent trying to write popularly in order to scale the heights of the New York Times Bestseller List, which in deed he did. It made reading the book a bit annoying at times though I found, repetitive, and almost simplistic in some of his examples and anecdotes that stand in stark contrast to the more analytical or rare numerical parts.
Most important though is the theory of the Black Swan he proposes. I will not go into much detail on this, there is a whole Wikipedia article on it and I am not sure I'd even be capable of properly explaining his argument. Quickly put, Taleb argues that most - especially financial - statistical analysis is based on a no-outlier model of the world reflected say in body sizes where frequency groups around an average and there are fewer and fewer people at very high and low heights. Yet, this kind of model fails in the world of social sciences where a single event (a systemic crisis in the markets let's say, or the First World War) can make irrelevant any heretofore dominant average. Any model that does not take these black swans into account is doomed to fail especially as these kind of events happen much more frequently and are much more important in determining stock market moves (or world history if you may) than we presume.
A super interesting argument that came out right before the financial crisis in 2008 and that I cannot say I have read a good argument against so far. Note that especially his (semi-)technical chapters at the end make for a very good read as they are far less anecdotal and thus (ironically and at least for me) far more interesting and easy to read.