Book Review: The Black Swan, by Nassim Nicholas Taleb
I'm not sure why this turned up in the work book sale, as it's the 2010 edition; maybe it was filler in part of a package shipped for review, or discovered at the back of a cupboard somewhere. Anyhow, it had been on my Amazon list for quite a while.
I had high hopes for this book, which were only partly fulfilled. The difficulty is not with the subject, so much as with its presentation. Taleb's writing frequently comes over as arrogant and without empathy (particularly ironic as in the appendix he has a section, devoid of sensibility, on Asperger's, in which he discusses this very trait); he also regularly exhibits a Gove-like disdain for "experts". This later becomes clarified to be specific to economics and social sciences, but the first impression already damaged the author in the mind of this reader. He also seems to be unable to resist any opportunity to take a dig at the French; even I find this out of place.
There are two main stems to his argument: that real-world probability distributions, particularly in economics and risk, are not Gaussian and therefore conventional wisdom is of little value as it discards outliers; and that "Black Swan" events (positive or negative) are the only events that matter. The first of these is reasonable; I once read that much of economics was based on classical thermodynamics, but I can't remember whether the article was dated 1st April. He presents some arguments for the second - in particular a graph showing the relative return of the stock market over 50 years, with and without the 10 days with greatest percentage changes, is compelling. But this doesn't entirely detract from the idea that, as these outliers happen rarely, most of the time the experts would probably be right. (In the appendix, he produces an example in which he appears to fail to distinguish between a "1 in 1000 chance" and a "1 in 1000-year chance" - unless one knows the underlying frequency of the event such a comparison is meaningless). There are a few real insights - drawing on the inapplicability of certain methods and distributions because events are not independent, but are in a complex feedback loop, and linking this to fractals and the work of Mandelbrot, for example - but these are buried quite deeply and probably don't impinge on readers not already with at least a passing familiarity with those concepts and work. His bar-bell strategy, combining the extremes of risk, seems valid, but it appears written from the perspective of someone with already substantial means; the ordinary person is unlikely to reap much real reward from it if applied rigorously.
The book was initially published in 2007. The appendix, added in 2010, moderates the tone to some extent. Taleb seems to have learnt some of the art of politics, but still can't resist lauding it over those whose models didn't predict the financial crisis. He doesn't offer alternative models; but rather than just saying "you're wrong", he now says "your model doesn't work well in these kinds of situations". It's an interesting book, but caveat emptor.