Monday, November 30, 2009

The Black Swan

I recently read "The Black Swan" by Nassim Nicholas Taleb. I found this to be an entertaining and thought provoking book that was worth reading. However it has definite flaws which others may find more off putting than I did.

The thesis of "The Black Swan" is that the course of events is greatly affected by rare, difficult to predict events, with large consequences. An example would be the 911 attacks. The title refers to the unexpected discovery of black swans in Australia surprising ornithologists who had come to believe from prior observations that all swans are white. Taleb calls unexpected events with large consequences "Black Swans".

Taleb develops this rather simple idea at great (300+ pages) length. Along the way he discusses some of the reasons the importance of rare events is underestimated. A particular target is statistical methods based on the normal (Gaussian) distribution which Taleb believes are routinely used in inappropriate ways.

Taleb has a provocative writing style. He clearly does not subscribe to the academic convention which requires you to carefully hedge and qualify your work and at least pretend to respect your opponents and their work. So Taleb routinely makes sweeping claims while dismissing contrary views as nonsense in the process casually insulting a wide range of people. I found this more entertaining than irritating even when I was in the line of fire (as on p. 204-205, "Furthermore, this trade-off between volatility and risk can show up in careers that give the appearance of being stable, like jobs at IBM until the 1990s. When laid off, the employee faces a total void: he is no longer fit for anything else. ..."). Still the academic convention exists for a reason.

One major flaw in the book is Taleb is prone to overstatement. The fact that forecasting is imperfect does not mean it is useless (as Taleb often seems to feel). For example on page 129-130 Taleb discusses risk management at a casino. They worried about and took measures to limit losses from lucky gamblers and cheaters only to be hit by among other things a large loss when a star performer was almost killed by a tiger something they had never considered as a possibility. Taleb concluded:

... The casino spent hundreds of millions of dollars on gambling theory and high-tech surveillance while the bulk of their risks came from outside their models.

But this is seriously misleading. It is quite likely the bulk of their risks would have come from within their models if they had not taken measures to reduce the risks they could foresee. This is like the claim that 40% of heat loss is through you head which is true when your head is the only uncovered part of your body.

So in summary lots of what Taleb says is true and often overlooked but things are not as one-sided as he claims.

1 comment:

  1. It's simply not true that the casino couldn't have predicted the performer's "accident" with the tiger since any fool knows tigers are dangerous. The casino just employed the wrong type of person to do the predictive model. If they had instead employed a Kindergarten teacher to observe, "Don't play tigers, they will eat you up," the story would have been quite different.