Wednesday, October 22, 2008

Books: The Black Swan; The Impact of the Highly Improbable

The Black Swan, by Nassim Nicholas Taleb is an engaging book about an important topic. However, if a reader who is familiar with the topic(s) will not find any new ideas.

This book is about risk management; specifically, how to deal with the risks arising from unpredictable events. To examine this issue, Taleb crafts a narrative that surveys a number of intellectual fields, including probability theory, complexity theory, epistemology (empiricism and philosophy of science), cognitive psychology, and social psychology. For a systematic summary, see Wikipedia.

The most striking thing about this book was Taleb's prescience with regard to the recent financial crisis. This book was published in April 2007, but Taleb (who made his "f*** you money" during the stock market crash of 1987) asserted that modern financial risk-assessment is total bullshit, and that when some unpredictable disruption occurs, the structure of the financial system will cause cataclysmic failure. Oh boy, was he right!

This book is supposed to be practical, so it's worth mentioning the advice provided. It can basically be summed up with two points:
  1. Know the limits of prediction (specifically, that some very important phenomena are fundamentally unpredictable). Don't fool yourself into thinking that you what the probabilities of events are (as if you were playing a game with well-defined rules)
  2. When you make predictions for unpredictable phenomena, ask yourself what are the consequences of being wrong. Don't estimate probabilities of being wrong--just avoid situations where being wrong will be fatal.
This book also has a few messages for how we think about social issues. One is that many "experts" are frauds, in that they pretend to be able to predict phenomena that are fundamentally unpredictable. The other is the randomness plays a major role in personal success, especially when we are looking at people who are extremely successful.

Taleb also warns us about putting much faith in predictions about what the world will be like in a few decades. He specifically dismissed predictions that the American Social Security system will be bankrupt in 40 years, implying that the funding of SS in 2048 is a concern for the people living in 2048, not for those of us living in 2008. By extension, we may also downplay predictions of radical climate change over the next few decades. However, both types of predictions do have value in that they provide us with a sense of the range of possible scenarios that we should be prepared for. [added after initial publication]

One more merit for this book is that it provides a good description of the life a scientist, including the methodological, psychological, and social aspects. I recommend this book for anyone interested in pursuing a career in science.

The main drawback of this book is that the author comes off as being quite arrogant. He is dismissive of many scientists, claiming that they make unjustifiable assumptions about the probability distribution of events. He even claimed that psychologists were "lucky" that they frequently used binary (Yes/No) measurements for which probability distributions ARE predictable. In fact, this is no accident. A careful scientist will design experiments in a way that the data conforms to certain assumptions, thereby allowing hypotheses to be tested. If you look in a book like Handbook of Parametric and Nonparametric Statistical Procedures, you will find extensive discussions about when it is appropriate to use various probability distributions.

Taleb also engaging in a form of name-dropping, describing how his arguments have gotten under the skin of a number of prominent intellectuals. This is the down-side of his narrative style of writing. However, the narrative of his life is what keeps this book interesting even if you are thinking "I know this already".

No comments: