Unfortunately, my view is that unless you have a professional or academic imperative, I would suggest that you do not bother reading The Black Swan, by Nassim Taleb. I would suggest reading two or three reviews (I have listed some at the end of this post), because there are some good points and interesting ideas in his book, but they are spread very thinly and surrounded by much that is sloppy and/or simply wrong.
Perhaps I should mention that there are declarations of interest which might lead some readers (or at least Taleb) to think I may be biased; I have listed these at the bottom of this post.
The good bits
Before I explain why I recommend that you don’t read the book, I’d like to highlight the good bits, because these are quite important.
Black Swans are Taleb’s name for unpredicted major events that change the world, some for good, some for bad. Amongst the Black Swans he cites are Google and 9/11. His choice of title comes from the fact that for many centuries Europeans believed that all swans were white. When Europeans became aware of Australia in the 17th Century they found, to their horror, black swans. This choice of language is quite powerful in that is specifically includes unexpected events that had been considered and ruled out.
Taleb has a really useful contribution on how humans seem programmed to look for and assume causality, and how this tendency can lead to errors. I will come back to this, in one of my upcoming notes, as it relates very strongly to market research.
He has some useful observations on survivor bias, but he weakens his position by not recognising that this is a widely accepted issue, and by inventing his own name for it.
I fully support much of his attack on the way funds are invested. During the ten years I spent as a pension fund trustee I didn’t come across a single fund manager, actuary, or consultant who seemed credible. Most of them seemed to have adopted betting shop strategies, but had then given them fancy names such as asset allocation strategies and hedging. However, most of this material can’t be generalised to non-fund situations.
When Taleb talks about luck and randomness he is onto something very important and under appreciated. In order to succeed you need to be good and lucky. When we analyse the success of something or somebody, we often pay too little heed to luck and therefore too much to what they actually did. If we accept that random events play a role in success, then we should assume that, even in theory, we can’t create a model which fully measures and predicts how to be successful.
The bad bits
Despite the fact that he claims his approach is based on scepticism and empiricism, and that he advises against categorising people, the book is packed with cases where he has generalised from the particular, including many cases where the even the particular seems suspect. For example, he asserts that all London taxi drivers, all the time, seek the route with the maximum amount of traffic, in order to charge more. Some taxi cab drivers, some of the time, would be a plausible observation, but his confidence in a) generalising to all taxi cab drivers, and b) in being confident about the causal connection between the intent of the driver and the route strike at the heart of what he claims his method to be.
In most of the cases where he makes an observation about the real world, in an area where I too have experience, I find his observation to be overstated, over generalised, and his ascribed causality either weak or unnecessary. The main exceptions are his comments about the finance industry, where I find many of his observations to tally with my own (much more limited) experience.
He asserts, repeatedly, that all statisticians, politicians, and economists use the Gaussian distribution, without fear and without exception. I have never attended nor taught a statistics course that did not spell out the limitations and assumptions of the Gaussian or normal distribution. Step one is using it to assess whether it will hold in the circumstances under examination. He is also wrong to suggest that it is used exclusively, non-Parametric statistics are taught at all levels. In market research we have seen a growth in the last 20 years of distributions such as the Logit distribution, which does have the scale invariance property Taleb is looking for.
Taleb highlights the problems with predictions and plans, but does not present an alternative and does not recognise that plans and predictions are often useful. For example, in my life as a local government councillor we have to make predictions about how many children will be attending school in 5, 10, and 15 years time. When we make predictions, we always get it wrong, but usually we get it wrong by a small enough amount that tactical initiatives can deal with the problem. For example, if we build a school that is too small, we end up adding temporary classrooms, which is less than ideal. If the school is too big we have wasted tax-payers money which is also less than ideal. But, if we had not planned then there would not be a school, which is worse.
As far as I am concerned, and in the opinion of many of the politicians and bureaucrats that I work with, the important thing about planning and predictions is to assume they are not correct. We make estimates and then we almost immediately start trying to find out if we were too high or two low. The prediction process gives us a strategic framework around which we can make tactical corrections. Speaking about the Normandy landings, during the Second World War, Eisenhower said of the planning process, “the plans were useless, the planning was invaluable”.
Perhaps my biggest problem with Taleb is that he is untruthful. In Chapter Two Taleb uses the example of an author Yevgenia Krasnova whose book turns out to be a Black Swan. In other chapters reference is made to this case and to a subsequent book. However, there is no such person, Taleb simply made her up. He acknowledges this in footnote in Chapter Three, something which could easily be missed (a footnote is for optional reading as far as I am concerned). As soon as I reached this point I realised a) no fact in the book can be trusted without being checked, b) Taleb is presumably too lazy to find real examples, or he can’t find real examples, c) I could never recommend that somebody read something so untruthful/sloppy.
The book is also written in the style of somebody in the pub sounding off, with clumsy slang, mindless stereotypes (Japanese culture is limited, people with poor French are to be laughed at, qualified people he disagrees with are idiot savants, and all history books are dull), and a general lack of respect for the reader.
My five key takes from the book
- Beware of Black Swans, they are real, they do happen, and they happen more than we tend to think.
- Most predictions are wrong, we need to do more to assess the size of our prediction errors.
- Luck (also known as random events) plays a major part in any success or failure.
- Beware of causality and survivor bias, just because a particular series of events for person A was followed by success, it does not mean the same series resulted in success for other people who did the same thing.
- The distribution of achievements by fund managers (e.g. the people looking after your pension) is about the same as you would get by random chance, i.e. there may be no such thing as a good fund manager, just some who have been lucky in the past.
Other reviews
Here are three reviews of the book, the first two are pretty negative, the third is positive.
The New York Times Book Review
The Scotsman
The Telegraph
Declarations of Interest
- I am by profession a market researcher, i.e. somebody who spends some of their time trying to predict what customers might do.
- I often act as a statistician and have taught statistics, mostly in the context of the market research industry.
- I do scenario planning and future related consultancy.
- I am in a small way a politician, having been an elected local government councillor since 1983. Taleb suggests that there are no legislators “with intellect, courage, vision, and perseverance”.
- I have a degree in economics, something he seems to associate with the dark arts.
- I have never met Nassim Taleb, and it is very rare for me to dislike anybody, but from his description of himself I think I would dislike him very much, mostly for his cruel and condescending attitude to some people, but also for his cavalier dismissal of the importance of truth.
I would hope that none of these ‘interests’ have coloured my thinking, but they may have, and readers may think they have, even if I don’t.
p.s. so why did I read the book all the way through, if I am so unhappy about it? If I like a book, I feel quite happy to recommend it without reading all of it, because I know some of it is good. Normally, if I don’t like a book, I don’t finish it and I don’t blog about it. This is the first case where I feel so disappointed by a book, probably because it promised so much (it could have been a really good book), that I felt the need to blog about it. And, if you feel the need to be negative about a book, I think you need to be 100% sure that the book does not fully redeem itself at some later point.
Thanks for helping to make a bit more sense of this book.
I found some of Nassim's journals that help explain some of the reasons for the overall impression being a bit loose and have appended them as a comment.
Posted by: Jason | June 25, 2007 at 06:18 AM