Black Swan and Induction Problem

Let me first confess that I have not finished the book yet and I have no authority to write on the entire book. However, I still believe that the parts I have read (the first three chapters) so far forced me to write this blog entry. Actually, I am supposed to write something against the nuclear reactors or on the bombing of Libya (I can hear the grumbling sound of a certain person) but since I am reading Nassim Nicholas Taleb’s “The Black Swan” whenever I have time from studying Financial Maths or Maths of Risk, I will write about his book and his ideas on induction.

The problem is induction. Since the ancient times, it is one of the biggest problems of epistemology: How do we know what we know? David Hume coined the term as “Induction Problem” as it is the inductive reasoning which seems creating the problem on the source of the knowledge. Many philosophers attacked this issue and attracted a lot of attention from the readers who already know what the writer would be talking about when it comes to induction problem. Basically, it can be rephrased by “How can knowing that sun has risen every morning for millions of years guarantee that it will rise tomorrow morning again?” The classic answer is of course “It does not”. But what misses in this answer is the continuation: “So what?”

Taleb attacks on the same issue like all other philosopher before him (by the way, he calls himself a philosopher but for me he is just another investment analyst with a few philosophical decorations in his sentences albeit he despises bankers, investors etc in the same fashion that push your own kind down so you will look superior. I can call him a “pop-philosopher” with bubble ideas which are not his own including the idea of “black swan” which can be considered as a page from one of the many books written on philosophy of science or history. He is not revolutionary enough to cut the branch he is sitting on.) and he creates this feeling that it is something new. The example he chooses actually used by Russell to identify the induction problem in philosophy of science. Here is the example:

Let’s think of a turkey (I chose a female one to differentiate from the male writers and the farmer) who thinks that the owner of the farm feeds her every morning at 8 am. After a certain amount of days, she will think that the owner will bring food every morning and she will have a comfortable life without worrying about what she will eat next day. Of course, when the turkey becomes fatty enough and the thanksgiving day reaches, the owner comes to the coop with a large knife and chops the head of the turkey for the dinner. Tadaaaa! The turkey was wrong. What she believed after experiencing n days was wrong as in n+1th day, the hypothesis failed and it cost her life.

No, it did not cost her life and it did not give her more harm than she would be having without her belief in induction. In fact, not believing induction would not give her any extra day to live (being a paranoid does not guarantee that no one chases you). The problem in this metaphor is to look at the story from the perspective of turkey’s life and her loss but not her gain. But let’s look at the other way around. Suppose the turkey is skeptical and after n days, she still suspects that next day there will be no food. This means, everyday she will spend her all time worrying about the food for the next day and do nothing else. At least, induction gives her a chance to worry about other things, maybe to increase the quality of her life or think about how to lay bigger eggs. Her biggest mistake is not to believe in induction but not to believe in induction enough. Because if she believed it enough, she would be reading the history of turkeys in the farm and learn that all turkeys are raised till a point and many of them lose their lives on thanksgiving day. This way, she can use non-stationary time series and create a periodical pattern. This pattern would help her to understand the periodical movement of lives of turkeys in the farm and can help her to create mortality tables. By the way, since Taleb believes that on a thanksgiving day the turkey will lose her life, he himself becomes a turkey in his own criteria… (how does he know a turkey lose her life on thanks giving day)

Yes, there is no way we can guarantee that n+1^st experiment will give us the same result as the first n experiments gave but this skeptical thought is never a problem for a practicing scientist. In fact, this is how science develops. If we give the turkey second chance (as scientists do not usually die after they fail in their hypotheses), I am sure she will be smarter in the next term and will start building an airplane to escape from the farm. With the wrong n+1th experiments, science develops; scientists purify their theories and refine them in the best way so that it can serve to human kind.

It is hard to be against his ideas on how history moves on (like the quantas in the subatomic world): It obeys the rules of normality till the revolutions occurs (Impact of highly improbable) then we turn to the realm of normality again... However, these ideas are not original either. So many modernist / postmodernist philosophers made their way with similar ideas and watching the news everyday (tsunamis in Japan shuts down nuclear reactors in other continents, a Tunisian poor man's fight for honor causes a democratic/social (but not political) revolution in Egypt and most probably the invasion of Libya by western powers) , it is impossible to claim the opposite.

I will write a longer review on Taleb’s book once I finish it but so far my impression is it is a bubble book with big words and small ideas. Yes, statistics is full of assumptions and it has so many flaws which are easily exploited by the analysts but at the same time there is nothing wrong with normal distributions and other common distributions. They are just models which fit to numerical perfections. Our world is not perfect and no real phenomena can fit into any mathematical model in 100% fashion. Statisticians, like all other scientists, learn by their errors and refine their models in long run. Denying their role in modern life will be denying the role of technology and science in our lives because without induction no science would ever be possible and we would be living in Stone Age now with fire and wheels (if we are lucky). Induction is a psychological result of our desire to have a comfortable tomorrow and it works perfect in that way. Whenever it fails, we tumble down and then we stand up again, stronger and bigger each time compared to the previous time.

The book seems criticizing other issues like over-confidence of analysts who believe in mathematical equations and/or efficient market hypothesis (and many others underlying assumptions) but unfortunately instead of mocking the assumptions they have and the system which allows people to exploit the other people (making money from money without producing anything: what is good about it?), he attacks the big picture without differentiating the good from the bad. I hope the rest of the book will have more balance on its logical arguments and the claims against the common knowledge of common men.

To be continued…