Imagine you are a turkey being fed comfortably on one of those mass production turkey farms. You may well assume that the good food, good company, and pleasant surroundings will go on forever. If you are a quant-savvy turkey, you might even gobble together a mathematical model that predicts good times well into the future, beyond not just Thanksgiving, but past Christmas and New Years as well. Suddenly in November, unexpectedly, with life-changing consequences...things change. You just didn't see it coming. Pass the cranberry sauce.
Financial planners, economists and other more sophisticated turkeys don't see it coming either, argues author Nassim Nicholas Taleb. His book highlights the danger of the unexpected. The unexpected will happen even if we have a comfortable model predicting only minor changes. After such a "black swan" catches us by surprise, we use our flawed hindsight to decide how we could have predicted the disaster using a better model. We are kidding ourselves, insists Taleb. We need better strategies to live in a world where truly random, unpredictable events occur. He goes to some trouble in this book, and his previous Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets, to educate us.
The flawed basis of many formal models is "the great intellectual fraud" of the bell curve. We learn that highly constrained variables like height and weight cluster around an average and that extreme variations from the average are unlikely. We just aren't going to meet anybody that's half a foot or half a mile tall. These "Mediocristan" models are fine until we misapply their assumptions to unconstrained "Extremistan" phenomena like stock values, book sales and such. We are slow to see this problem. We persistently commit the "Ludic Fallacy," clinging to our formal models because they seem more real to us than the messy, real-world events they are meant to explain. Taleb illustrates this point with examples ranging from the events of 9/11 to the "off model" problems that cost casinos money. S. I. Hyakawa warned us in the early `70s that "the map is not the territory," but we haven't learned.
Taleb also warns us of the narrative fallacy based on our love of stories. We feel we understand something when we can tell a story about why it happened--even after the fact, with only part of the relevant information. When musicians achieve fame and dramatic financial success, we backtrack through their histories, explaining success by what we see along the path. We don't see the hidden cemetery of failed garage bands and starving artists who did all the same things to no avail. Because we believe this artificial story, we don't have to face the role of randomness in success or failure. Or consider its impact on our own plans.
Taleb offers some suggestions--though fewer than I'd hoped for. He advises us to be open to positive black swans and guard against negative ones. Lending money at interest, for example, opens us only to a high impact negative. This worst case is that the borrower will go bankrupt and we won't get our money back. But the very best outcome is that the loan will be simply repaid. If the borrower's entrepreneurial effort is wildly, off-the-scale successful, the lender doesn't get any more than this. An investor, on the other hand, suffers the same risk of loss, but participates fully in an "Extemistan" success. Readers are left to ponder the implications--and perhaps to hire Taleb as an investment consultant.
Although Taleb does not venture there, some of his ideas are useful in applied psychology. Personnel tests, for example, rely on the principle of "behavioral consistency," assuming that our past actions best predict our future actions. If someone is a poor performer, the safe bet is that this person will perform poorly in future employment. This may be fit a general model, but employers--and psychologists who advise them--might consider whether we commit Taleb's fallacies. Are we so comfortable with are general predictive models, with our stories about how people "are," that we close ourselves to possible change? Wouldn't it be better to seek the occasional "gray swan" of improvement and hire the flawed job applicant? The author has convinced me that this is worth considering. My time reading this book was well spent.
One final note: The author's condescending tone has been mentioned by other reviewers. It's there all right. Yes, he is condescending. Yes, he sneers at his fellow financial analysts. Yes, his citations veer into name dropping. And, yes, he finds ways to not-so-subtly complement himself as he praises Benoit Mandelbrot. But none of this matters. Taleb's message is valuable. I recommend you ignore his tone--or perhaps even be entertained by it. Stay on task and learn something about the nature of randomness and prediction.