The Black Swan

by Nassim Nicholas Taleb

  Mandelbrotian Gray Swan: Black Swans that we can somewhat take into account—earthquakes, blockbuster books, stock market crashes—but for which it is not possible to completely figure out the properties and produce precise calculations.

  Mediocristan: the province dominated by the mediocre, with few extreme successes or failures. No single observation can meaningfully affect the aggregate. The bell curve is grounded in Mediocristan. There is a qualitative difference between Gaussians and scalable laws, much like gas and water.

  Narrative discipline: the discipline that consists in fitting a convincing and well-sounding story to the past. Opposed to experimental discipline.

  Narrative fallacy: our need to fit a story or pattern to a series of connected or disconnected facts. The statistical application is data mining.

  Nerd knowledge: the belief that what cannot be Platonized and studied does not exist at all, or is not worth considering. There even exists a form of skepticism practiced by the nerd.

  Platonic fold: the place where our Platonic representation enters into contact with reality and you can see the side effects of models.

  Platonicity: the focus on those pure, well-defined, and easily discernible objects like triangles, or more social notions like friendship or love, at the cost of ignoring those objects of seemingly messier and less tractable structures.

  Probability distribution: the model used to calculate the odds of different events, how they are “distributed.” When we say that an event is distributed according to the bell curve, we mean that the Gaussian bell curve can help provide probabilities of various occurrences.

  Problem of induction: the logical-philosophical extension of the Black Swan problem.

  Randomness as incomplete information: simply, what I cannot guess is random because my knowledge about the causes is incomplete, not necessarily because the process has truly unpredictable properties.

  Retrospective distortion: examining past events without adjusting for the forward passage of time. It leads to the illusion of posterior predictability.

  Reverse-engineering problem: It is easier to predict how an ice cube would melt into a puddle than, looking at a puddle, to guess the shape of the ice cube that may have caused it. This “inverse problem” makes narrative disciplines and accounts (such as histories) suspicious.

  Round-trip fallacy: the confusion of absence of evidence of Black Swans (or something else) for evidence of absence of Black Swans (or something else). It affects statisticians and other people who have lost part of their reasoning by solving too many equations.

  Scandal of prediction: the poor prediction record in some forecasting entities (particularly narrative disciplines) mixed with verbose commentary and a lack of awareness of their own dire past record.

  Scorn of the abstract: favoring contextualized thinking over more abstract, though more relevant, matters. “The death of one child is a tragedy; the death of a million is a statistic.”

  Statistical regress argument (or the problem of the circularity of statistics): We need data to discover a probability distribution. How do we know if we have enough? From the probability distribution. If it is a Gaussian, then a few points of data will suffice. How do we know it is a Gaussian? From the data. So we need the data to tell us what probability distribution to assume, and we need a probability distribution to tell us how much data we need. This causes a severe regress argument, which is somewhat shamelessly circumvented by resorting to the Gaussian and its kin.

  Uncertainty of the deluded: people who tunnel on sources of uncertainty by producing precise sources like the great uncertainty principle, or similar, less consequential matters, to real life; worrying about subatomic particles while forgetting that we can’t predict tomorrow’s crises.

  I

  LEARNING FROM MOTHER NATURE, THE OLDEST AND THE WISEST

  How to make friends among walking people—On becoming a grandmother—The charms of eco-Extremistan—Never small enough—Harvard-Soviet chic

  I am writing this essay three years after the completion of The Black Swan—which I have kept intact except for a few clarifying footnotes. Since then, I’ve written a dozen “scholarly” papers around some aspects of the Black Swan idea. These are very, very boring to read, since almost all academic papers are made to bore, impress, provide credibility, intimidate even, be presented at meetings, but not to be read except by suckers (or detractors) or, even worse, graduate students. Also, I am making the “what to do next” more salient here—you can take a horse to water and, in addition, you may have to make it drink. So this essay will allow me to go deeper into some points. Like the main text itself, the beginning will be what is called literary, and progressively turn technical.

  I owe the idea of this book-length essay to Danny Kahneman, toward whom I (and my ideas) have more debt than toward anyone else on this planet. He convinced me that I had obligations to try to make the horse drink.

  ON SLOW BUT LONG WALKS

  Over the past three years, my life experienced a bit of change, mostly for the better. Like parties, a book puts you on the envelope of serendipity; it even gets you invited to more parties. During my dark days, I was called a trader in Paris (something extremely vulgaire), a philosopher in London (meaning too theoretical), a prophet in New York (dissingly, because of my then false prophecy), and an economist in Jerusalem (something very materialistic). I now saw myself dealing with the stress of having to live up to the wholly undeserved designations of a prophet in Israel (a very, very ambitious project), a philosophe in France, an economist in London, and a trader in New York (where it is respectable).

  Such exposure brought hate mail, at least one death threat (by former employees of the bankrupt firm Lehman Brothers*), which I found extremely flattering, and, worse than any threat of violence, hourly requests for interviews by Turkish and Brazilian journalists. I had to spend a lot of time writing personalized and courteous notes declining invitations to dinner with suit-wearing current hotshots, suit-wearing archeo-hotshots, suit-wearing proto-hotshots, and the nasty brand of suit-wearing namedroppers. But it also brought some benefits. I was contacted by like-minded persons, people I would have never dreamed of meeting in the past, or those I did not think existed before, in disciplines completely outside my normal circles, who helped me further my quest with the most unexpected of ideas. I was often reached by people I admired and whose work I knew well, and who became natural collaborators and critics; I will always remember the thrill of getting an unexpected e-mail from Spyros Makridakis of the M-Competition described in Chapter 10, the great debunker of misforecasting, or another one from Jon Elster, the scholar of rare erudition and insights who integrated the wisdom of the ancients into modern social science thinking. I’ve met novelists and philosophical thinkers whose works I had read and admired, like Louis de Bernières, Will Self, John Gray (the philosopher, not the pop psychologist), or Lord Martin Rees; in all four cases I had the peculiar need to pinch myself upon hearing them talking to me about my own book.

  Then, through a chain of friends of friends, cappuccinos, dessert wines, and security lines at airports, I got to partake of and understand the potency of oral knowledge, as discussions are vastly more powerful than just correspondence. People say things in person they would never put in print. I met Nouriel Roubini (to my knowledge the only professional economist who really predicted the crisis of 2008, and perhaps the only independent thinker in that business). I also found a variety of people I did not know existed, good economists (i.e., with scientific standards), like Michael Spence and Barkley Rosser. Also Peter Bevelin and Yechezkel Zilber kept feeding me the papers I was looking for without knowing it, the first in biology, the second in cognitive science—thus they nudged my thinking in the appropriate direction.

  So I have been dialoguing with many people. My problem is that I found only two persons who can have a conversation during a long walk (and walk slowly): Spyros Makridakis and Yechezkel Zilber. Most people, alas, walk too fast, mis
taking walking for exercise, not understanding that walking is to be done slowly, at such a pace that one forgets one is walking—so I need to keep going to Athens (where Spyros lives) in order to indulge in my favorite activity, being a flâneur.

  My Mistakes

  And of course people will scrutinize the text. After examining messages and reports, I do not feel I need to retract anything in the initial version, or to correct any error (outside of typos and minor factual mistakes), except for two related matters. The first fault was pointed out to me by Jon Elster. I had written that the narrative fallacy pervades historical analyses, since I believed that there was no such thing as a test of a historical statement by forecasting and falsification. Elster explained to me that there are situations in which historical theory can escape the narrative fallacy and be subjected to empirical rejection—areas in which we are discovering documents or archeological sites yielding information capable of countering a certain narrative.

  So, in relation to his point, I realized that the history of Arabic thought was not so definitive and that I had fallen into the trap of ignoring the continuous changes in past history, that the past too was largely a prediction. I (accidentally) discovered that I had fallen for conventional wisdom in textbook scholarship on Arabic philosophy, a wisdom that was contradicted by existing documents. I had exaggerated the import of the debate between Averroës and Algazel. Like everyone I thought that 1) it was a big deal and, 2) it killed Arabic falsafah. It turned out to be one of the mis-conceptions being recently debunked by researchers (such as Dimitri Gutas and George Saliba). Most of those who theorized about Arabic philosophy did not know Arabic, so they left many things to their imagination (like Leo Strauss, for example). I am a bit ashamed, because Arabic is one of my native languages, and here I was reporting from tenth-hand sources developed by scholars illiterate in Arabic (and sufficiently overconfident and lacking in erudition to not realize it). I fell for the confirmation bias seen by Gutas: “It seems that one always starts with a preconception of what Arabic philosophy should be saying, and then concentrating only on those passages which seem to be supporting such a bias, thereby appearing to corroborate the preconception on the basis of the texts themselves.”

  Once again, beware of history.

  ROBUSTNESS AND FRAGILITY

  Upon the completion of The Black Swan, I spent some time meditating on the items I raised in Chapter 14 on the fragility of some systems with large concentration and illusions of stability—which had left me convinced that the banking system was the mother of all accidents waiting to happen. I explained in Chapter 6, with the story of the old elephants, that the best teachers of wisdom are naturally the eldest, simply because they may have picked up invisible tricks and heuristics that escape our epistemic landscape, tricks that helped them survive in a world more complex than the one we think we can understand. So being old implies a higher degree of resistance to Black Swans, though, as we saw with the turkey story, it is not a guaranteed proof—older is almost always more solid, but older is not necessarily perfect. But a few billion years is vastly more proof than a thousand days of survival, and the oldest system around is clearly Mother Nature.

  That was, in a way, the reasoning behind the epilogism argument of the medical empiricists of the post-classical Levant (like Menodotus of Nicomedia), who were the only practitioners to merge skepticism and decision-making in the real world. They are also the only group of people to use philosophy for anything useful. They proposed historia: maximal recording of facts with minimal interpretation and theorizing, describing of facts without the why, and resisting universals. Their form of nontheoretical knowledge was degraded by the medieval Scholastics, who favored more explicit learning. Historia, just the recording of facts, was inferior to philosophia or scientia. Even philosophy, until then, had more to do with decision-making wisdom than it does today, not with impressing a tenure committee, and medicine was where such wisdom was practiced (and learned): Medicina soror philosophiae: “Medicine, sister of Philosophy.”*

  Giving an ancillary status to a field that prefers particulars to universals is what formalized knowledge since the Scholastics has been doing, which necessarily gives short shrift to experience and age (too much accumulation of particulars), in favor of those who hold a PhD like Dr. John. This may work in classical physics, but not in the complex domain; it has killed a lot of patients in the history of medicine, particularly before clinical medicine was born, and is causing a lot of damage in the social domain, particularly at the time of writing.

  The central things the old teachers communicate to you are, to use religious terms, dogmas (rules you need to execute without necessarily understanding them) not kerygmas (rules you can understand and that have a purpose clear to you).

  Mother Nature is clearly a complex system, with webs of interdependence, nonlinearities, and a robust ecology (otherwise it would have blown up a long time ago). It is an old, very old person with an impeccable memory. Mother Nature does not develop Alzheimer’s—actually there is evidence that even humans would not easily lose brain function with age if they followed a regimen of stochastic exercise and stochastic fasting, took long walks, avoided sugar, bread, white rice, and stock market investments, and refrained from taking economics classes or reading such things as The New York Times.

  Let me summarize my ideas about how Mother Nature deals with the Black Swan, both positive and negative—it knows much better than humans how to take advantage of positive Black Swans.

  Redundancy as Insurance

  First, Mother Nature likes redundancies, three different types of redundancies. The first, the simplest to understand, is defensive redundancy, the insurance type of redundancy that allows you to survive under adversity, thanks to the availability of spare parts. Look at the human body. We have two eyes, two lungs, two kidneys, even two brains (with the possible exception of corporate executives)—and each has more capacity than needed in ordinary circumstances. So redundancy equals insurance, and the apparent inefficiencies are associated with the costs of maintaining these spare parts and the energy needed to keep them around in spite of their idleness.

  The exact opposite of redundancy is naïve optimization. I tell everyone to avoid attending (orthodox) economics classes and say that economics will fail us and blow us up (and, as we will see, we have proofs that it failed us; but, as I kept saying in the original text, we did not need them; all we needed was to look at the lack of scientific rigor—and of ethics). The reason is the following: It is largely based on notions of naïve optimization, mathematized (poorly) by Paul Samuelson—and this mathematics contributed massively to the construction of an error-prone society. An economist would find it inefficient to maintain two lungs and two kidneys: consider the costs involved in transporting these heavy items across the savannah. Such optimization would, eventually, kill you, after the first accident, the first “outlier.” Also, consider that if we gave Mother Nature to economists, it would dispense with individual kidneys: since we do not need them all the time, it would be more “efficient” if we sold ours and used a central kidney on a time-share basis. You could also lend your eyes at night since you do not need them to dream.

  Almost every major idea in conventional economics (though a lesser number of minor ones) fails under the modification of some assumption, or what is called “perturbation,” when you change one parameter, or take a parameter heretofore assumed by the theory to be fixed and stable, and make it random. We call this “randomization” in the jargon. This is called the study of model error and examination of the consequences of such changes (my official academic specialty is now model error or “model risk”). For instance, if a model used for risk assumes that the type of randomness under consideration is from Mediocristan, it will ignore large deviations and encourage the building of a lot of risk that ignores large deviations; accordingly, risk management will be faulty. Hence the metaphor of “sitting on a barrel of dynamite” I used concerning Fannie Mae (now bust)
.

  For another example of egregious model error, take the notion of comparative advantage supposedly discovered by Ricardo and behind the wheels of globalization. The idea is that countries should focus, as a consultant would say, on “what they do best” (more exactly, on where they are missing the smallest number of opportunities); so one country should specialize in wine and the other in clothes, although one of them might be better at both. But do some perturbations and alternative scenarios: consider what would happen to the country specializing in wine if the price of wine fluctuated. Just a simple perturbation around this assumption (say, considering that the price of wine is random, and can experience Extremistan-style variations) makes one reach a conclusion the opposite of Ricardo’s. Mother Nature does not like overspecialization, as it limits evolution and weakens the animals.

  This also explains why I found current ideas on globalization (such as those promoted by the journalist Thomas Friedman) one step too naïve, and too dangerous for society—unless one takes into account side effects. Globalization might give the appearance of efficiency, but the operating leverage and the degrees of interaction between parts will cause small cracks in one spot to percolate through the entire system. The result would be like a brain experiencing an epileptic seizure from too many cells firing at the same time. Consider that our brain, a well-functioning complex system, is not “globalized,” or, at least, not naïvely “globalized.”