Antifragile: Things That Gain from Disorder

by Taleb, Nassim Nicholas

  Let me express my idea in Lebanese dialect first. When you see a young and an old human, you can be confident that the younger will survive the elder. With something nonperishable, say a technology, that is not the case. We have two possibilities: either both are expected to have the same additional life expectancy (the case in which the probability distribution is called exponential), or the old is expected to have a longer expectancy than the young, in proportion to their relative age. In that situation, if the old is eighty and the young is ten, the elder is expected to live eight times as long as the younger one.

  Click here for a larger image of this table.

  Now conditional on something belonging to either category, I propose the following (building on the so-called Lindy effect in the version later developed by the great Benoît Mandelbrot):2

  For the perishable, every additional day in its life translates into a shorter additional life expectancy. For the nonperishable, every additional day may imply a longer life expectancy.

  So the longer a technology lives, the longer it can be expected to live. Let me illustrate the point (people have difficulty understanding it at the first go). Say I have for sole information about a gentleman that he is 40 years old and I want to predict how long he will live. I can look at actuarial tables and find his age-adjusted life expectancy as used by insurance companies. The table will predict that he has an extra 44 to go. Next year, when he turns 41 (or, equivalently, if applying the reasoning today to another person currently 41), he will have a little more than 43 years to go. So every year that elapses reduces his life expectancy by about a year (actually, a little less than a year, so if his life expectancy at birth is 80, his life expectancy at 80 will not be zero, but another decade or so).3

  The opposite applies to nonperishable items. I am simplifying numbers here for clarity. If a book has been in print for forty years, I can expect it to be in print for another forty years. But, and that is the main difference, if it survives another decade, then it will be expected to be in print another fifty years. This, simply, as a rule, tells you why things that have been around for a long time are not “aging” like persons, but “aging” in reverse. Every year that passes without extinction doubles the additional life expectancy.4 This is an indicator of some robustness. The robustness of an item is proportional to its life!

  The physicist Richard Gott applied what seems to be completely different reasoning to state that whatever we observe in a randomly selected way is likely to be neither in the beginning nor in the end of its life, most likely in its middle. His argument was criticized for being rather incomplete. But by testing his argument he tested the one I just outlined above, that the expected life of an item is proportional to its past life. Gott made a list of Broadway shows on a given day, May 17, 1993, and predicted that the longest-running ones would last longest, and vice versa. He was proven right with 95 percent accuracy. He had, as a child, visited both the Great Pyramid (fifty-seven hundred years old), and the Berlin Wall (twelve years old), and correctly guessed that the former would outlive the latter.

  The proportionality of life expectancy does not need to be tested explicitly—it is the direct result of “winner-take-all” effects in longevity.

  Two mistakes are commonly made when I present this idea—people have difficulties grasping probabilistic notions, particularly when they have spent too much time on the Internet (not that they need the Internet to be confused; we are naturally probability-challenged). The first mistake is usually in the form of the presentation of the counterexample of a technology that we currently see as inefficient and dying, like, say, telephone land lines, print newspapers, and cabinets containing paper receipts for tax purposes. These arguments come with anger as many neomaniacs get offended by my point. But my argument is not about every technology, but about life expectancy, which is simply a probabilistically derived average. If I know that a forty-year-old has terminal pancreatic cancer, I will no longer estimate his life expectancy using unconditional insurance tables; it would be a mistake to think that he has forty-four more years to live, like others in his age group who are cancer-free. Likewise someone (a technology guru) interpreted my idea as suggesting that the World Wide Web, being currently less than about twenty years old, will only have another twenty to go—this is a noisy estimator that should work on average, not in every case. But in general, the older the technology, not only the longer it is expected to last, but the more certainty I can attach to such a statement.5

  Remember the following principle: I am not saying that all technologies do not age, only that those technologies that were prone to aging are already dead.

  The second mistake is to believe that one would be acting “young” by adopting a “young” technology, revealing both a logical error and mental bias. It leads to the inversion of the power of generational contributions, producing the illusion of the contribution of the new generations over the old—statistically, the “young” do almost nothing. This mistake has been made by many people, but most recently I saw an angry “futuristic” consultant who accuses people who don’t jump into technology of “thinking old” (he is actually older than I am and, like most technomaniacs I know, looks sickly and pear-shaped and has an undefined transition between his jaw and his neck). I didn’t understand why one would be acting particularly “old” by loving things historical. So by loving the classics (“older”) I would be acting “older” than if I were interested in the “younger” medieval themes. This is a mistake similar to believing that one would turn into a cow by eating cow meat. It is actually a worse fallacy than the inference from eating: a technology, being informational rather than physical, does not age organically, like humans, at least not necessarily so. The wheel is not “old” in the sense of experiencing degeneracy.

  This idea of “young” and “old” attached to certain crowd behavior is even more dangerous. Supposedly, if those who don’t watch prepackaged 18-minute hyped-up lectures on the Web paid attention to people in their teens and twenties, who do, and in whom supposedly the key to the future lies, they would be thinking differently. Much progress comes from the young because of their relative freedom from the system and courage to take action that older people lose as they become trapped in life. But it is precisely the young who propose ideas that are fragile, not because they are young, but because most unseasoned ideas are fragile. And, of course, someone who sells “futuristic” ideas will not make a lot of money selling the value of the past! New technology is easier to hype up.

  I received an interesting letter from Paul Doolan from Zurich, who was wondering how we could teach children skills for the twenty-first century since we do not know which skills will be needed in the twenty-first century—he figured out an elegant application of the large problem that Karl Popper called the error of historicism. Effectively my answer would be to make them read the classics. The future is in the past. Actually there is an Arabic proverb to that effect: he who does not have a past has no future.6

  A FEW MENTAL BIASES

  Next I present an application of the fooled by randomness effect. Information has a nasty property: it hides failures. Many people have been drawn to, say, financial markets after hearing success stories of someone getting rich in the stock market and building a large mansion across the street—but since failures are buried and we don’t hear about them, investors are led to overestimate their chances of success. The same applies to the writing of novels: we do not see the wonderful novels that are now completely out of print, we just think that because the novels that have done well are well written (whatever that means), that what is well written will do well. So we confuse the necessary and the causal: because all surviving technologies have some obvious benefits, we are led to believe that all technologies offering obvious benefits will survive. I will leave the discussion of what impenetrable property may help survival to the section on Empedocles’ dog. But note here the mental bias that causes people to believe in the “power of” some techno
logy and its ability to run the world.

  Another mental bias causing the overhyping of technology comes from the fact that we notice change, not statics. The classic example, discovered by the psychologists Daniel Kahneman and Amos Tversky, applies to wealth. (The pair developed the idea that our brains like minimal effort and get trapped that way, and they pioneered a tradition of cataloging and mapping human biases with respect to perception of random outcomes and decision making under uncertainty). If you announce to someone “you lost $10,000,” he will be much more upset than if you tell him “your portfolio value, which was $785,000, is now $775,000.” Our brains have a predilection for shortcuts, and the variation is easier to notice (and store) than the entire record. It requires less memory storage. This psychological heuristic (often operating without our awareness), the error of variation in place of total, is quite pervasive, even with matters that are visual.

  We notice what varies and changes more than what plays a large role but doesn’t change. We rely more on water than on cell phones but because water does not change and cell phones do, we are prone to thinking that cell phones play a larger role than they do. Second, because the new generations are more aggressive with technology, we notice that they try more things, but we ignore that these implementations don’t usually stick. Most “innovations” are failures, just as most books are flops, which should not discourage anyone from trying.

  Neomania and Treadmill Effects

  You are driving on the highway in your two-year-old Japanese car when you are overtaken by a vehicle of the same make, the latest version, that looks markedly different. And markedly better. Markedly better? The bumper is slightly larger and the taillights are wider. Other than these cosmetic details (and perhaps some hidden technical improvements) representing less than a few percentage points in variation, the car looks the same, but you can’t tell by just looking at it. You just see the lights and feel that you are due an upgrade. And the upgrade will cost you, after you sell your car, about the third of the price of a new vehicle—all that motivated by small, mostly cosmetic variations. But switching cars is a small cost compared to switching computers—the recovery value of an old computer is so negligible.

  You use an Apple Mac computer. You just bought a new version a week before. The person on the plane next to you just pulled out of his bag an older version. It has a family resemblance to yours, but looks so inferior. It is thicker and has a much less elegant screen. But you forget the days when you used to have the same model and were thrilled with it.

  The same with a cell phone: you look down at those carrying older, larger models. But a few years ago you would have considered these small and slick.

  So with so many technologically driven and modernistic items—skis, cars, computers, computer programs—it seems that we notice differences between versions rather than commonalities. We even rapidly tire of what we have, continuously searching for versions 2.0 and similar iterations. And after that, another “improved” reincarnation. These impulses to buy new things that will eventually lose their novelty, particularly when compared to newer things, are called treadmill effects. As the reader can see, they arise from the same generator of biases as the one about the salience of variations mentioned in the section before: we notice differences and become dissatisfied with some items and some classes of goods. This treadmill effect has been investigated by Danny Kahneman and his peers when they studied the psychology of what they call hedonic states. People acquire a new item, feel more satisfied after an initial boost, then rapidly revert to their baseline of well-being. So, when you “upgrade,” you feel a boost of satisfaction with changes in technology. But then you get used to it and start hunting for the new new thing.

  But it looks as though we don’t incur the same treadmilling techno-dissatisfaction with classical art, older furniture—whatever we do not put in the category of the technological. You may have an oil painting and a flat-screen television set inhabiting the same room of your house. The oil painting is an imitation of a classic Flemish scene made close to a century ago, with the dark ominous skies of Flanders, majestic trees, and an uninspiring but calmative rural scene. I am quite certain that you are not eager to upgrade the oil painting but that soon your flat-screen TV set will be donated to the local chapter of some kidney foundation.

  The same with dishes—recall that we try to replicate nineteenth-century dinner customs. So there is at least one other domain in which we do not try to optimize matters.

  I am initially writing these lines longhand, using a seasoned fountain pen. I do not fuss over the state of my pens. Many of them are old enough to cross decades; one of them (the best) I have had for at least thirty years. Nor do I obsess over small variations in the paper. I prefer to use Clairefontaine paper and notebooks that have hardly changed since my early childhood—if anything, they have degraded in quality.

  But when it comes to transcribing my writing into electronic form, then I get worried that my Mac computer may not be the best tool for the job. I heard somewhere that the new version had a longer-lasting battery and I plan to upgrade soon, during my next impulse buying episode.

  Note here is a strange inconsistency in the way we perceive items across the technological and real domains. Whenever I sit on an airplane next to some businessman reading the usual trash businessmen read on an e-reader, said businessperson will not resist disparaging my use of the book by comparing the two items. Supposedly, an e-reader is more “efficient.” It delivers the essence of the book, which said businessman assumes is information, but in a more convenient way, as he can carry a library on his device and “optimize” his time between golf outings. I have never heard anyone address the large differences between e-readers and physical books, like smell, texture, dimension (books are in three dimensions), color, ability to change pages, physicality of an object compared to a computer screen, and hidden properties causing unexplained differences in enjoyment. The focus of the discussion will be commonalities (how close to a book this wonderful device is). Yet when he compares his version of an e-reader to another e-reader, he will invariably focus on minute differences. Just as when Lebanese run into Syrians, they focus on the tiny variations in their respective Levantine dialects, but when Lebanese run into Italians, they focus on similarities.

  There may be a heuristic that helps put such items in categories. First, the electronic on-off switch. Whatever has an “off” or “on” switch that I need to turn off before I get yelled at by the flight attendant will necessarily be in one category (but not the opposite as many items without an on-off switch will be prone to neomania). For these items, I focus on variations, with attendant neomania. But consider the difference between the artisanal—the other category—and the industrial. What is artisanal has the love of the maker infused in it, and tends to satisfy—we don’t have this nagging impression of incompleteness we encounter with electronics.

  It also so happens that whatever is technological happens to be fragile. Articles made by an artisan cause fewer treadmill effects. And they tend to have some antifragility—recall how my artisanal shoes take months before becoming comfortable. Items with an on-off switch tend to have no such redeeming antifragility.

  But alas, some things we wish were a bit more fragile—which brings us to architecture.

  ARCHITECTURE AND THE IRREVERSIBLE NEOMANIA

  There is some evolutionary warfare between architects producing a compounded form of neomania. The problem with modernistic—and functional—architecture is that it is not fragile enough to break physically, so these buildings stick out just to torture our consciousness—you cannot exercise your prophetic powers by leaning on their fragility.

  Urban planning, incidentally, demonstrates the central property of the so-called top-down effect: top-down is usually irreversible, so mistakes tend to stick, whereas bottom-up is gradual and incremental, with creation and destruction along the way, though presumably with a positive slope.

  Further, things tha
t grow in a natural way, whether cities or individual houses, have a fractal quality to them. Like everything alive, all organisms, like lungs, or trees, grow in some form of self-guided but tame randomness. What is fractal? Recall Mandelbrot’s insight in Chapter 3: “fractal” entails both jaggedness and a form of self-similarity in things (Mandelbrot preferred “self-affinity”), such as trees spreading into branches that look like small trees, and smaller and smaller branches that look like a slightly modified, but recognizable, version of the whole. These fractals induce a certain wealth of detail based on a small number of rules of repetition of nested patterns. The fractal require some jaggedness, but one that has some method to its madness. Everything in nature is fractal, jagged, and rich in detail, though with a certain pattern. The smooth, by comparison, belongs to the class of Euclidian geometry we study in school, simplified shapes that lose this layer of wealth.

  Alas, contemporary architecture is smooth, even when it tries to look whimsical. What is top-down is generally unwrinkled (that is, unfractal) and feels dead.

  Sometimes modernism can take a naturalistic turn, then stop in its tracks. Gaudi’s buildings in Barcelona, from around the turn of the twentieth century, are inspired by nature and rich architecture (Baroque and Moorish). I managed to visit a rent-controlled apartment there: it felt like an improved cavern with rich, jagged details. I was convinced that I had been there in a previous life. Wealth of details, ironically, leads to inner peace. Yet Gaudi’s idea went nowhere, except in promoting modernism in its unnatural and naive versions: later modernistic structures are smooth and completely stripped of fractal jaggedness.