Antifragile: Things That Gain from Disorder

by Taleb, Nassim Nicholas

  But Montaigne and Seneca were a bit too indulgent toward self-interest and missed something quite central. They clearly got the point that economic life does not necessarily depend on altruistic motives, and that the aggregate works differently from the individual. Remarkably, Seneca was born about eighteen centuries before Adam Smith, and Montaigne about three, so we should be quite impressed with their thinking while retaining a certain abhorrence of the fundamental dishonesty of men. We have known since Adam Smith that the collective does not require the benevolence of individuals, as self-interest can be the driver of growth. But all this does not make people less unreliable in their personal opinions about the collective. For they are involving the skin of others, so to speak.

  What Montaigne and Seneca missed, in addition to the notion of skin in the game, was that one can draw the line with public affairs. They missed the agency problem—although the problem was known heuristically (Hammurabi, golden rules), it was not part of their consciousness.

  The point isn’t that making a living in a profession is inherently bad; rather, it’s that such a person becomes automatically suspect when dealing with public affairs, matters that involve others. The definition of the free man, according to Aristotle, is one who is free with his opinions—as a side effect of being free with his time.

  Freedom in this sense is only a matter of sincerity in political opinions.

  The Greeks saw the world in three professions. The banausikai technai, the artisans; the craft of war, polemike techne; and that of farming, georgia. The last two professions, war and farming, were worthy of a gentleman—mainly because they were not self-serving and were free of conflicts of interest with the collective. But the Athenians despised the banausoi, the artisans who worked for a living in dark rooms making objects—generally sitting down. For Xenophon, such crafts degraded the craftsmen’s bodily strength, softened his spirit, and left him no time for his friends and city. The illiberal arts confine one to the workshop and narrow one’s interests to his own welfare; the crafts of war and farming give one a wider scope so that he can attend to his friends and city. To Xenophon, farming is the mother and nurse of the other technai. (The ancients did not have corporations; if Xenophon were alive today he would transfer his distrust from artisans to corporate employees.)

  There are Arabic and Hebrew sayings, Yad el hurr mizan / Yad ben horin moznayim—“the hand of the free is a scale.” It is just that the definition of the free is not well understood: he is free who owns his own opinion.

  For Metternich, humanity started at the rank of baron; for Aristotle, as well as, though in a separate form, the English up until the twentieth century, it started at the rank of idle freeman, unpreoccupied with work. It never meant not working; it just meant not deriving your personal and emotional identity from your work, and viewing work as something optional, more like a hobby. In a way your profession does not identify you so much as other attributes, here your birth (but it could be something else). This is the f*** you money that allowed Thales of Miletus to gauge his own sincerity. For the Spartans, it was all about courage. For Fat Tony, humanity started at the level of “self-ownership.”

  Now self-ownership for our horizontal friend was vastly more democratic than for his thinking predecessors. It simply meant being the owner of your opinion. And it has nothing to do with wealth, birth, intelligence, looks, shoe size, rather with personal courage.

  In other words, for Fat Tony, it was a very, very specific definition of a free person: someone who cannot be squeezed into doing something he would otherwise never do.

  Consider this leap in sophistication from Athens to Brooklyn: if for the Greeks, only he who is free with his time is free with his opinion, for our horizontal friend and advisor, only he who has courage is free with his opinion. Sissies are born, not made. They stay sissies no matter how much independence you give them, no matter how rich they get.

  Another facet of the difference between abstract modernistic nation-states and local government. In an antique city-state, or a modern municipality, shame is the penalty for the violation of ethics—making things more symmetric. Banishment and exile, or, worse, ostracism were severe penalties—people did not move around voluntarily and considered up-rooting a horrible calamity. In larger organisms like the mega holy nation-state, with a smaller role for face-to-face encounters, and social roots, shame ceases to fulfill its duty of disciplinarian. We need to reestablish it.

  And aside from shame, there is friendship, socialization in a certain milieu, being part of a group of people that have diverging interests from the collective. Cleon, the hero of the Peloponnesian War, advocated the public renouncement of friends upon taking up public affairs—he paid for it with some revilement by historians.

  A simple solution, but quite drastic: anyone who goes into public service should not be allowed to subsequently earn more from any commercial activity than the income of the highest paid civil servant. It is like a voluntary cap (it would prevent people from using public office as a credential-building temporary accommodation, then going to Wall Street to earn several million dollars). This would get priestly people into office.

  Just as Cleon was reviled, in the modern world, there seems to be an inverse agency problem for those who do the right thing: you pay for your service to the public with smear campaigns and harassment. The activist and advocate Ralph Nader suffered numerous smear campaigns as the auto industry went after him.

  THE ETHICAL AND THE LEGAL

  I felt ashamed not having exposed the following scam for a long time. (As I said, if you see fraud …) Let us call it the Alan Blinder problem.

  The story is as follows. At Davos, during a private coffee conversation that I thought aimed at saving the world from, among other things, moral hazard and agency problems, I was interrupted by Alan Blinder, a former vice chairman of the Federal Reserve Bank of the United States, who tried to sell me a peculiar investment product that aims at legally hoodwinking taxpayers. It allowed the high net worth investor to get around the regulations limiting deposit insurance (at the time, $100,000) and benefit from coverage for near-unlimited amounts. The investor would deposit funds in any amount and Prof. Blinder’s company would break it up into smaller accounts and invest in banks, thus escaping the limit; it would look like a single account but would be insured in full. In other words, it would allow the super-rich to scam taxpayers by getting free government-sponsored insurance. Yes, scam taxpayers. Legally. With the help of former civil servants who have an insider edge.

  I blurted out: “Isn’t this unethical?” I was then told in response “It is perfectly legal,” adding the even more incriminating “we have plenty of former regulators on the staff,” (a) implying that what was legal was ethical and (b) asserting that former regulators have an edge over citizens.

  It took a long time, a couple of years, before I reacted to the event and did my public J’accuse. Alan Blinder is certainly not the worst violator of my sense of ethics; he probably irritated me because of the prominence of his previous public position, while the Davos conversation was meant to save the world from evil (I was presenting to him my idea of how bankers take risks at the expense of taxpayers). But what we have here is a model of how people use public office to, at some point, legally profit from the public.

  Tell me if you understand the problem in its full simplicity: former regulators and public officials who were employed by the citizens to represent their best interests can use the expertise and contacts acquired on the job to benefit from glitches in the system upon joining private employment—law firms, etc.

  Think about it a bit further: the more complex the regulation, the more bureaucratic the network, the more a regulator who knows the loops and glitches would benefit from it later, as his regulator edge would be a convex function of his differential knowledge. This is a franchise, an asymmetry one has at the expense of others. (Note that this franchise is spread across the economy; the car company Toyota hired former U.S. regula
tors and used their “expertise” to handle investigations of its car defects.)

  Now stage two—things get worse. Blinder and the dean of Columbia University Business School wrote an op-ed opposing the government’s raising the insurance limit on individuals. The article argued that the public should not have the unlimited insurance that Blinder’s clients benefit from.

  A few remarks.

  First, the more complicated the regulation, the more prone to arbitrages by insiders. This is another argument in favor of heuristics. Twenty-three hundred pages of regulation—something I can replace with Hammurabi’s rule—will be a gold mine for former regulators. The incentive of a regulator is to have complex regulation. Again, the insiders are the enemies of the less-is-more rule.

  Second, the difference between the letter and the spirit of regulation is harder to detect in a complex system. The point is technical, but complex environments with nonlinearities are easier to game than linear ones with a small number of variables. The same applies to the gap between the legal and the ethical.

  Third, in African countries, government officials get explicit bribes. In the United States they have the implicit, never mentioned, promise to go work for a bank at a later date with a sinecure offering, say $5 million a year, if they are seen favorably by the industry. And the “regulations” of such activities are easily skirted.

  What upset me the most about the Alan Blinder problem is the reactions by those with whom I discussed it: people found it natural that a former official would try to “make money” thanks to his former position—at our expense. Don’t people like to make money? goes the argument.

  Casuistry as Optionality

  You can always find an argument or an ethical reason to defend an opinion ex post. This is a dicey point, but, as with cherry-picking, one should propose an ethical rule before an action, not after. You want to prevent fitting a narrative to what you are doing—and for a long time “casuistry,” the art of arguing the nuances of decisions, was just that, fitting narratives.

  Let me first define a fraudulent opinion. It is simply one with vested interests generalized to the public good—in which, say a hairdresser recommends haircuts “for the health of people,” or a gun lobbyist claims gun ownership is “good for America,” simply making statements that benefit him personally, while the statements are dressed up to look as if they were made for the benefit of the collective. In other words, is he in the left column of Table 7? Likewise, Alan Blinder wrote that he opposed generalized deposit insurance, not because his company would lose business, but because of the public good.

  But the heuristic is easy to implement, with a simple question. I was in Cyprus at a conference dinner in which another speaker, a Cypriot professor of petrochemical engineering in an American university, was ranting against the climate activist Lord Nicholas Stern. Stern was part of the conference but absent from the dinner. The Cypriot was extremely animated. I had no idea what the issues were, but saw the notion of “absence of evidence” mixed with “evidence of absence” and pounced on him in defense of Stern, whom I had never met. The petrochemical engineer was saying that we had no evidence that fossil fuels caused harm to the planet, turning his point semantically into something equivalent in decision making to the statement that that we had evidence that fossil fuels did not harm. He made the mistake of saying that Stern was recommending useless insurance, causing me to jump to ask him if he had car, health, and other insurance for events that did not take place, that sort of argument. I started bringing up the idea that we are doing something new to the planet, that the burden of evidence is on those who disturb natural systems, that Mother Nature knows more than he will ever know, not the other way around. But it was like talking to a defense lawyer—sophistry, and absence of convergence to truth.

  Then a heuristic came to mind. I surreptitiously asked a host sitting next to me if the fellow had anything to gain from his argument: it turned out that he was deep into oil companies, as an advisor, an investor, and a consultant. I immediately lost interest in what he had to say and the energy to debate him in front of others—his words were nugatory, just babble.

  Note how this fits into the idea of skin in the game. If someone has an opinion, like, say, the banking system is fragile and should collapse, I want him invested in it so he is harmed if the audience for his opinion are harmed—as a token that he is not an empty suit. But when general statements about the collective welfare are made, instead, absence of investment is what is required. Via negativa.

  I have just presented the mechanism of ethical optionality by which people fit their beliefs to actions rather than fit their actions to their beliefs. Table 8 compares professions with respect to such ethical backfitting.

  Click here for a larger image of this table.

  There exists an inverse Alan Blinder problem, called “evidence against one’s interest.” One should give more weight to witnesses and opinions when they present the opposite of a conflict of interest. A pharmacist or an executive of Big Pharma who advocates starvation and via negativa methods to cure diabetes would be more credible than another one who favors the ingestion of drugs.

  BIG DATA AND THE RESEARCHER’S OPTION

  This is a bit technical, so the reader can skip this section with no loss. But optionality is everywhere, and here is a place to discuss a version of cherry-picking that destroys the entire spirit of research and makes the abundance of data extremely harmful to knowledge. More data means more information, perhaps, but it also means more false information. We are discovering that fewer and fewer papers replicate—textbooks in, say, psychology need to be revised. As to economics, fuhgetaboudit. You can hardly trust many statistically oriented sciences—especially when the researcher is under pressure to publish for his career. Yet the claim will be “to advance knowledge.”

  Recall the notion of epiphenomenon as a distinction between real life and libraries. Someone looking at history from the vantage point of a library will necessarily find many more spurious relationships than one who sees matters in the making, in the usual sequences one observes in real life. He will be duped by more epiphenomena, one of which is the direct result of the excess of data as compared to real signals.

  We discussed the rise of noise in Chapter 7. Here it becomes a worse problem, because there is an optionality on the part of the researcher, no different from that of a banker. The researcher gets the upside, truth gets the downside. The researcher’s free option is in his ability to pick whatever statistics can confirm his belief—or show a good result—and ditch the rest. He has the option to stop once he has the right result. But beyond that, he can find statistical relationships—the spurious rises to the surface. There is a certain property of data: in large data sets, large deviations are vastly more attributable to noise (or variance) than to information (or signal).1

  FIGURE 18. The Tragedy of Big Data. The more variables, the more correlations that can show significance in the hands of a “skilled” researcher. Falsity grows faster than information; it is nonlinear (convex) with respect to data.

  There is a difference in medical research between (a) observational studies, in which the researcher looks at statistical relationships on his computer, and (b) the double-blind cohort experiments that extract information in a realistic way that mimics real life.

  The former, that is, observation from a computer, produces all manner of results that tend to be, as last computed by John Ioannides, now more than eight times out of ten, spurious—yet these observational studies get reported in the papers and in some scientific journals. Thankfully, these observational studies are not accepted by the Food and Drug Administration, as the agency’s scientists know better. The great Stan Young, an activist against spurious statistics, and I found a genetics-based study in The New England Journal of Medicine claiming significance from statistical data—while the results to us were no better than random. We wrote to the journal, to no avail.

  Figure 18 shows the swelling number
of potential spurious relationships. The idea is as follows. If I have a set of 200 random variables, completely unrelated to each other, then it would be near impossible not to find in it a high correlation of sorts, say 30 percent, but that is entirely spurious. There are techniques to control the cherry-picking (one of which is known as the Bonferoni adjustment), but even then they don’t catch the culprits—much as regulation doesn’t stop insiders from gaming the system. This explains why in the twelve years or so since we’ve decoded the human genome, not much of significance has been found. I am not saying that there is no information in the data: the problem is that the needle comes in a haystack.

  Even experiments can be marred with bias: the researcher has the incentive to select the experiment that corresponds to what he was looking for, hiding the failed attempts. He can also formulate a hypothesis after the results of the experiment—thus fitting the hypothesis to the experiment. The bias is smaller, though, than in the previous case.

  The fooled-by-data effect is accelerating. There is a nasty phenomenon called “Big Data” in which researchers have brought cherry-picking to an industrial level. Modernity provides too many variables (but too little data per variable), and the spurious relationships grow much, much faster than real information, as noise is convex and information is concave.