Cherry-picking meets ignorance of human nature meets naive interpretation of history meets erroneous assumptions

By Alex Lynch  on February 24, 2018

[Unlike those who complain about Taleb’s unresolved teenage angst, his thin-skinned hubris, or his lack of civility, I couldn’t care less about his crass remarks. My problem is with the ideas in this book, not its author, although I do question the intelligence of its author when his prose lapses into pseudoscientific drivel.]

If you cherry-pick the data, you can make ANY ridiculous hypothesis sound convincing.

Most of the ideas in Nassim Nicholas Taleb’s ‘Skin in the Game’ are characterized by a shameless lack of nuance, are supported only by dishonest misrepresentation and overgeneralization of samples, and will probably make the world much worse if implemented. The only other book I can think of which more strongly exemplifies confirmation bias, and which is more blind towards overwhelming contradictory evidence, is Rhonda Byrne’s ‘The Secret’, and parallels between these two books run deep—deep enough to call this ‘Taleb’s The Secret’. (Even Byrne’s book draws heavily upon ancient mythologies to make one absurd point after another). As much as I’d like to offer a page-by-page refutation of all the specious ideas in this book, Amazon has a 5000-word limit on reviews; I prefer to save the words for analyzing the central idea of this book (“skin in the game/SITG”) and a recurring idea (“the Lindy effect”). But if I am accusing the author of statistical subterfuge, at least one example is in order.

p.11: “The idea of skin in the game is woven into history: historically, all warlords and warmongers were warriors themselves, and, with a few curious exceptions, societies were run by risk takers, not transferors.” This is followed by a paragraph in which Taleb melodramatically glorifies the bravery of a few emperors who died on the battlefield while fighting shoulder-to-shoulder with their soldiers or taking “frontline position in battle,” and then he continues: “These are not isolated anecdotes. The statistical reasoner in this author is quite convinced: less than a third of Roman emperors died in their beds—and one can argue that given that only few of these died of really old age, had they lived longer, they would have fallen either to a coup or in battle.”

Ironically, Taleb here is employing the very same journalistic trickery that he claims to loathe. This trick is disparagingly called “weasel words” by Wikipedia editors, and it’s the art of making vague but misleading statements that can be denied later if questioned. Taleb first prefixes the argument with “these are not isolated anecdotes” and then classifies “less than one-third of Roman emperors who died in their beds” into one category. This gives the reader a strong but utterly false impression that roughly two-thirds of them died in battle. (Analyzed below; it’s not even close this number.) But to maintain deniability about ever stating that two-thirds of them died in battle, he very subtly slips in that word at the end: coup. (“I never said that” is a usual Talebian denial after being exposed for implying exactly “that”). The reality, of course, is much less awe-inspiring/sensational and much more nuanced. What he is too dishonest to tell you is that in his data on the deaths of Roman emperors, he is actually COMBINING the minority of deaths in battle AND the majority of deaths from covert assassinations, suicides, and post-coup executions into a single unstated category to give it the overall appearance of brave warriors with SITG.

Now, we can all agree that emperors who, while they were emperors, fought alongside their soldiers on the battlefield were indeed ballsy and had a lot of SITG. But in other forms of unnatural death, does merely dying from coups and assassinations while going about the business of daily life possess battlefield-style SITG? It does not.

Usurpers who engage in foul-play assassinations, conspiracies, and coup attempts are always a threat to ANYBODY in power—Roman emperors, modern dictators, presidents—and if they are rarer now than in ancient history, it’s because our security systems are a lot more sophisticated at detecting and eliminating such threats. Merely being the target of an assassination/coup is not, by any stretch of the definition, battlefield-style SITG. And nor are post-failure suicides. Even Hitler committed suicide after losing WW2 without fighting in a single battle of this war.

So to analyze this further, and to keep the analysis logically consistent with the contents of p.11, I’ll first define ‘an emperor dying in battle’ as either 1) dying while fighting shoulder-to-shoulder alongside the emperor’s own soldiers on the battlefield, while executing his own plans, or 2) being captured while fighting in this manner and dying in captivity.

The moment you separate other forms of death from dying on the battlefield as defined above, the data ceases to be impressive. Following are the data on the deaths of Roman emperors. Verify, if you must, by looking up “list of Roman emperors”.

1. Julio-Claudian dynasty (27BC – 68AD) – 5 emperors; 0 deaths in battle; 1 natural death; 1 cause unknown; 1 suicide; 2 assassinations

2. Year of the Four Emperors and Flavian dynasty (68–96) – 6 emperors; 1 death in battle; 2 natural; 1 suicide; 2 assassinations

3. Nerva–Antonine dynasty (96–192) – 7 emperors; 0 deaths in battle; 6 natural; 1 assassination

4. Year of the Five Emperors and Severan dynasty (193–235) – 8 emperors; 1 death in battle; 1 natural; 4 assassinations; 2 executions

5. Crisis of the Third Century (235–285) – 22 emperors; 6 deaths in battle; 3 natural; 2 suicides; 11 assassinations

6. Tetrarchy and Constantinian dynasty (284–364) – 15 emperors; 3 deaths in battle; 6 natural; 2 suicides; 2 assassinations; 2 executions

7. Valentinian dynasty (364–392) – 4 emperors; 1 death in battle; 1 natural; 1 possibly suicide; 1 assassination

8. Theodosian dynasty (392–455) – 7 emperors; 2 deaths in battle; 3 natural; 1 assassination; 1 executions

9. Last emperors of the Western Empire (455–476) – 10 emperors; 0 deaths in battle; 3 natural; 3 assassinations; 2 executions; 2 unknown causes

All that adds up to:

Battle-related deaths: 14

Natural deaths: 26

Suicides: 7

Assassinations: 27

Executions after coup: 7

Unknown causes: 3

Total: 84

So there you have it. Only 17% of all the deaths of Roman emperors took place while fighting alongside their soldiers on a battlefield. Taleb takes 14 battlefield deaths, clubs them with 44 other unnatural deaths, and RUNS with it to make grandiose proclamations about SITG and sweeping generalizations about how it was too common back then.

Secondly, there’s ANOTHER weasel word in that quote. Without specifying what “really old age” is, Taleb says of Roman emperors who died naturally that “only few of these died of really old age.” This obviously leads the reader to assume that most emperors of this group died naturally at a “young” age and would’ve bravely fought to the death otherwise, but quite the opposite is true. 14 of the 26 emperors in this group lived past the age of 60 (sixty!), with the average for these 14 being 67 years—and this was almost 1700 years ago! Of course, the advantage that weasel words afford is that one can always retort, “by ‘really old age’, I meant [insert new favorite number here].” Every time he hides behind the veil of mendacious ambiguity, beware: he might be grossly distorting words to cover up a disconfirming fact. Twisting facts to fit his pet narrative is more prominent in this book than building an objective narrative around the entire set of facts.

Finally, and most importantly, there’s a good reason why high-ranking commanders of modern armies, the “decision-makers,” are precluded from joining their armies on the battlefield regardless of how good their plans are: Strong military acumen is an extremely rare but highly valuable commodity. Those who possess it are to be protected so that their wisdom can be applied to future missions. But even when their battle plans are carefully thought-out and diligently maneuvered, there WILL be unavoidable causalities. In executing these plans, if these commanders are out there on the battlefield with their men, they could potentially take a hit and all their acumen would die with them. The real loss, then, is to the future missions that would be left without the guidance of that fallen commander. Both Gen. Albert Sidney Johnston and Gen. Stonewall Jackson of the Confederate Army used to personally lead their men on the battlefield, and both died in two separate friendly fires. The deaths of these two generals are considered by military historians to be two of the twelve turning points in the American Civil War. Learning from such mistakes, high-ranking commanders of the Allied troops issued orders from highly secure locations, and won WW2. Ultimately, winning the war is more important than showing imbecilic heroism; the fate of the people, not just the soldiers, lies in the outcome of the war.

So in summary, the fact that SOME historical commanders had skin in the battle was actually a reckless error that has been fixed by modern armies. But that is exactly the kind of backwardness Taleb fantasizes about returning the world to.

And yes, there’s THAT much BS to unpack in just one page from the book, and we are still at p.11. Unfortunately, the rest of the book only gets worse as erroneous arguments rise both in number and in scale.

Anyway, to get on with the meat of the book: SITG. (Taleb all but begs the reader to take note of his SITG chivalry. Yes, good Sir Knight, your chivalry is noted.)

SITG isn’t just a reward-punishment model; punishment is what a centralized justice system does AFTER screwing up. Rather, it’s a decentralized, reward-or-punishment-through-risk-exposure model where your exposure to the consequences is ensured BEFORE the implementation, so that screwing up automatically punishes on its own. It’s (supposedly) a self-corrective model.

Now, there are not two but four combinations of idea-consequence scenarios that can be neatly represented as below.

The premise: You present an idea to the world, which is then implemented. In all four scenarios listed below, other people are respectively affected as a result of the implementation, but the ramifications for you are different in each.

1) Symmetry: You gain something valuable (to you) if it works, and you lose something valuable if it doesn’t.

2) Positive asymmetry: You gain something valuable if it works, but you lose nothing if it doesn’t.

3) Negative asymmetry: You gain nothing if it works, but you lose something valuable if it doesn’t.

4) Neutral: You gain nothing if it works, and you lose nothing if it doesn’t.

(1 and 3 are SITG scenarios; 2 and 4, not)

The book is rather disingenuous in its front-cover illustration and subtitle, which make it seem that this book is somehow a crusade against positive asymmetries—the “heads I win, tails you lose” bets. I would have showered this book with so much praise as to exhaust the nation’s supply of accolades if this book REALLY were about replacing only asymmetries with symmetries.

But since nuance isn’t Taleb’s forte, he goes all the way to the other extreme and says that EVERY idea-consequence situation must be symmetrical. (Along with numerous instances throughout the book, he ends the book by suggesting “[do] nothing without skin in the game.”) In a nutshell, Taleb argues that SITG eliminates bad ideas by disfiguring both the reputation and the bank accounts of those who concocted the ideas. An investment advisor who is investing your money with his ideas should have a significant personal stake in the same fund. If the idea fails, he almost drowns in bankruptcy and nobody will ever take his investment advice seriously again. Over time, many similar events will eliminate other bad ideas and the people who parented those ideas. As a result, the system overall is better off, and it is precisely SITG that allowed these self-corrections to happen. In a non-SITG environment, such people can persist.

Sounds great, and symmetries are indeed well suited to some situations. But the problem is that this solution is not at all generalizable and is very restricted in its applicability. Recall that there are two kinds of non-SITG scenarios, and if applied to the wrong one, Taleb’s model harms the system more than it rehabilitates it.

Which brings us to the dark side of SITG.

Many decades ago, Stanley Kubrick, the acclaimed filmmaker, pronounced his verdict on human nature in this eloquent quote: “We are capable of the greatest good and the greatest evil, but the problem is that we often can't distinguish between them when it suits our purpose.”

Paraphrased to befit the context of this review, the above quote simply says that if a man has his SITG, he will do just about ANYTHING to save his skin. He will lie, cheat, deceive, exaggerate, lobby, wield power, or do a million other wicked things just to save his skin.

Here are some ways in which SITG, by incapacitating the ability of the skin-owners to tell the difference between good and evil, can harm the system:

1. Taleb maintains that SITG and conflict of interest should not be conflated, but he fails to grasp that if, as he demands, politicians were to have their SITG, it would INEVITABLY lead to conflict of interest as a nasty side-effect. The reason why the powers-that-be, economic advisers to the president, and top-level bureaucrats are required NOT to have any SITG is because it’s a textbook example of conflict of interest—they could use the power of their office to recommend or implement only those policies which save their own skin, while the benefits for others might not be as, or at all, profitable. Carl Icahn, who is currently under federal investigation, briefly served as Adviser to the President and attempted to use the power of his office to save himself $200 million in taxes through a biofuel company that he owned. (He was allowed to have SITG because of bureaucratic loopholes; normally, this is rightly prohibited). If a man has SITG AND the government-given power to save his skin, he will do ANYTHING to save his own skin. [Additional checks, which currently do not exist, must be in place to ensure that even a conflict-of-interest-free public servant doesn’t directly profit from the policies they implemented, AFTER they leave the office]. However, someone who has the official power but who has nothing to gain or lose (as in the case of pure neutrals), either in the present or in the future, is more likely to do good to others rather than serve himself like Icahn did.

2. Financial SITG is the reason why tobacco companies, despite their own research showing that smoking tobacco is strongly correlated with lung cancer, suppressed those findings, lied to the public for decades that there is no evidence, let millions die of preventable cancer, got caught lying, and were sued for billions—all in a misguided attempt to save their invested skin. And unsurprisingly, owing to SITG, something very similar is happening with oil companies now. (Read Oreskes and Conway’s ‘Merchants of Doubt’ for more on this). Along with the rest of mankind, these people and their descendants will also be exposed to the downside risk of carbon emission—so there’s both financial AND literal SITG, but that doesn’t stop them from defrauding the public. All these companies lose a lot of money should things not go in their favor, and make a lot of money otherwise, so they are never honest about their data or their true intentions—a typical trait of those with SITG. Taleb himself stood to make a lot more money in 2007 had all the Big Banks been allowed to fail; he had placed bets that they would fail. Only the truly gullible can fail to see why he (fruitlessly) demanded that the Fed let those banks fail.

3. NOT having any SITG game lets you think objectively about a situation in a way that having your skin at stake hardly can. The slave-holding states of the American antebellum South wanted to secede from the Union primarily, though not solely (I am not nuance-averse), because of the issue of slavery. The abolitionists of the North had no skin in the cotton production game; only the southern cotton plantation/industry owners did, and cotton was the prime mover of the Southern economy. Slavery was crucial to the cotton business, and the slave-holding states of the South would have taken a huge economic hit if slavery were abolished. Small wonder, then, that the South wanted to keep slavery alive by seceding from the Union, thus initiating the Civil War. There was nothing inherently evil or stupid about the Southerners; they were driven by an inability to tell the difference between good and evil because their own interests were involved. Slavery did not resolve itself at the hands of those with skin in the cotton game. It took the intelligence and objectivity of non-slaveholders—the abolitionists of the North who, if the logic of this book were applied, would be labeled “virtue signalers”—to rid the US of slavery and better the system. It was Lincoln and his cohorts, not slaveholders or Southerners, who ratified the Thirteenth Amendment to the Constitution, the one abolishing slavery.

The best way to tell the difference between good and evil is to take your skin OUT of the game and then decide.

But the point is, both SITG and no SITG have their place; it’s a matter of what works better given the context. Sadly, such nuanced analysis is unapologetically foreign to Taleb’s simplistic methods. When Abraham Maslow noted that “to a man with only a hammer, every problem looks like a nail,” he must’ve foreseen Taleb violently conking at everything with his SITG sledgehammer.

Coming to the second recurring theme in the book, the Lindy effect: Here, Taleb’s loose grasp of reality takes on a life of its own. This idea is mathematically beautiful but ultimately stands on quick soil. (Taleb generally likes to point out that you cannot question the mathematics of his published papers without having your own sanity questioned, but the problem isn’t with the mathematics; it’s with the assumptions that get you started. Kurt Gödel once “proved” that God exists using mathematical logic; see Gödel's ontological proof. What’s not above criticism, though, are his assumptions—nor anybody else’s!) In essence, it states that the projected lifespan of non-perishable cultural entities is in direct correlation with its current age. If a book has survived for 100 years in print, it will likely survive another 100.

If you are familiar with ‘the Wisdom of the Crowds,’ it only takes a small leap of imagination to arrive at the Lindy effect: Lindy is nothing but the Wisdom of the Crowds applied across time. At its foundation, both ideas require people—lots of common, hardworking people—who make collective decisions about accepting or rejecting an idea through small decisions that accrue. In the Wisdom, the decisions accrue across space; in Lindy, across time. But in both, it is the hoi polloi—and not the academics, the bureaucrats, or some other group of chosen experts—who truly put the ideas to the test.

Studying the Wisdom sheds light on the nature of Lindy, and to that end I’ll quote an insightful excerpt from an essay by Warren Buffett that decries the Efficient Market Hypothesis, an absurd, absolutistic theory built on the Wisdom of the Crowds: “EMH was embraced not only by academics, but by many investment professionals and corporate managers as well. Observing correctly that the market was FREQUENTLY efficient, they went on to conclude incorrectly that it was ALWAYS efficient. The difference between these propositions is night and day.”

Lindy is indeed good at eliminating some bad cultural objects from the past. But since Taleb is fond of saying “Lindy and Lindy alone is the real expert,” I think Lindy’s consistency is worth examining. Is Lindy only FREQUENTLY or ALWAYS effective? The difference matters. A lot.

My first brush with the core assumption underlying this idea—though not the idea itself—was in Taleb’s ‘Fooled by Randomness,’ which I read back when I was an admirer of his. In that book he is careful to distinguish between survival through chance and survival through competence. A stockbroker can have a long career making successful bets, despite being clueless about stocks. The laws of stochastic probability make room for such anomalies. However, a dentist or a doctor can have a long career if and only if they are competent, and no law of probability will rescue them otherwise. It’s not really malpractice lawsuits or losing medical license that removes them from the profession, although that contributes, too; rather, it’s public verdict that nails their metaphorical coffin: You cannot fool people for long stretches of time in a profession where luck plays no role.

Or so I thought.

I learned this many unfortunate years later: The case he makes for non-stochastic professions turns out not to be true at all and illuminates a rot in the assumptions that Lindy stands on. Not only CAN incompetent doctors have decades-long career, but there actually IS a precedent for it.

The noise caused by the placebo effect can sometimes deafen people to the fraudulence of most alternative medicines which generally treat non-life-threatening conditions. But there is one particular case of a “doctor” in South Asia whose “cure” for the most intractable of human miseries—cancer—essentially makes it impossible to fail to tell the difference between success and failure of the medicine for long periods. If any alternative medicine fraud claims to have a cure for cancer, the claim can be put to the test as easily by the public as by scientists. People should, given a decade or more of hearsay, arrive at a verdict about the efficacy of the treatment—if Taleb is to be believed.

This “doctor” goes by the name of Vaidya Narayana Murthy who, along with his ancestors for centuries, has been “curing” all forms of cancer and other incurable ailments by making people ingest pieces of tree barks grown in his native village. He boasts of a success rate of 60%, clearly fabricated, since he would be awarded the Nobel Prize in Medicine if he could cure ANY cancer, let alone ALL forms of cancer, with that level of success. Every week, an exodus of benighted, gullible, illiterate, and even semi-literate people from all across the country arrive at his doorstep and stand in miles-long queue for hours to get a 10-second appointment with him. If Lindy were to be an “expert,” such oddities would necessarily have to be eliminated over time as people realize this man’s fraud and stop seeking his appointments, regardless of how much he charges for the appointments—but quite the opposite is happening, as his patient numbers rise every year. Even his online ratings are consistently high. Nothing about their behavior even remotely suggests that you can’t fool them for long, even at something so basic as the efficacy of a cure for cancer.

Vox populi? Vox humbug!

Many more such examples abound. When Lindy cannot even eliminate fraud in simple systems like detecting the success of a miraculous cure for cancer, to expect it to arrive at reliable heuristics in complex systems in the form of time-tested aphorisms is naïve wishful thinking. Aphorisms survive because of their rhetorical effect, not necessarily because they are agents of truth. Only by woefully cherry-picking them can you present them in a positive light. Superstitions survive for thousands of years, and horrible myths that are demonstrably untrue are inherited through generations of descendants, completely unfiltered by Lindy. (Conversely, many great books of science and math from the antiquity, including five books by Euclid, have been irretrievably lost, unprotected by Lindy.) In India, a practice called “sati”—in which a widowed wife would be cremated alive with her husband’s corpse—prevailed for more than 1500 years before it was forcefully abolished in 1821 through government intervention. Lindy tolerated it for 1500 years; bureaucrats and reformers ended it in just 15. If you wish to make hard life choices based on one-liners handed down from the social “wisdom” of the ancients, the Romans, or any other people who owned human slaves and committed atrocities for recreation, be my guest.

Page after page of this book is filled with vignettes from classical literature, to give it the feel of Lindyness. It never ceases to amuse me how Taleb combs through historical mythologies to find stories that vaguely metaphorically resemble an agenda he has already made up his mind about. (Even the typeface of this book is given a historical context for, geez!) Taleb likes to chastise psychologists, but psychologists have also committed the same error that Taleb is committing in abundance here: Drawing a little too much inspiration from ancient vignettes. Freud was inspired by the vignette of Oedipus when he came up with his ridiculous hypothesis of Oedipus complex. Jung produced an equally ridiculous variant called the Electra complex after the Greek mythological character. Another perverse complex, also inspired by classical Greek stories, goes by Jocasta complex.

Romanticizing the genius of the “elders” can produce idiotic filth, not always profundity. Even evolution by natural selection, which Taleb claims is a sophisticated form of SITG, is only a crude method of problem-solving that doesn’t eliminate all errors, no matter how much time passes. Cancer genes can survive in a species for millions of years.

However, none of this is to say that Lindy is totally useless. In the philosophy of science, consilience is a method of converging on the truth through multiple, independent sources of evidence that are themselves imperfect and prone to errors. We know that the theory of evolution is true not just because fossils hint at it, but because seven independent sources of evidence converge at the same conclusion. A theory which is supported only by one form of evidence is a lot weaker than a theory that is vindicated by multiple sources that do not depend on each other. In the event of a disagreement between sources—which is bound to happen given that each source is imperfect—all it means is that further investigation is needed, not that one source is necessarily better than the other, or that the other source must be discarded altogether. In consilience, Lindy can act as ONE of these independent sources, rather than replacing other sources. But by clownishly interjecting that “Lindy alone is the real expert,” Taleb only makes the cavernous depths of his ignorance official.

I have so much more to say, enough to fill a book, but I’ve reached the end of my 5000 words. I will make this last, somewhat amusing point, though. Murray Gell-Mann, Nobel Laureate physicist, is rightly praised in this book. Gell-Mann wrote an influential non-fiction book called The Quark and the Jaguar, which contains illuminating discussions on simple and complex systems.

Now, Taleb likes to mouth off that climate change is a non-linear, complex system and therefore we cannot, as climate scientists often do, predict the future trajectory of climate change with sufficient confidence. But here let me quote an excerpt from a 2011 interview with Gell-Mann precisely on this topic, which can be viewed on YouTube under the title ‘A chat on climate literacy with Murray Gell-Mann’. Gell-Mann’s answer is seething with righteous condescension, as he expresses these words with a look of incomprehension towards people who misguide the public on the predictability of climate change:

“Is it REALLY, REALLY so EXTREMELY difficult to persuade people that climate—which is not weather but average weather—can have three contributions that add to each other? That is, some cyclical effects, some random noise, and a secular, steadily rising trend from human activity. … This isn’t even science; it’s ordinary arithmetic. … Can people REALLY not grasp this trivially simple idea that you have the sum of these three terms, and if we WAIT until the secular term—the anthropogenic term—gets REALLY, REALLY big so that it DROWNS out the other two, it’s too late to do anything? … Isn’t it just an arithmetic problem, the idea that you are taking the sum of these three terms and that they behave differently? I mean, if you are used to graphs, you can appreciate it immediately. Is it THAT hard to introduce the public to the notion of a GRAPH? … The thing that I’m talking about is not the whole issue; just this trivial point about the sum of the three terms: 1) the random fluctuations, 2) the cyclical behavior—because the deniers talk about cycles, and they’re right, there are cycles and there are random terms—and 3) there is also the secular term. And I don’t understand why the people who are supposed to educate the public cannot get this across.”

Oh, dear Dr. Gell-Mann, the reason why there is so much confusion about the predictability of climate change is that “people who are supposed to educate the public” are actively engaged in misleading the public that, because climate change is a complex system, it cannot possibly be the arithmetic sum of three simple terms. That’s some circular irony right there: While Taleb is busy hijacking Gell-Mann’s work to sell his own agenda, Gell-Mann himself is out there trying to put out the fire started by charlatans who are spreading gross misapplications of his writings on complex systems.

If intellectuals can be idiots, Taleb is its most shining example. He is better suited for trolling on Twitter and peddling conspiracy theories about GMOs than for sermonizing on how societies should function.