Review: Skin in the Game by Nassim Nicholas Taleb

As with most of Taleb’s books, this one is poorly organized, full of childish insults and bravado, and makes some totally absurd claims. But this book also contains some startling, deep insights and ideas. It’s frustrating to have to wade through a lot of bullshit to get to these interesting ideas, but when you finally get there, the pay off is pretty damn good. Here are some of the biggest insights I got from this book:

1. The central thesis is that “skin in the game” is essential for systems to work efficiently and fairly.

Any actor who makes a decision must bear the risk of that decision. Otherwise, you end up with an asymmetry where one actor may get all the upside, and everyone else is stuck with all the potential downside.

Examples:

• No skin in the game: politician who argues for war, but has no risk of actually dying in that war.

• Skin in the game: activist, dissident, revolutionary.

• No skin in the game: banker who makes bad investments and is bailed out with tax dollars.

• Skin in the game: hedge fund traders. “Don’t tell me what you think; tell me what’s in your portfolio.”

• No skin in the game: bankers or financial advisors who tell others how to invest (e.g., on TV), but don’t invest that way themselves.

• Skin in the game: hedge fund traders.

• No skin in the game: consultants who dish out lots of advice, and then leave, so that only the customer feels the consequences of following that advice.
• Skin in the game: entrepreneurs, investors.

Creating a world where everyone has skin in whatever game they are playing may be more effective than adding more laws.

2. Systems work completely differently at different scales and you can’t generalize from one scale to another

You can’t always understand larger systems from smaller parts. Examples:

• You can’t understand the emergent behavior of an ant colony from the behavior of one ant.
• You can’t understand the behavior of the mind from the behavior of one neuron.
• You can’t understand the behavior of an entire country from the behavior of one person.
• “A country is not a large city, a city is not a large family, and, sorry, the world is not a large village.”
• A utopian socialist society, where everyone chips in for the good of the tribe, may be possible, but only below a certain scale; beyond that scale, groups of people behave completely differently, and you start to run into issues (e.g., tragedy of the commons) that cause socialism to fail.

Due to the curse of dimensionality, small increases in size can lead to massive increases in complexity. For example, we still don’t fully understand how the 300 neurons of the tape worm’s central nervous system work—it’s just too complicated. But going up to 301 neurons may double that complexity; going to 302 neurons may double it again; and so on. The human brain, for comparison, has 100 billion neurons.

While our TV-brand of politics pretends like it’s all about “left” vs “right”, Taleb has this amazing quote from the brothers Geoff and Vince Graham:

“I am, at the Fed level, libertarian;
at the state level, Republican;
at the local level, Democrat;
and at the family and friends level, a socialist.”

3. The Lindy effect: for some things, mortality rate decreases with time.

Broadway actors used to gather at Lindy’s delicatessen after each performance, and they noticed a pattern where plays that had been running for 20 days were likely to survive another 20 days; plays that had ran for 100 days were likely to survive another 100 days; those that ran for 200 days would run another 200; and so on. The same effect can be seen in many other places: e.g., a book that has been in print for 500 years is likely to be in print another 500 years. In other words, the life expectancy of some things is proportional to their current age. The longer that thing has survived, the longer it’s likely to keep surviving, so the mortality rate decreases with time.

As a general rule, this is why you should prefer things that have survived longer to those that are newer: e.g., a book that has been in print for hundreds of years is likely to be better than a book that came out last week. That old book has survived so long for a reason! The same goes for ideas, art, and traditions that have been passed down from generation to generation. “90% of what your grandma says is true.”

4. Rule of the minority

In many situations, a tiny minority (e.g., 1-3% of the population) can impose its will on a far larger majority. Examples:

• A large percentage of food in the US is kosher, even though only a tiny percentage of the population follows kosher practices
• Schools and planes ban peanuts even though only a tiny percentage of the population is allergic to them

This arises due to an asymmetry where the minority group can only tolerate one option but the majority group can tolerate either, and so we often accede to the demands of the minority group: e.g., people who are allergic to peanuts can’t attend an event at all if there are peanuts there, whereas everyone else can attend events either with or without peanuts, so even though the allergic group is ~1% of the population, we still typically go with their preferences.

This is why “Merry Christmas” became “Happy Holidays”, even in a country where the vast majority of people celebrate Christmas. This is also why all it takes is a single angry old lady complaining regularly to change the laws in a small town. More generally, many great changes in society happen not via consensus or voting, but because some small passionate minority (typically with skin in the game) pushes for it.

This is also why, even in a society that preaches tolerance, you cannot be tolerant of intolerance. That is, even if you firmly believe in tolerance of all viewpoints and free speech, you cannot be tolerant of, for example, Nazi movements. Even a small minority of intolerant people can overrun a tolerant society.

5. Ergodicity

To understand ergodicity, you must first understand the difference between time averaging and ensemble averaging. Time averaging is taking the average value of one process running over a long period of time; ensemble averaging is taking the average of many copies of the process (representing all the different states of that process) running for a fixed amount of time. Example: if the process is rolling two 6-side dice, then time averaging is where a single person rolls the dice thousands of times, and you take the average value of all of those rolls, and ensemble averaging is where you look at all the possible states from one roll (1-1, 1-2, 1-3, … 6-1, 6-2, 6-3…) and take the average of those.

A system is ergodic if the time average equals the ensemble average. The example with the two dice is ergodic, because the average value of all the possible states will equal the average value of one person rolling the dice over and over on a long enough timeline. However, not all systems are ergodic! Example from this blog post:

• Play a game where you flip a coin, and if you get heads, you win 50% of your bet, and if you get tails, you lose 40% of your bet.
• Ensemble average: If you were to compute the expected value of all possible states of the system (e.g., heads-heads-tails, tails-heads-heads-tails, etc), you’d expect to make roughly +5%.
• Time average: If you actually play this game as one person, on a long enough time line, you end up losing all of your money!
• Example playthrough: you bet $100, get heads, and end up with $150; you bet the $150, get heads again, and end up with $225; on the next flip, you get tails, and end up at $135; the next flip ends up tails too, and now you’re at $81. You got heads twice and tails twice, but have less money than you started with!
• The ensemble average does not equal the time average, so this system is not ergodic. This can lead to counter-intuitive results: “When many people play the game a fixed number of times, the average return is positive, but when a fixed number of people play the game many times, they should expect to lose most of their money.”

Taleb looks at ergodicity in society. If a society has perfect ergodicity, then, over a long enough timeline, everyone would spend some amount of time in each of the lower, middle, and upper classes (i.e., roughly 1% of your life would be spent in the top 1%, 50% of your life in the top 50%, etc); if a society has no ergodicity, then whatever class you start in, is where you stay—there’s no social mobility.

In order for social mobility to be possible, it’s not enough for the lower classes to be able to move up. It also needs to be possible for the upper classes to move down. How? By ensuring they have skin in the game! The upper class must be exposed to risk and downside; for if there’s no downside for them, that means there’s no upside for the everyone else.

“Consider that about 10 percent of Americans will spend at least a year in the top 1 percent, and more than half of all Americans will spent a year in the top 10 percent. This is visibly not the same for the more static—but nominally more equal—Europe. For instance, only 10 percent of the wealthiest five hundred American people or dynasties were so thirty years ago; more than 60 percent on the French list are heirs and a third of the richest Europeans were the richest centuries ago. In Florence, it was just revealed that things are even worse: the same handful of families have kept the wealth for five centuries.”

6. Risk and ruin

In systems that contain a risk of total ruin—e.g., a risk of losing all your money, or even more severely, a risk of losing your life—standard cost/benefit analysis doesn’t work. This is especially true on repeated playthroughs and in systems that are not ergodic. In such system, risks from seemingly independent events add up! For example, you can’t use normal “expected value” calculations on Russian roulette. The ensemble average across all possible states of multiple playthroughs of Russian roulette is positive; but the time average for a single person playing the game repeatedly is almost always death.

We often don’t properly take into account these ideas of ruin, ergodicity, and risks adding up over repeated events. Taleb has some great examples of this:

“All risks are not equal. We often hear that “Ebola is causing fewer deaths than people drowning in their bathtubs,” or something of the sort, based on “evidence.” This is another class of problems that your grandmother can get, but the semi-educated cannot. […] The probability that the number of people who drown in their bathtubs in the United States doubles next year—assuming no changes in population or bathtubs—is one per several trillions lifetimes of the universe. This cannot be said about the doubling of the number of people killed by terrorism over the same period.

[…]

So we often see the headline that many more American citizens slept with Kim Kardashian than died of Ebola. Or that more people were killed by their own furniture than by terrorism. Your grandmother’s logic would debunk these claims. Just consider that: it is impossible for a billion people to sleep with Kim Kardashian (even her), but that there is a non-zero probability that a multiplicative process (a pandemic) causes such a number of Ebola deaths. Or even if such events were not multiplicative, say, terrorism, there is a probability of actions such as polluting the water supply that can cause extreme deviations. The other argument is one of feedback: if terrorism casualties are low, it is because of vigilance (we tend to search passengers before boarding planes), and the argument that such vigilance is superfluous indicates a severe flaw in reasoning. Your bathtub is not trying to kill you.”

7. Random insights

The book meanders quite a bit, so I had a few fun insights in my notes that seem largely disconnected from everything else:

• Does a merchant have the moral obligation to to divulge all relevant information to a buyer? Example: the merchant is bringing grain from Egypt to a city experiencing famine. The merchant knows he can get a high price for his grain now, but he also knows more boats with grain are on the way, so the price will drop. Does he have the moral obligation to tell the buyer that those other boats are on the way?

• All other factors being equal, if you’re picking, say, a doctor, you should typically prefer the doctor who does not look the part—i.e., does not look like your stereotype of a doctor (e.g., older man with gray hair, refined appearance, silver-rimmed glasses, delicate hands). That’s because the doctor who doesn’t fit the stereotype, to build a successful career, had to get there through ability and accomplishment, without being able to lean on looks. The same is true of restaurants (the hole-in-the-wall on a side street can only survive by having amazing food, rather than due to location/appearance), leaders (have you noticed that all CEOs always look the same, but entrepreneurs always look different?), and books (don’t judge a book by its cover).

• History is often told by focusing on major, dramatic events, such as wars; the absence of such events is typically ignored by historians, which leads to an incomplete picture of what life was like. “Reading a history book, without putting its events in perspective, offers a similar bias to reading an account of life in New York seen from an emergency room at Bellevue Hospital.”

8. Random bullshit

This book also contained a lot of childish, absurd, bullshit. Some of it was clearly Taleb having a personal feud with a critic or rival. Some of it was him making claims about things where he is not an expert or where he has no skin in the game: e.g., his praise of Putin and Trump, his comparisons of different religions and atheism. It takes an order of magnitude more energy to refute bullshit than to spout it, so I won’t waste time arguing against Taleb’s more absurd claims, and merely note down here that I don’t agree with many of them.

Rating: 4 stars