The Black Swan: The Impact of the Highly Improbable

The summary

Before Australia was discovered, every swan anyone had seen was white — and one bird undid a belief built on millions of confirming sightings. Nassim Nicholas Taleb takes that as the shape of the events that actually run your life: rare, enormous in consequence, and declared obvious only after they land. A handful of them explain almost everything — the Great War, the collapse of the Soviet bloc, the spread of the Internet — while we spend our attention on the small repeatable things we can measure. The trouble isn’t that we fail to see them coming. It’s that we act as though they don’t exist.

Mediocristan and Extremistan

Line up a thousand random people in a stadium and add the heaviest human alive. He accounts for maybe 0.6 percent of the group’s total weight — a rounding error. That’s Mediocristan, where no single observation moves the aggregate: height, calorie intake, a dentist’s income. Now add the richest person on the planet, whose net worth Taleb puts near $80 billion: everyone else becomes a rounding error on his portfolio’s movement over the last second. For weight to do that, someone would have to weigh fifty million pounds. That’s Extremistan: wealth, book sales, city populations, market moves.

The dividing line is scalability. A dentist is paid by the hour and held down by gravity; a novelist writes the book once and sells it to one reader or three hundred million. In Mediocristan a hundred observations tell you most of what there is to know; in Extremistan your data will never be enough, because the observation that decides everything hasn’t shown up yet.

The turkey problem

Taleb’s version of a puzzle from Bertrand Russell: a turkey is fed every day by people who appear to have its interests at heart, and every feeding firms up the belief. On the Wednesday afternoon before Thanksgiving, the belief gets revised. The sharp part isn’t that the past misled the bird — it’s that its sense of safety peaked exactly when the risk was highest. Experience had negative value.

Generalize it to any situation where the same hand that feeds you can wring your neck. In the summer of 1982, large American banks lost roughly everything they had ever earned in the history of American banking, all at once, when the countries they had been lending to defaulted together. They had looked conservative right up to the end.

History hides the evidence

Psychologists once asked women to pick a favorite from twelve pairs of nylon stockings. The pairs were identical; the women confidently cited texture, feel, and color. Explanation isn’t something you do, it’s something that happens to you — and it works by compression. “The king died and the queen died” is two facts. “The king died, and then the queen died of grief” is one, lighter to carry and easier to pass on. Every satisfying story has thrown out the randomness, and the Black Swan is what got thrown out.

Then history deletes the evidence. Cicero tells of Diagoras, shown painted tablets of worshippers who prayed and survived a shipwreck; he asked where the pictures were of those who prayed and drowned. Studies of self-made millionaires find courage, optimism, and appetite for risk — so would a study of the graveyard of people who tried the same things and lost.

The scandal of prediction

Ask people for a range they are 98 percent confident contains some number. Two percent should fall outside it; when Albert and Raiffa first ran the test, on Harvard Business School students, 45 percent did. Taleb asked sixty Londoners to bracket the number of books in Umberto Eco’s library — thirty thousand — and not one range was wide enough. The problem isn’t how little you know, it’s the distance between that and how much you think you know. The Sydney Opera House was to open in early 1963 for AU$7 million; it opened a decade late, in a smaller form, for around AU$104 million.

Some experts really are experts: chess masters, physicists, test pilots. The list with no measurable edge is longer and better paid — stockbrokers, economists, clinical psychologists, risk experts. Things that move rarely have experts; things that sit still often do. Extra information only makes it worse: give bookmakers ten more variables per horse race and their accuracy stays flat while their confidence climbs. The casino Taleb visited spent hundreds of millions modeling gamblers, then took its four biggest hits from a tiger, a would-be bomber, a clerk hiding tax forms under his desk, and a kidnapping.

Be prepared instead of predictive

Since you can’t forecast these events, change your exposure. Decide by consequences, which you can picture, rather than probabilities, which you can’t compute. You don’t know the odds of an earthquake, but you know what one would do to San Francisco.

In practice that’s the barbell. Put 85 to 90 percent in the safest instruments you can find, Treasury bills, and the other 10 to 15 percent in extremely speculative bets. No broken risk model can reach your floor, and you keep the whole upside. Medium risk is the dangerous place, because “medium” was calculated by someone using a bell curve.

Then sort your surprises. Banking and catastrophe insurance carry only downside. Publishing, venture capital, and research have small losses and unbounded gains — there, not knowing is an asset, so make many small aggressive bets and stay in the path of accidents. Fleming found penicillin while cleaning his lab. Go to parties.

The bottom line

You will not see the big one coming, and neither will the person charging you for the forecast. Stop buying better predictions: cap the losses that could end you, stay exposed to the gains you can’t foresee, and treat any model that makes rare events look negligible as the one most likely to hurt you. Read it if you make decisions where the worst case matters more than the average.