7 ideas
A Black Swan is an event with three properties: it is an outlier lying outside regular expectations, it carries an extreme impact, and human nature makes us concoct explanations for it after the fact, rendering it explainable and predictable in hindsight. The combination of rarity, extreme consequence, and retrospective predictability is what makes these events both rare and structurally invisible before they occur.
Some domains (height, weight, calorie consumption) belong to Mediocristan, where no single observation can meaningfully change the aggregate and outcomes cluster around an average. Other domains (wealth, book sales, financial losses) belong to Extremistan, where a single observation can dominate the total — and applying Mediocristan's bell-curve tools to Extremistan is the root of catastrophic miscalculation.
We build theories from the winners and survivors we can see, while the failures that would falsify those theories are dead, silent, and absent from the dataset. Because the drowned worshippers cannot testify alongside the survivors who prayed, any inference drawn only from visible success systematically overstates the role of skill, strategy, or causation.
A turkey is fed every day for a thousand days, and each feeding statistically reinforces its confidence that humans act in its interest — until the day before Thanksgiving, when its belief is maximally high and catastrophically wrong. The point at which past data felt most reassuring was precisely the point of greatest vulnerability, showing how induction from a finite history can betray you exactly when you trust it most.
Instead of taking medium risk, which is exposed to model errors you can't see, put roughly 85–90% of resources in extremely safe holdings and the rest in highly speculative bets with open-ended upside. This caps your exposure to negative Black Swans while leaving you exposed to positive ones. The strategy doesn't depend on predicting any event, only on shaping your payoff to its consequences.
In fields dominated by rare, high-impact events, such as economics, finance, political forecasting, and security analysis, expert predictions do little better than naive baselines. The experts also fail to learn from their errors, because they explain misses away as exceptions. Their credentials and precise models produce confidence without accuracy, which is more dangerous than admitting ignorance.
People compress sequences of facts into causal stories because stories are easier to remember and feel like understanding. After a Black Swan, we build an explanation that makes it look predictable in hindsight. That retrospective coherence hides how random the event was and makes us overconfident that we could foresee the next one.