Mediocristan vs. Extremistan
Two provinces of randomness. Mistaking one for the other is the root error.
Taleb's framing. In Mediocristan, no single observation moves the total. Heights, weights, exam scores. If you sample 1,000 people, the tallest still doesn't matter much for the total. The bell curve works. Averages are meaningful.
In Extremistan, one observation can dominate. Wealth, market returns, book sales, war casualties. The top 0.1% of any of these can be larger than the bottom 99.9% combined. Averages mislead. The variance is often infinite.
The root error is applying Mediocristan tools (bell curves, averages, standard deviations) to Extremistan phenomena. Most of the worst forecasting failures in history come from this exact mistake.
Examples in the wild
Revenue per customer in B2B SaaS is usually Extremistan. The top 5% of customers often produce 50% of revenue. Modelling them with averages produces wildly wrong forecasts.
Most investment returns over a lifetime come from a handful of decisions. Trying to be 'average' on each one misses the entire point of Extremistan.
Friendships are roughly Mediocristan. Romantic partners are roughly Extremistan. One bad partner can ruin a decade; few friendships have that asymmetry.
Mediocristan vs. Extremistan is one of the mental models we apply through real cases inside the Pareto MBA — a part-time program for professionals who want to think clearly about business.