Black SwanMental model

Extremistan

Экстремистан · Mediocristan vs Extremistan

Domains where a single observation can dominate the total - wealth, book sales, pandemic spread, social-media reach, war casualties.

Plain explanation

Add 1000 random people's net worth. Now add Jeff Bezos. The sample is now 99%+ about one person. Averages mean very little. Past data understates the future tail. Sample sizes lie.

Why it matters

Most modern wealth, fame, technology adoption, market crashes, viral events, political careers - all Extremistan. Treating them with Mediocristan tools is the single most common analytical error.

Practical example

«Average startup outcome» is meaningless - a few unicorns dwarf thousands of failures. «Median tweet engagement» tells you almost nothing about how a viral one will perform.

How to use

1If the domain is Extremistan, stop trying to predict averages. Plan for the tail.
2Limit exposure on the bad-tail side. Expose yourself to the good-tail side.

Related tools

Related concepts

Black Swan

MediocristanDomains where no single observation dramatically changes the total - physical attributes, repeating natural processes, well-bounded quantities.

Black Swan

Thick tail (and thin tail)In thick-tailed distributions, extreme events are more frequent and more extreme than the bell curve predicts. The tail does most of the work.

Black SwanAntifragility

ExposureThe shape of what happens to you under different outcomes - not the probability of those outcomes. Manage exposure, not forecast.

Black SwanAntifragility

RuinThe class of outcomes from which you cannot recover. Ruin doesn't average out, doesn't get reversed by a winning streak, and ends the game.

Black Swan

Black SwanA rare, high-impact event that is unpredictable beforehand and feels obvious afterwards.