CONCEPT
Fat Tails
The property of power-law distributions where extreme events contribute disproportionately to variance and expected value — rare outcomes in the tail dominate the distribution's character despite their rarity.
Fat tails describe the shape of probability distributions where extreme values occur far more frequently than Gaussian (thin-tailed) distributions predict. In a thin-tailed Gaussian, events beyond three standard deviations from the mean are vanishingly rare — six-sigma events are expected once every few million observations. In a fat-tailed power-law distribution, the probability decreases polynomially rather than exponentially, meaning extreme events, while still uncommon, are statistically significant. The consequence is counterintuitive: in fat-tailed distributions, rare extreme events contribute more to the total variance than common events near the median. For risk assessment, strategic planning, and policy design, this reverses the usual logic — you cannot safely ignore the tails because the tails contain the events that dominate outcomes.
In The You On AI Field Guide
Nassim Taleb popularized fat-tail awareness through his 'black swan' framework, but the mathematical foundation was established by Benoit Mandelbrot in the 1960s and given theoretical grounding by Per Bak's self-organized criticality in the 1980s. Mandelbrot demonstrated that cotton price fluctuations, rather than