Definition
A fat-tailed distribution is one whose extreme values — the tails — are much thicker than those of a normal bell curve. Rare, large events happen far more often than the familiar bell-shaped intuition suggests, and a single such event can dwarf everything else combined.
Many human-shaped domains behave this way: financial markets, wealth, city sizes, book sales, the impact of pandemics. In these domains the average is a poor summary, because a handful of outliers dominate the total. In a fat-tailed world, the tail is not the exception — it is the story.
Why it matters
How it works
When you suspect a fat-tailed domain, stop reasoning from the average and start reasoning from the extremes. Ask how bad the worst plausible case is and whether you could survive it. The right strategy emphasizes robustness — limiting downside exposure — over fine-tuning the everyday outcome.
The danger is applying bell-curve thinking, built for thin-tailed phenomena like height, to thick-tailed phenomena like markets. The two demand different mental models.