The Collapse of the Bell Curve and the Age of Power Laws
Why winner-take-all distributions keep appearing across industries — the logic of power laws and what it means for strategy in an attention economy.
In statistics, a normal distribution — the bell curve — describes a world where most outcomes cluster near the average, with extremes tapering off symmetrically on both sides. Height follows a normal distribution. Test scores approximate it.
A power law distribution looks different: a tiny number of outcomes capture most of the results, with a very long tail of small outcomes. Earthquakes follow power laws — most are barely detectable, but occasionally one is catastrophic. Wealth distribution in most countries follows a power law. So does social media attention. And startup valuations.
Why Power Laws Keep Appearing
Power laws emerge wherever there are feedback loops — where early success creates conditions for more success.
A startup that raises its first million finds it easier to raise its second. A song that gets played once is more likely to get played again because it's already in circulation. A city that attracts one skilled worker creates the conditions that attract the next one.
Normal distributions describe worlds where outcomes are independent. Power laws describe worlds where outcomes compound.
The internet dramatically accelerated power law dynamics by:
- Reducing distribution costs to near zero (one network effect can reach everyone)
- Making quality comparisons instantaneous and global
- Removing the geographic constraints that previously forced people to use local alternatives even when global alternatives were better
What This Means
If you're operating in a domain with power law dynamics, average performance is increasingly irrelevant. The question is whether you can reach the zone where compounding takes over.
This has obvious implications for media (a few platforms capture most attention), for labor markets (a few skills command most of the premium), and for companies (most industries are tending toward oligopoly).
The uncomfortable implication: a lot of what we were told about the value of being "pretty good" at many things may be a legacy of the normal distribution era. In a power law world, "pretty good at many things" competes against "excellent at one thing" — and loses.
This isn't fully deterministic. Niches have power laws too. You can be the best in a small, well-defined domain and still compound. But you have to actually be the best in that domain, not merely above average.