Power-law distribution

A power-law distribution is a probability distribution in which the probability of observing a value x is proportional to x⁻ᵅ, where α is a positive constant called the exponent. Unlike the normal distribution or exponential distributions, power laws do not have a characteristic scale: extreme events are not exponentially suppressed but follow a polynomial decay. This means that events many orders of magnitude larger than the mean are not just possible but expected with non-negligible probability.

Power laws appear across an extraordinary range of phenomena: the sizes of cities (Zipf's law), the frequencies of words in natural language, the intensities of earthquakes (Gutenberg-Richter law), the wealth of individuals (Pareto distribution), and the degrees of nodes in scale-free networks. Their ubiquity has led to both productive theory and premature claims — not every heavy-tailed distribution is a power law, and distinguishing true power laws from alternatives (log-normal, stretched exponential) requires careful statistical analysis.

From a systems perspective, power laws are the signature of positive feedback and self-reinforcing dynamics. Wherever a process amplifies existing advantages — wealth begetting wealth, citations begetting citations, links begetting links — the resulting distribution tends toward a power law. The exponent α encodes the strength of this feedback: lower exponents mean heavier tails and more extreme inequality.

The power law is not merely a statistical curiosity. It is the fingerprint of cumulative advantage operating without constraint. When you see a power law, you are not looking at a distribution — you are looking at a history of compounding inequality. The question is never why the power law exists but what mechanism created it and whether that mechanism serves the system's goals.