This article explores the concept of power laws, contrasting them with normal distributions, and explains their presence in natural and human systems. It discusses the implications of power laws in various domains such as income distribution, natural disasters, and investing, emphasizing the importance of recognizing the type of system you are in to make better decisions and take intelligent risks.
Some phenomena in the world follow what we call a normal distribution, where most data clusters around an average value. Examples include human height, IQ, or the size of apples on a tree. However, many important aspects of life do not follow this pattern. Instead, they follow power laws, which have very different properties and implications.
Power laws describe relationships where the frequency of an event varies as a power of some attribute of that event. Unlike normal distributions, power laws have heavy tails, meaning extreme events are more common than expected. For example, the size of world wars measured by casualties follows a power law, with a much higher likelihood of very large events than a normal distribution would predict.
In the late 1800s, Italian engineer Vilfredo Pareto discovered a power law in income distribution across several European countries. He found that the number of people earning more than a certain income decreases as a power of that income. This means there are a few people earning vastly more than the average, a pattern that persists today.
Pareto plotted income data on a log-log scale and found a straight line with a gradient around -1.5, indicating a power law relationship:
Number of people earning ≥ X ∝ 1 / X^1.5
This pattern was consistent across countries, showing the universality of power laws in income.
In a simple coin toss game where you win $1 for each head in 100 tosses, the expected payout is $50. The outcomes cluster around this average, forming a bell-shaped curve known as the normal distribution. This distribution arises when many independent random additive effects combine.
In a game where your winnings multiply by 1.1 for heads and 0.9 for tails over 100 tosses, the expected payout remains $1. However, the distribution of outcomes is skewed, with a long tail of very large payouts possible but unlikely. When plotted on a logarithmic scale, this distribution appears normal, hence the name log-normal. This distribution arises from multiplicative random effects.
In a game where you toss a coin until you get heads, and your payout doubles with each toss, the expected payout is theoretically infinite. The probability of large payouts decreases exponentially, but the payout increases exponentially, resulting in a power law distribution:
Probability of payout x ∝ 1 / x
This distribution has no finite average or standard deviation, making it fundamentally different and more unpredictable than normal or log-normal distributions.
Power laws imply that rare, extreme events dominate averages and outcomes. For example, the presence of billionaires skews the average wealth in a room dramatically. This phenomenon also appears online, where a few servers hold data for millions, and breaches can have widespread effects.
Power laws often emerge in systems at a critical point, where the system is poised between order and disorder. Examples include:
These systems exhibit fractal, self-similar structures and are scale-free, meaning no characteristic size dominates.
The concept of self-organized criticality explains how some systems naturally evolve to a critical state without external tuning. The sandpile model illustrates this: grains of sand added to a pile cause avalanches of various sizes, following a power law distribution. This behavior resembles forest fires and earthquakes, suggesting a universal mechanism.
At critical points, diverse systems exhibit universal behavior independent of their microscopic details. This universality allows simple models to capture the essential dynamics of complex phenomena, from magnets to ecosystems to human social systems.
Power laws appear in many human contexts:
For example, venture capital firms often see most investments fail, but a few outliers produce extraordinary returns that drive overall profits. Similarly, book publishing and streaming platforms see a few hits dominate attention and revenue.
Understanding whether you are operating in a normal or power law environment is crucial:
This means embracing uncertainty and making many bets, knowing most will fail but a few can change everything.
Power laws can arise from mechanisms like preferential attachment, where entities that are already successful are more likely to gain further success. This explains the distribution of links on the internet and many social phenomena.
The world is often shaped by power laws, where rare events dominate outcomes. Recognizing this changes how we approach risk, success, and decision-making. Instead of avoiding risk, the goal is to make repeated, intelligent bets, understanding that one wild success can outweigh many failures. This perspective helps us better understand natural disasters, economic inequality, innovation, and much more.
All simulations and models discussed are available for free use, and further exploration can deepen understanding of these fascinating phenomena.
