Power laws sound abstract until you see them everywhere.

Once you start looking, you realize that most quantities in the social and economic world follow power laws. A few extreme observations dominate the rest. There's no "typical" value. The distribution has a long tail.

Here are the most compelling examples.

Earthquakes: The Gutenberg-Richter Law

Seismologists discovered that earthquake magnitudes follow a perfect power law: each unit increase in magnitude corresponds to roughly ten times fewer earthquakes.

This isn't a statistical approximation. It's a physical law that holds across different regions and time periods.

Practically:

• A region might experience thousands of small tremors (magnitude 3-4) per year.
• A few magnitude 5 earthquakes.
• Perhaps one magnitude 6 every decade.
• A major magnitude 7 every century.
• A catastrophic magnitude 8 or 9 every few centuries.

The pattern is consistent and predictable in frequency, but the timing is not. You can say with confidence that a magnitude 7 will occur in the region within the next 100 years. You cannot predict which year.

Why it matters: If you're building infrastructure in an earthquake zone, you need to design for the full distribution, not just the earthquakes you've observed recently. The biggest quake in the last 50 years tells you almost nothing about the biggest quake possible in that region.

Book Sales and the Long Tail

Before the internet, books were distributed through bookstores constrained by shelf space. A typical bookstore carried roughly 10,000 titles. These were books with the broadest appeal.

Amazon and digital distribution removed the shelf-space constraint. They can now carry millions of titles.

What happened? The long tail of books that never made it into physical bookstores now collectively account for an enormous share of total sales.

Practically:

• A tiny number of books (Harry Potter, Twilight, etc.) sell tens of millions of copies.
• A small number of moderately popular books sell hundreds of thousands.
• The vast majority of books sell hundreds or fewer copies.
• But the aggregate of millions of low-selling books exceeds the sales of the hits.

The distribution is power-law: a few massive bestsellers, then a gradually declining tail of decreasing sales.

Why it matters: The old retail model (focus on the top hits because shelf space is limited) was forced by constraint, not by the actual distribution. Once constraint is removed, the economics flip. Niche books, serving small audiences, become economically valuable in aggregate.

City Populations: Zipf's Law

City populations across a country or the world follow Zipf's Law: the n-th largest city has population proportional to 1/n times the largest city.

This isn't exact, but it's remarkably consistent.

Practically:

• If the largest city has 10 million people, the second-largest has about 5 million, the third about 3.3 million, and so on.
• This pattern holds whether you're looking at one country, multiple countries, or different historical eras.
• There is no "typical" city size.

The pattern emerges from the dynamics of city growth: larger cities attract more people because of more opportunities, which makes them larger still. This preferential-attachment dynamic produces the power-law distribution.

Why it matters: If you're designing urban policy or infrastructure and someone talks about "the average city," they're thinking in Gaussian terms. Policy designed for typical cities will fail at the extremes. A policy that works for cities of a million people won't work for cities of 5 million or for towns of 50,000. You need policy that scales with the power law.

War Casualties: The Power-Law Distribution

Wars vary wildly in size. Most wars kill relatively few people. A small number of wars dominate total casualties.

The distribution of war deaths by conflict follows a power law. The biggest wars (World War II, World War I) account for a disproportionate share of total war casualties across history.

Practically:

• Most military conflicts kill fewer than a few thousand people.
• Some kill hundreds of thousands.
• A rare few kill tens of millions.

The worst conflicts dominate the statistics. A history of wars is dominated by a small number of massive conflicts that would be catastrophically worse than the "average" war.

Why it matters: If you're estimating the risk of a future conflict based on "typical" past conflicts, you're ignoring the tail. The "typical" war is far smaller than the distribution's potential. Preparation needs to account for the possibility of the large wars that sit in the tail.

Wealth: The Original Pareto Observation

Vilfredo Pareto observed in the 19th century that wealth in a country wasn't distributed evenly. The richest 20% of people held 80% of wealth.

But more interesting: if you looked within that richest 20%, the same pattern held. The top 20% of the rich held 80% of the wealth within the richest 20%.

This recursive structure is the defining characteristic of a power law. The distribution looks the same at every scale.

Practically:

• A tiny fraction of people hold the vast majority of wealth.
• A larger fraction holds modest wealth.
• The distribution has no "typical" person.
• One billionaire in a sample of a thousand middle-class people dominates the average.

Why it matters: Using "average wealth" as a metric is meaningless in countries with high wealth inequality. The median (the middle value) is far more informative. When designing policy about wealth and taxation, you need to acknowledge that the distribution is power-law and that extreme wealth concentration is a feature of the distribution, not an anomaly.

Financial Returns: A Case Study

Stock market returns don't follow a Gaussian distribution. Mandelbrot showed this decades ago. The tail is fat. Extreme moves are far more common than the bell curve predicts.

A crash day (10%+ loss) should happen once every million trading days under Gaussian assumptions. In reality, they happen every few years.

Practically:

• Most days are calm with small moves (less than 2%).
• Some days have moderate moves (2-5%).
• Rare days have large moves (5-10%).
• Very rare days have catastrophic moves (10%+).

The distribution is power-law. The tail events where most of the economic damage concentrates are far more likely than standard models predict.

Why it matters: Banks, investment firms, and insurance companies that model returns using Gaussian assumptions will catastrophically underestimate tail risk. When the crisis arrives (not if, but when), they'll be unprepared. Institutions that acknowledge the power-law structure build in buffer for the tail they know they haven't observed but is likely to arrive.

The Cognitive Challenge

We're hardwired for Gaussian intuition. Our everyday experience — heights, weights, IQs, most biological measurements — follows something close to the bell curve. We intuitively expect a peak, a typical value, and rapid falloff.

Power laws feel wrong. They don't have a typical value. They have every scale simultaneously. One extreme observation dominates. This violates our intuition.

But intuition trained on biological Mediocristan is anti-calibrated for economic and social Extremistan. The fix isn't willpower. It's training yourself to ask the right question before reasoning:

"Can one observation dominate the total?"

If yes, assume power law. If no, assume Gaussian. Most of the time, that question determines which framework is appropriate.