Two musicians leave the same conservatory in the same year, equally gifted, equally hungry. One gets a mildly better first review — luck, a critic in a good mood, nothing more. That review nudges a few extra people to the next show; a fuller room draws a booker; the booking earns a bigger review; the bigger review fills a bigger room. Five years on, one is playing arenas and the other is teaching scales to teenagers. Ask anyone to explain the gap and they’ll reach for talent: the arena act must simply be that much better. But they started even. The gap wasn’t there at the start — it was manufactured, one self-reinforcing step at a time, from a difference so small it was basically a coin flip.
That is not a story about music. It is the default way that advantage compounds wherever getting ahead makes it easier to get further ahead — and once you can see it, you’ll find it running underneath wealth, fame, cities, companies, scientific careers, and the app on your phone that “everyone” uses. It has a name, and a signature shape it leaves on the world.
The idea, named
The engine is cumulative advantage: a reinforcing loop in which having more of something makes you more likely to get even more of it. More money earns more return; more followers get more recommended; more citations attract more citations; more links bring more traffic that brings more links. The advantage doesn’t just sit there being rewarded once — it feeds back into itself and grows.
The one-sentence version
Cumulative advantage (the Matthew effect, preferential attachment, “the rich get richer”) is a reinforcing feedback loop acting on a quantity: the more of it you already have, the faster you get more — so small, often random, early leads compound into vast, self-perpetuating gaps.
The name Matthew effect comes from sociologist Robert Merton, who borrowed a line from the Gospel of Matthew — “for unto every one that hath shall be given… but from him that hath not shall be taken away.” Merton noticed that famous scientists get disproportionate credit for work done with unknown collaborators: fame begets fame. Network scientists later gave the same loop a mechanical name — preferential attachment — for how new links in a network attach preferentially to already-popular nodes. Different fields, one loop. And that loop leaves a fingerprint you can spot from across the room.
The fingerprint: a power law
When cumulative advantage runs long enough, the resulting spread of outcomes is not a gentle bell curve clustered around an average. It’s a power law — a distribution with a few gargantuan winners and a long, thinning tail of everyone else. Here’s the tell that makes it a different animal from the familiar bell curve:
| Bell curve (normal) | Power law (scale-free) | |
|---|---|---|
| Typical example | Human height, exam scores | City size, wealth, book sales, links |
| Is there an “average” case? | Yes — most things sit near the mean | No — no single scale is “typical” |
| How rare are extremes? | Astronomically rare (no 100‑foot people) | Common enough to dominate the total |
| A single outlier can… | never move the average much | outweigh everyone else combined |
| Ratio of #1 to #1000 | small — a couple of factors at most | enormous — hundreds or thousands× |
Heights are bell-curved: nobody is ten times the average height, and the tallest person barely nudges the mean of a stadium. Wealth is power-law: one person in that same stadium can hold more than the other fifty thousand put together. The difference isn’t a detail — it decides whether the average tells you anything at all, and whether the story of the system is in its comfortable middle or its monstrous tail. (You met exactly this in fat tails; a power law is the canonical fat-tailed shape, and cumulative advantage is one of the main ways nature and society build one.)
Before you read — take a guess
Two brand-new streaming songs are, by any blind measure, equally catchy. One happens to land on a big playlist first and gets a few thousand early plays; the other doesn't. A year later the first has 40 million plays and the second has 60,000. What's the most precise explanation?
Why this earns a top spot in the latticework
Most people carry a single, bell-curve-shaped intuition about how the world distributes outcomes: a big comfortable middle, symmetric and rare extremes, an “average” that summarises everything. That intuition is correct for heights, shoe sizes, and measurement errors — and catastrophically wrong for wealth, fame, firm size, city size, word frequencies, war deaths, market moves, and pandemic sizes. Reach for the average in a power-law world and you’ll under-plan for the extreme that dominates the total, over-credit the winner’s skill, and mistake a reinforcing loop for a level playing field.
So the payoff of this course is a switch you learn to throw before you reason about any distribution of outcomes: which world am I in? A bell-curve world, where the middle is the story and the average means something — or a power-law world, where a handful of winners hold most of everything, the average is a distraction, and a compounding loop has quietly been picking winners the whole time.
Feel it in your hands
Here’s the engine itself, stripped to its bones. Fourteen piles start out exactly equal. Tokens drop one at a time, and each token picks a pile to join — but you control how much a pile’s current size boosts its odds of grabbing the next token. Slide that all the way down and it’s pure luck: every pile is equally likely, and they stay roughly even, a boring near-bell-curve. Slide it up and the loop switches on — bigger piles pull harder, so whichever pile stumbles into an early lead starts winning more, which extends its lead, which wins it more. Watch one tower rise out of a field of scraps. Then flip to the rank–size view and watch that lopsidedness straighten into the diagonal line that is the unmistakable signature of a power law.
Cumulative-advantage engine
Watch a winner get manufactured
Fourteen equal piles compete for tokens dropped one at a time. Set how strongly a pile’s current size boosts its odds of grabbing the next token, then keep dropping. At low strength luck keeps the piles even; crank it up and a tiny early lead snowballs into a runaway winner.
Press “Drop 300 tokens” to start feeding the piles.
- Tokens dropped
- 0
- Biggest pile
- —
- Top pile’s share
- —
- Inequality (Gini)
- —
Two things to notice, because they preview the whole course. First, the winner is chosen early and largely by luck — reset the run a few times at high strength and a different pile wins each time, none of them more “deserving” than the others. Second, strength controls inequality: nudge the loop’s power up and the top pile’s share and the Gini coefficient climb together. Turn the loop off and inequality collapses back toward zero. The distribution of outcomes isn’t handed down by how good the piles are — it’s generated by how strong the “more begets more” loop is.
The map of the course
Six teaching lessons build the model from the shape up to its hard limits, then one exam locks it in:
- What a Power Law Is — the scale-free distribution, the log–log straight line, and why it’s a different animal from the bell curve, with the everyday menagerie (cities, words, wealth, quakes, sales).
- The Engine: Preferential Attachment — the generative mechanism, step by step: the Matthew effect, “more begets more,” and why it manufactures a power law out of near-equal starts.
- The 80/20 Principle — Pareto’s rule as the friendly face of a power law, why “vital few, trivial many” recurs, and how to actually use it (and where 80/20 becomes 90/10, or nests inside itself).
- Winner-Take-All — superstar and tournament markets: why a razor-thin quality edge, made scalable by technology and network effects, pays off a hundred to one, not ten percent more.
- The Tyranny of the Tail — why the average is meaningless here, the sample mean never settles, one event can outweigh all history, and outcomes are luck-compounded far more than merit-proportional.
- Where the Model Lies — the honest limits: the log–log eyeball test proves little (log-normals in disguise), survivorship and the Matthew effect inflate the winner’s apparent skill, and the tail exponent is shakier than it looks.
Then a Final Exam — graded, one question at a time, one-way: once you answer, it locks. No back button, no retries, 70% to pass.
How to use this course
One rule does most of the work: guess before you peek. Commit to an answer on every exercise before you reveal the explanation — the small sting of being wrong is what welds the idea into memory. And keep coming back to the engine above; drag the strength slider from luck to runaway and back until one thing feels obvious in your gut: the shape of the outcomes is set by the strength of the loop, not the merit of the players. That single instinct is what the whole course is trying to install.
Next up: lesson 1, What a Power Law Is — the strange, scale-free shape that a bell-curve mind refuses to believe until it sees the log–log line go straight.