AI Wealth Truth (03): Why 'Fair' Markets Make Inequality Worse

I. We usually believe fair competition reduces inequality. If the rules treat everyone the same, with no corruption, no discrimination, no inheritance. The strong win, the weak lose. Resources end up distributed by "ability". It sounds reasonable.

II. But math tells us: perfectly fair competition leads to perfectly unfair outcomes. This is not a paradox. It is the inevitable logic of network effects.

III. In 1968, sociologist Robert K. Merton systematically studied the "Matthew Effect". The name comes from a line in the Gospel of Matthew: "For to everyone who has, more will be given, and he will have an abundance; but from the one who has not, even what he has will be taken away." Merton found that in science, papers by famous scholars receive more citations. Less famous scholars can write papers of the same quality and still get ignored.

IV. Behind it is a positive feedback loop. You have reputation. So more people read your work. So more people cite you. So you gain more reputation. You have no reputation. So nobody reads you. So nobody cites you. So you lose even the little you have. Small initial differences are amplified into huge gaps.

V. In 1999, network scientist Albert-László Barabási found a precise mathematical form for this mechanism. He studied the link structure of the World Wide Web and found that link distribution follows a power law. A few sites (Google, Facebook) accumulate massive inbound links. Most sites have only a handful.

VI. Barabási proposed the preferential attachment model. When a new page chooses where to link, it tends to link to pages that are already popular. Why? Because those pages are easier to find. This is completely "fair". No one is discriminating against anyone. The new page is simply making the most reasonable choice. But the result is: the rich get richer, the poor get poorer.

VII. Mathematical proofs show that a power law becomes inevitable as long as two conditions hold:

1. The system keeps growing (new nodes keep joining).
2. New nodes prefer to connect to nodes that already have many connections (preferential attachment). These two conditions hold in almost every social system.

VIII. Wealth distribution follows the same logic. You have money. So you access better opportunities that ordinary people never see. So you make more money. So you have more money. You have no money. So you can only deposit in a bank at 2% a year. So you lose to inflation. So you have even less. Your starting point decides which loop you enter.

IX. Follower distribution on social platforms obeys the same law. If you have a million followers, the algorithm shows you to more people. You gain more followers. If you have a hundred followers, the algorithm barely shows you. You stay at a hundred. The algorithm is "fair". It promotes content that "performs well". But performance is shaped by accumulated history.

X. There is a key epistemic problem here. We cannot distinguish "truly excellent" from "excellence amplified by the Matthew Effect". Is a famous scholar's paper actually ten times better, or is it cited more because the author is famous? Is a creator with ten million followers actually one hundred thousand times better than a creator with a hundred followers, or are they recommended more because they are already big? We cannot know. The signal has already been distorted.

XI. In 1989, economist W. Brian Arthur studied positive feedback in technology adoption. Which was better, VHS or Betamax? By technical specs, Betamax was slightly better. But VHS won, because early on it gained a slightly larger market share. Then studios prioritized VHS. Consumers preferred VHS. VHS gained more share. In the end, the slightly worse technology dominated the market.

XII. Was that process "fair"? Completely fair. Every consumer made a rational choice. No one forced anyone. But the result is: one slightly leading product eats the entire market. Free choice plus positive feedback equals winner-take-all.

XIII. The QWERTY keyboard layout is another case. It was designed to prevent typewriters from jamming. Typewriters are long gone. There is evidence that other layouts (like Dvorak) can be more efficient. But QWERTY became the standard. Everyone learns QWERTY. Every keyboard ships with QWERTY. Switching costs are too high. So it stays the standard. First-mover advantage gets locked into permanent advantage.

XIV. This destroys faith in "fair competition". We think fair competition makes the best win. In reality, fair competition makes the first to lead win. Then positive feedback amplifies a tiny lead into a crushing advantage. Being excellent and being ahead are not the same thing.

XV. Worse, positive feedback is hard to reverse. Once the gap forms, latecomers almost never catch up. The leader's advantage self-reinforces over time. Unless an external shock hits (policy intervention, technological discontinuity), the structure does not change.

XVI. AI makes this effect even more extreme. In the traditional economy, physical constraints slow the Matthew Effect down. Even the most successful company still needs time and labor to produce, ship, and serve. AI removes those constraints.

XVII. Once an AI model is trained, it can be copied infinitely. It can serve a billion users, with near-zero marginal cost. That means first place can take 99% of the market. Second place might not even get 1%. Network effects plus AI's near-zero marginal cost creates the most extreme winner-take-all dynamic in history.

XVIII. Look at the competitive landscape in AI. Once ChatGPT leads, it gets more users and more data. More data makes it better. A better product attracts more users. That is a positive feedback loop. How does a latecomer catch up? You need data, but the users are already using ChatGPT. First-mover advantage is being locked into permanent advantage.

XIX. How did Google Search become a monopoly? At the beginning it was only slightly better than other search engines. Then more people used it. It collected more query data. The algorithm improved. More people used it. Twenty years later, Google holds more than 90% of global search share. A slightly better product becomes an absolute monopoly under positive feedback.

XX. Social networks work the same way. Facebook was not the first social network. MySpace came earlier. But at a certain moment, Facebook gained a slightly larger user base. Your friends were on Facebook. So you joined. So more people joined. Network effects amplified a tiny lead into a monopoly.

XXI. What does this mean for you? If you are not among the first to enter a field, you might never catch up. Not because you are not good enough, but because the feedback mechanisms have locked the structure. Your effort gets swallowed by the Matthew Effect.

XXII. The only escape hatch is: change the game. Grinding inside a field where the feedback loop has already formed is futile. You should look for a new field where the feedback loop has not formed yet. There, you can become the tiny early lead. Then you can let the positive feedback work for you, instead of fighting against it.

XXIII. But this also means most people are destined to lose. Each field can only support a tiny number of winners. Most people become fuel for the Matthew Effect, contributing to the winners' success. This is not conspiracy. It is math.

XXIV. The "fairer" the market, the more pronounced this effect becomes. Because "fair" means no external interventions to block positive feedback. Progressive taxes, antitrust enforcement, inheritance taxes are interventions. Remove them to let the market "compete freely", and the result is: winner-take-all accelerates and inequality worsens. "Fair competition" is the most efficient machine for producing inequality. In the AI era, this machine becomes ten thousand times more efficient.