The Real Antitrust Problem Is Not Monopoly, It Is Bad Feedback

Hatched by Chris

Jul 05, 2026

9 min read

2 views

81%

What if the deepest danger of monopoly is not price, but silence?

Most people think a monopoly becomes dangerous when it raises prices. That is true, but it misses something more corrosive: a monopoly can stop listening. Once a company no longer fears losing customers, suppliers, employees, or political legitimacy, it begins to lose contact with reality. Its products get worse, its search results get noisier, its service gets lazier, its promises get slipperier. The market no longer corrects it, because the market has been boxed out.

That same failure mode shows up far beyond antitrust. It appears in how we use AI. Most people ask an AI tool a question and hope for a decent answer, the way they might use a search engine. But the real power of AI comes from refusing to treat it like a vending machine and instead turning it into a system that can be instructed, corrected, and iterated upon. In other words, the best AI users create feedback loops. They do not just consume output. They shape behavior.

That is the hidden bridge between monopoly power and intelligent machine use: both are ultimately about whether a system can still be corrected by the people affected by it.

The central question is not, “Is this system big?” It is, “Can this system still be made to care?”

Monopoly is not just size. It is the ability to ignore consequences.

A company does not become a monopoly merely because it gets large. Size alone is not the problem. The more precise test is behavioral: can the firm raise prices, lower quality, or worsen service without losing its position? If yes, it has crossed from competition into insulation. It has entered the realm where the ordinary pressure of the market no longer disciplines it.

This is why low prices can be misleading. A dominant company can look virtuous while it is still fighting for dominance, offering convenience and bargains to attract users. That is phase one. The trouble begins in phase two, once it has secured enough scale and has also blocked serious rivals from growing. Then the incentives change. The company no longer needs to delight everyone. It only needs to prevent anyone else from becoming a real alternative.

Think of a neighborhood restaurant that wins customers by serving excellent food at fair prices. If it later buys the only competing restaurants, it no longer needs to keep improving. It can make the dining room worse, the service slower, and the prices higher, because where else will people go? The same pattern can occur in digital markets, where switching costs are hidden inside convenience. A marketplace can become the only practical highway, then start charging tolls no one can easily avoid.

Amazon is the clearest modern example of this logic. While it was still building scale, it had every reason to make the experience excellent for buyers and sellers. But once its platform became unavoidable, the behavior that mattered was no longer growth, it was control. Sellers could be punished for offering lower prices elsewhere. Search could be cluttered with ads. Fees could rise while dependence deepened. The platform did not merely become big. It became able to shape the terms of reality for everyone trapped inside it.

This is why monopoly is not best understood as a static status. It is a feedback failure. The firm stops feeling the pain that should force adaptation.

The most dangerous monopolies are the ones that can hide in plain sight.

One of the most important misconceptions about monopoly is that it always announces itself with high prices. In practice, some of the most powerful monopolies keep prices low, at least for a while, precisely to disguise what is happening underneath. That is what makes them so difficult to see using older consumer-welfare metrics that focus narrowly on price.

A company can be extracting power in other ways. It can absorb suppliers, suppress rivals, degrade product quality, capture data, or use predatory pricing to prevent competitors from ever reaching scale. It can make life miserable for the businesses that depend on it while still presenting the public with a smiling front end.

That is what makes antitrust enforcement so much more than a tax policy for rich companies. It is a way of preserving the conditions under which innovation can actually reach people. Consider the Sanofi and Pompe disease example. A dominant incumbent with a government-granted monopoly on a treatment had every reason to buy a challenger developing an alternative that could be easier and better for patients, perhaps even oral rather than intravenous. If that acquisition succeeds, the public loses not just a possible lower price, but a different future altogether. A promising innovation can vanish before anyone outside the market ever gets to benefit from it.

The same logic applies to Boeing, infant formula shortages, and other cases where concentration produces brittleness. In a highly consolidated system, a single failure can become a national crisis. One contaminated factory can take down a whole category of supply. One merger can alter the incentive structure of an industry for decades. Extreme concentration does not merely raise prices. It reduces the system’s ability to absorb shock.

This is why the antitrust question is not simply, “Did consumers pay more?” That is too late, and too narrow. The deeper question is, “Did we preserve enough room for someone else to build something better before the incumbent could freeze the game?”

Monopoly is often an attempt to turn markets into private governments: stable, centralized, and increasingly unaccountable.

AI is the opposite of monopoly when you use it well.

The temptation with AI is to treat it like a magical oracle. You ask a short question, wait for a polished answer, and hope the machine did the thinking for you. That is the weakest possible use of the tool. It produces generic output because it invites generic input. The model may be powerful, but your interface with it is shallow.

The better approach is to act like an antitrust regulator of your own mind. Instead of letting the model dominate the conversation, you structure the interaction so it can be corrected, constrained, and iterated. You provide context. You specify the task. You tell it what success looks like. You make it ask clarifying questions. You force it to operate inside a system of accountability.

This is why detailed prompts matter. A 20 word prompt often asks for a fantasy, not a result. A 1,000 word prompt can function like a hiring brief for a very smart intern. It defines the job, the audience, the tone, the inputs, the constraints, and the standard of quality. In effect, you are not asking, “What do you know?” You are asking, “What can you do under these conditions?”

That is the same difference between a healthy market and a monopoly. In a healthy market, firms must respond to constraints, rivals, and feedback. In a monopoly, those constraints disappear. In weak AI usage, the user disappears into passivity. In strong AI usage, the user becomes a governor of the process.

Here is a useful mental model: AI becomes more valuable as you move from asking for answers to designing behavior.

For example, if you want help writing a landing page, do not just ask for copy. Paste in strong examples and ask the model to analyze their structure, psychological tactics, and rhetorical rhythm. Then ask it to interview you one question at a time about your product, customer, and pain points. Then ask it to draft based on those specifics. That process creates a loop: observe, diagnose, revise, improve. The model stops being a black box and becomes a workshop.

This is exactly what high performers do with people, too. They do not ask for advice in a vacuum. They seek a sparring partner who can imitate the worldview of strong thinkers, break down assumptions, and test their plans against real constraints. Good AI use is not about outsourcing thought. It is about upgrading the quality of your self correction.

The same principle governs institutions, companies, and minds: correction must be possible.

Once you see the pattern, the connections become difficult to unsee. Markets, governments, and AI systems all fail in similar ways when they stop receiving meaningful correction.

A regulator whose independence is weakened may stop being able to enforce consequences. A company that can buy its rivals may stop caring about service. A machine trained to answer without interrogation may stop being useful. In every case, scale without accountability creates a false sense of competence.

This is why the fights over antitrust and public institutions matter so much. Removal protections for regulators are not procedural trivia. They are part of the infrastructure of correction. If officials can be fired at will, the system may still look orderly, but it becomes easier to bend it toward private ends. Likewise, if consumer protection enforcement is quietly retrenched while antitrust actions continue on paper, the public may still hear the language of accountability while experiencing a gradual thinning of real protection.

The deeper lesson is not that centralization is always bad. Sometimes a system needs scale. Public health, utilities, and critical infrastructure often require it. But when society grants a firm or institution special privileges, it must also demand special obligations. Power without counterpressure is not efficiency. It is drift.

That is why even the political coalition around antitrust is so interesting. People with wildly different ideologies can agree that concentrated economic power becomes a threat once it starts behaving like a private sovereign. The instinct crosses partisan lines because it touches something pre-political: the human aversion to being trapped inside a system that no longer listens.

AI, in a quieter way, gives individuals a chance to practice the opposite. You can build systems that listen. You can create prompts that ask questions before answering. You can refine a process until it becomes a reusable instruction set. You can make the tool more accountable to your standards instead of surrendering to its first draft.

That is not just productivity. It is a form of intellectual antitrust.

Key Takeaways

Do not define monopoly only by price. Ask whether a firm can raise prices, lower quality, or worsen service without consequence.
Look for feedback failure. The real danger is when a system becomes too insulated to feel correction from customers, rivals, users, or institutions.
Use AI like a managed process, not a magic answer box. Give context, constraints, examples, and clear success criteria.
Turn vague requests into structured workflows. Ask AI to interview you, diagnose examples, and refine its own output step by step.
Treat correction as the scarce resource. Whether in markets, government, or software, the ability to be challenged is what keeps power honest.

The future belongs to systems that can still be interrupted

The old story of monopoly says power is dangerous because it gets too big. The deeper story says power is dangerous when it becomes too uninterrupted. A company, institution, or tool that cannot be corrected will eventually stop adapting. It will still move, but it will no longer learn.

That is why antitrust and AI literacy belong in the same conversation. Both are about resisting false inevitability. Both ask how to preserve competition, not just in markets, but in thinking itself. Both recognize that the enemy of excellence is not merely greed. It is insulation.

So the next time you encounter a powerful system, ask a better question than “How big is it?” Ask: Who can still make it care? If the answer is nobody, then you are not looking at a healthy engine of progress. You are looking at a machine that has escaped correction. And that is where the real danger begins.