The Paradox of Capitalism: Is Money Still a Reward?

Getting your Trinity Audio player ready...

Capitalism is commonly defended as a system of incentives. A person produces something valuable, society purchases it, and the person receives money. In this simplified picture, wealth is not merely property: it is a reward for useful action.

But modern economic life increasingly contradicts this interpretation.

People may earn enormous sums from speculation, inherited assets, monopoly power, persuasive advertising, organizational politics, or control of platforms. At the same time, a mathematician, free-software developer, caregiver, teacher, independent scientist, or environmental volunteer may create substantial social value and receive little or nothing.

This creates a profound paradox:

If money no longer rewards good or useful actions, what moral meaning does capitalism retain?

At first sight, the conclusion appears obvious. If capitalism cannot connect contribution with reward, perhaps it has lost its justification and should be replaced by socialism or communism.

That conclusion, however, is too fast. Money may have lost one meaning while retaining another.

In contemporary capitalism, money often functions less as a reward for usefulness and more as evidence of a person’s demonstrated dedication to organizing and sustaining projects. That is an imperfect signal—but still an economically important one.

The proposed AI Internet-Meritocracy, or AIIM, attempts to restore the stronger meaning of money: payment as a reward for deeds that are actually useful. In a later economy dominated by superintelligence, money may acquire a third meaning—as a mechanism through which human beings exercise governance over artificial intelligence.

The historical progression can therefore be summarized as follows:

1. Money was a practical reward for useful work.
2. Money became an imperfect score for demonstrated project dedication.
3. AIIM could make money a practical reward for usefulness again.
4. Superintelligence may transform earning into a form of voting and human governance.

The original meaning of money as a practical reward

In a relatively simple economy, the relationship between work and reward is visible.

A farmer grows food. A carpenter builds furniture. A tailor makes clothing. A merchant transports goods from where they are abundant to where they are scarce. Customers directly experience the result and pay for it.

Money therefore serves several functions simultaneously:

• a medium of exchange;
• a store of purchasing power;
• a signal of demand;
• and, at least approximately, a reward for useful production.

The connection was never perfect. History contains inheritance, exploitation, rent seeking, slavery, conquest, fraud, and monopoly. Nevertheless, in small-scale production, the relationship between contribution and payment was often comparatively direct.

The baker who repeatedly produced bad bread lost customers. The craftsperson who produced durable tools acquired a reputation. Economic feedback was local, observable, and relatively rapid.

In this environment, money could plausibly be interpreted as a practical social message:

“You did something that another person considered useful.”

This is closely related to the conventional defense of capitalism. Private ownership, voluntary exchange, prices, competition, and profit are supposed to coordinate decentralized decisions. Prices direct labor and capital toward uses for which others are willing to pay.

However, willingness to pay is not the same thing as social usefulness.

Why money has lost its meaning as a reward for good actions

The modern economy separates value creation, value measurement, and value capture.

A person can create value without being able to charge for it. Another person can capture money without creating an equivalent amount of value. The larger and more complicated the economy becomes, the weaker the moral interpretation of income becomes.

Markets reward monetizable demand, not moral value

Markets answer a particular question:

How much will somebody with purchasing power pay?

They do not directly answer:

How much does this action benefit humanity?

These questions sometimes produce similar answers, but they can also diverge radically.

A cure for a neglected disease may have tremendous social importance but little commercial demand if the affected population is poor. A luxury status product may produce limited social benefit but attract high prices from wealthy consumers.

Markets weigh preferences by money. A billionaire’s minor preference may have more market influence than the urgent needs of thousands of poor people.

Therefore, market revenue measures effective demand, not usefulness in any complete ethical or civilizational sense.

Public goods are systematically under-rewarded

Scientific knowledge, free software, mathematical theorems, public datasets, educational materials, and environmental protection often benefit people who never pay their creators.

Once knowledge is publicly available, one person’s use usually does not prevent another person from using it. It may also be difficult to exclude non-payers. These characteristics make basic research and other public goods difficult to finance through ordinary commercial transactions.

The person who creates a public good may generate enormous value while capturing only a tiny part of it.

This is especially visible in science. Foundational discoveries can enable entire industries decades later, while the original researcher receives neither ownership of those industries nor payment proportional to their contribution.

Ownership is rewarded separately from contribution

Modern income frequently comes from owning assets rather than performing new socially useful actions.

A person may receive:

• rent from land;
• dividends from inherited shares;
• capital gains caused by general market growth;
• royalties from intellectual property acquired from another person;
• or profit from controlling infrastructure that others must use.

Some ownership performs a useful economic function. Investors accept risk, allocate capital, and sometimes exercise productive oversight. But ownership income cannot automatically be interpreted as a reward for the owner’s good deeds.

The asset may have been inherited. Its price may rise because of scarcity, network effects, regulation, or the work of unrelated people.

Money therefore accumulates partly according to prior ownership, not present usefulness.

Institutions reward measurable proxies

Large organizations cannot directly observe every employee’s true contribution. They rely on proxies:

• hours worked;
• sales figures;
• publication counts;
• citations;
• credentials;
• job titles;
• quarterly revenue;
• standardized performance indicators.

Once a proxy becomes a target, people learn to optimize the proxy rather than the underlying purpose. This is the structure commonly associated with Goodhart’s law: intense optimization of an imperfect measure can weaken or even reverse its relationship with the true objective. Research on machine-learning reward systems demonstrates the same general difficulty: an engineered reward is normally a proxy for the real goal, and over-optimizing that proxy can produce worse outcomes.

A university wants important research but counts publications.

A company wants durable customer value but rewards quarterly sales.

A government wants effective services but measures completed forms and budget utilization.

Money is then attached to the metric—not necessarily to the useful result.

Persuasion can be more profitable than production

Advertising and marketing can communicate real value. They help customers discover products and allow innovators to reach users.

Yet marketing can also distort allocation. A mediocre product with excellent promotion may outperform an excellent product that remains unknown. In attention markets, the capacity to attract clicks, stimulate emotion, or dominate distribution may be more profitable than solving an important problem.

Consequently, market success may indicate that a project was effectively sold, not that it was exceptionally useful.

Bargaining power affects pay

Income depends not only on contribution but also on the ability to claim part of the resulting value.

Workers and creators with rare credentials, legal protection, institutional authority, strong networks, or control over distribution can negotiate higher compensation. Those lacking these advantages may be underpaid even when their work is valuable.

The question is therefore not merely, “What did this person contribute?”

It is also:

“What portion of the contribution could this person successfully capture?”

Financial rewards may alter motivation

Money does not always strengthen socially beneficial motivation. Research on motivational crowding suggests that tangible rewards can, under certain conditions, weaken intrinsic motivation—particularly for activities that people already find meaningful. The effect is context-dependent and does not imply that payment is generally harmful, but it undermines the simplistic assumption that more financial reward always produces more genuine commitment.

The modern monetary system may therefore fail twice:

1. It may reward the wrong actions.
2. Its rewards may sometimes displace the non-monetary motives that supported the right actions.

Does this mean capitalism has lost all meaning?

Not necessarily.

The argument that money no longer reliably measures usefulness does weaken one moral defense of capitalism. It becomes difficult to claim that every wealthy person deserves wealth because society has objectively judged their actions to be good.

But capitalism does not depend exclusively on that claim.

Its practical strength also comes from decentralized experimentation, private initiative, transferable resources, freedom to establish organizations, competition between projects, and the ability to reinvest accumulated capital.

Modern capitalism may therefore be interpreted in a more modest way:

Money is not a reliable score of human goodness or social usefulness. It is an imperfect record of demonstrated capacity and dedication within monetizable projects.

This interpretation is less morally satisfying, but more defensible.

Money as evidence of dedication to projects

Running a successful project usually requires more than having a valuable idea.

It may require:

• sustained attention;
• planning;
• coordination;
• recruitment;
• negotiation;
• risk tolerance;
• adaptation to failure;
• communication;
• financial discipline;
• and years of repetitive work.

A person who has built and maintained an organization has demonstrated some combination of these capacities.

The resulting money does not prove that the project was morally good. It does not prove that the person created all the value. It does not eliminate the roles of luck, inheritance, privilege, timing, or monopoly.

Nevertheless, it may contain information.

Past project performance is often relevant when predicting future project performance. Someone who has repeatedly coordinated teams, managed budgets, delivered products, and responded to changing conditions may be more likely than an untested person to execute another project successfully.

Money can therefore work as a crude certificate:

“This person has previously participated in projects capable of attracting and retaining resources.”

It resembles an academic score. A high examination grade does not prove wisdom, creativity, morality, or future greatness. It indicates that the student performed well under a particular evaluation system.

Likewise, wealth does not prove social merit. It indicates success under a particular economic system.

Why this limited meaning still favors capitalism

This interpretation helps explain why capitalism can outperform rigid central planning even when monetary rewards are morally inaccurate.

Capitalism permits individuals who have accumulated resources to initiate further projects without obtaining permission from a central authority. Some of these people accumulated money through genuine productive competence. Allowing them to reinvest can therefore create a feedback loop:

1. A person demonstrates dedication by sustaining a project.
2. The project generates income or attracts investment.
3. The person gains control over additional resources.
4. Those resources finance another project.
5. Successful execution produces further evidence of capability.

This is not a system for rewarding virtue. It is a system for selecting and reproducing organizational capacity.

Socialist planning can theoretically allocate resources according to social need rather than private profit. But planners face severe informational and incentive problems. A ministry or committee must decide which projects deserve resources, who is competent to run them, when a failing project should be closed, and how unconventional ideas should be assessed.

Capitalist markets distribute these decisions among many actors. Multiple investors, founders, workers, and consumers can make conflicting judgments. Failed experiments normally lose resources, while successful projects gain opportunities to expand.

This decentralized selection mechanism is one reason capitalist economies have supported technological innovation and specialization, despite serious inequality and market failure.

The strongest argument for capitalism is therefore not:

“Rich people are necessarily good or deserving.”

It is:

“Control of resources is partly assigned to people who have shown that they can organize projects, and decentralized reassignment is often more adaptive than centralized administrative selection.”

That is a narrower and more realistic defense.

The limitation: dedication is not usefulness

A person can be extremely dedicated to a harmful, wasteful, deceptive, or trivial project.

History offers many examples of disciplined organizations pursuing bad objectives. Dedication is morally neutral. It measures persistence and organizational ability, not the worth of the goal.

This produces the central defect of modern capitalism:

It can identify who is effective at operating inside the reward system, but not reliably whether the rewarded result is good for humanity.

A casino operator, manipulative advertising network, monopolistic intermediary, medical researcher, and open-source developer may all exhibit dedication. The market may reward the first three more than the last two because their value is easier to monetize.

Capitalism is therefore efficient at some forms of selection while being poorly aligned with social merit.

It is an engine with a distorted objective function.

AIIM: restoring money as a reward for useful deeds

AI Internet-Meritocracy proposes a different allocation mechanism.

Instead of rewarding only what can be sold, an AI-based system would evaluate a person’s publicly documented contributions—such as scientific research, inventions, educational work, or free-software development—and allocate available funds according to their estimated value.

The related AIIM economic model describes the system as an automated scientific prize: rather than requiring each researcher or developer to win a conventional grant, sell a proprietary product, or persuade a small committee, the system would periodically evaluate contributions and distribute treasury funds proportionally.

The intended transformation is fundamental.

Under conventional capitalism:

Money follows transactions.

Under bureaucratic grant systems:

Money follows applications, credentials, institutional approval, and committee decisions.

Under AIIM:

Money is intended to follow useful contributions themselves.

This could reward activities that markets systematically undervalue:

• basic scientific research;
• mathematical theories;
• independent scholarship;
• maintenance of free software;
• replication studies;
• negative experimental results;
• open educational resources;
• scientific communication;
• and discoveries created outside established institutions.

AIIM would not abolish markets. People could still found companies, sell goods, invest, and trade. Instead, it would add a new layer of economic feedback specifically designed for public contributions.

In that sense, AIIM is neither conventional capitalism nor classical socialism.

It preserves freedom of initiative: people choose what to investigate or build. But the payment is not determined solely by sales or by a central planning ministry. It is allocated through algorithmic evaluation, auditable rules, a shared treasury, and human governance.

From market score to merit score

AIIM attempts to correct a category mistake in modern economics.

Market price answers:

“How much can this result earn through exchange?”

AIIM aims to answer:

“How useful is this result, including benefits that cannot easily be sold?”

The distinction is particularly important for knowledge. The social value of a theorem, algorithm, scientific dataset, or free-software library may be distributed across millions of future uses. No individual transaction captures the total.

An AI system can, at least in principle, examine broader evidence:

• dependencies between projects;
• citations and technical adoption;
• originality;
• downstream applications;
• expert criticism;
• replication;
• software usage;
• historical priority;
• and benefits to other contributors.

Instead of treating each sale as a vote, it could model the contribution’s place in a network of human knowledge and production.

Money would again become practical rather than ceremonial. It would finance the person who performed the useful work, allowing that person to continue working.

AIIM does not automatically solve the measurement problem

AIIM should not be described as an infallible judge.

An AI evaluator can inherit bias from training data, favor visible contributors, misunderstand unconventional research, be manipulated through search-engine optimization, or optimize inaccurate proxies. Prompt injection and fabricated evidence create additional risks.

The AIIM proposal itself recognizes this problem and includes human voting over sanctions such as banning and unbanning malicious participants. It argues that human oversight is necessary because automated systems can share correlated vulnerabilities and cannot safely serve as their own final judges.

A credible merit-based system would therefore require:

• transparent evaluation criteria;
• explanations for allocations;
• appeal procedures;
• adversarial auditing;
• plural AI models;
• independent human review;
• anti-Sybil safeguards;
• correction of false information;
• and governance capable of changing the rules.

The objective is not to create an omniscient machine. It is to construct a better approximation of social usefulness than sales, inherited ownership, institutional prestige, or grant-writing skill alone.

AIIM remains a proposed and developing model, not proof that the measurement problem has been solved.

The temporary restoration of practical reward

Suppose AIIM succeeds.

For a period, money could again carry a relatively direct practical meaning:

“You performed a useful deed, and society is financing you because of that deed.”

A scientist who discovers an important principle would receive resources.

A programmer who maintains essential free infrastructure would be paid even if the software remains freely available.

An educator whose materials improve learning could receive compensation without placing them behind a paywall.

An independent researcher would not need a university position merely to become economically visible.

This would restore the productive loop that the moral interpretation of capitalism promised but never fully delivered:

1. Perform useful work.
2. Receive money.
3. Use the money to live and continue working.
4. Produce further useful results.

Yet even this restoration may be temporary.

Superintelligence and the declining economic value of human deeds

A sufficiently capable superintelligence could outperform humans in many intellectual activities.

It might generate scientific hypotheses, prove theorems, design software, diagnose technical failures, optimize infrastructure, and produce educational materials faster than human specialists. Human work would then become less economically scarce.

This creates another paradox.

AIIM is intended to reward useful human contributions more accurately. But if artificial intelligence eventually performs nearly all useful cognitive labor, there may be progressively fewer uniquely human contributions to reward.

Money could no longer function primarily as compensation for productive human labor because human labor would no longer be the dominant productive input.

A society facing this condition might adopt universal basic income, public ownership of AI infrastructure, social dividends, universal capital shares, or some combination of these mechanisms.

But AIIM suggests another possible role for human beings.

Money as a means of voting over AI

Even when AI systems perform most productive work, society will still face normative questions:

• Which goals should AI pursue?
• Which risks are acceptable?
• Which projects should be prohibited?
• How should resources be distributed?
• Which scientific directions deserve priority?
• What constitutes abuse?
• Who may participate in governance?
• How can an automated decision be appealed?

These are not merely optimization problems. They determine whose values and interests govern civilization.

The AIIM model therefore gives humans a continuing role as judges and voters. Its proposed governance includes voting on whether participants should be banned or restored, with identity verification intended to limit manipulation through fake accounts.

For a broader discussion, see AI Requires Human Judgment: A Future Beyond Terminator.

In such a system, earning money could evolve into earning governance weight, eligibility, reputation, or some related capacity to influence the AI-mediated economy.

Money would then stand between its previous meanings.

It would not be purely practical payment for labor, because machines would perform much of the labor.

It would not be merely an impractical social score, because control over resources and governance would have real consequences.

Instead, it would become a constitutional instrument:

a socially recognized claim to participate in decisions about the automated systems that produce and distribute wealth.

Why governance rights should not simply equal wealth

This idea requires an important qualification.

Unrestricted “one dollar, one vote” governance would allow the richest participants to control the AI. That would reproduce oligarchy rather than preserve human agency.

Therefore, AI governance should distinguish among:

• money for consumption;
• money for investment;
• reputation for verified contribution;
• identity-based voting rights;
• delegated expertise;
• and universal political rights.

AIIM’s current concept of one verified user receiving one vote on banning decisions points in this direction. Economic rewards and basic governance rights need not be identical.

A robust future system might combine several principles:

• one person, one vote for fundamental rights;
• earned reputation for technical or scientific judgments;
• limited delegation to trusted experts;
• transparent AI recommendations;
• and constitutional constraints that neither money nor majority voting can override.

Money might support participation without becoming the sole source of political power.

Four stages in the meaning of money

The development can be represented as four economic stages.

Stage	Primary meaning of money	Main allocation mechanism	Central weakness
Early exchange economy	Practical reward for visible useful work	Direct exchange	Limited scale and unequal bargaining
Modern capitalism	Score of market success and project dedication	Markets, ownership, institutions	Weak connection to total social usefulness
AIIM economy	Practical reward for evaluated useful contribution	AI evaluation with human governance	Bias, manipulation, and measurement errors
Superintelligent economy	Claim to resources and participation in AI governance	Automated production plus constitutional voting	Risk of technocracy, oligarchy, or loss of human control

The transition is not necessarily linear. All four meanings may coexist.

A worker may receive wages for practical labor. An investor may receive money because of ownership. A scientist may receive AIIM funding for public value. A citizen may use tokens or identity-based rights to govern an automated institution.

Money is not one thing. Its social meaning depends on the institutions that issue, allocate, and recognize it.

Capitalism should be repaired, not interpreted as moral truth

The failure of money to measure goodness does not prove that every feature of capitalism should be abolished.

It proves that market income should not be confused with moral desert.

Capitalism remains useful as a decentralized mechanism for testing initiatives, organizing production, and identifying some forms of sustained project competence. But it should not be expected to measure every kind of contribution.

Socialist institutions can correct market failures, guarantee basic services, fund public goods, and reduce destructive inequality. Yet centralized systems can also become bureaucratic, politically captured, conservative toward unconventional ideas, and dependent on imperfect administrative metrics.

The relevant choice is therefore not necessarily capitalism or socialism.

A more productive question is:

Which allocation mechanism is suitable for each type of resource and decision?

Markets may remain effective for ordinary goods.

Public provision may be necessary for universal services.

Democratic institutions should govern fundamental rights.

AI-assisted merit systems may be useful for rewarding public intellectual and technical contributions.

The future economy may be a synthesis rather than an ideological victory by one nineteenth- or twentieth-century model.

The deeper paradox

Capitalism works partly because money does not perfectly measure virtue.

It rewards persistence, adaptation, persuasion, risk taking, coordination, and the ability to maintain an organization under real constraints. These capabilities are useful for running projects, even when the project’s social value is uncertain.

But capitalism becomes morally dangerous when success under its rules is treated as proof of personal worth.

Wealth may indicate dedication. It may indicate ownership. It may indicate bargaining power, luck, monopoly, institutional position, or inherited advantage. Usually it reflects several of these factors at once.

AIIM proposes to add missing information: an explicit evaluation of usefulness.

Its purpose is not merely to distribute more money to researchers and developers. It is to change what receiving money communicates.

Under the present system, wealth often says:

“This person successfully captured economic value.”

Under AIIM, a payment would aim to say:

“This person produced a contribution that benefits others.”

Under superintelligence, an earned economic position might eventually say:

“This human remains an acknowledged participant in governing the intelligence that operates society.”

Conclusion: money can recover meaning

Money has not become meaningless. Its meaning has changed.

It was once closely associated with direct, visible, useful production. In modern capitalism, it has become an indirect and often distorted signal of market success, ownership, and dedication to projects. This is similar to a score: informative within a system, but easily mistaken for a complete judgment of merit.

That remaining signal helps explain capitalism’s resilience. People who have demonstrated the ability to sustain projects are sometimes well positioned to run future projects. Decentralized reinvestment can use this information more flexibly than centralized allocation.

But dedication alone is insufficient. Society also needs to know whether a project is useful.

AI Internet-Meritocracy proposes to restore this missing connection by evaluating and rewarding scientific, technical, and other public contributions that markets frequently fail to finance.

For a transitional period, this could make money practical again: useful work would provide the means to continue useful work.

Superintelligence may eventually weaken that model by making human productive deeds economically less important. Yet it would not eliminate the need for human judgment. Money, reputation, identity, and governance rights could then converge into a new institution through which people supervise AI, contest its decisions, and determine the ends toward which automated production is directed.

The future meaning of earning may therefore be neither ordinary wages nor a meaningless score.

It may be the right to remain a participant in civilization.

---

Frequently Asked Questions

Has money stopped rewarding useful work?

Not completely. Markets still reward many useful products and services. The problem is that income also depends on ownership, scarcity, bargaining power, marketing, institutional position, and the ability to monetize a contribution. Money is therefore an unreliable measure of total social usefulness.

Does inequality prove that capitalism has failed?

Inequality alone does not establish that every capitalist mechanism has failed. It does show that market outcomes cannot automatically be interpreted as fair rewards. The relevant questions include how wealth was acquired, whether opportunities were equal, and whether the resulting allocation improves or harms society.

Why can capitalism be effective if money does not measure merit?

Money can still contain information about project execution. Building a profitable organization may demonstrate persistence, coordination, adaptability, and resource management. These qualities can help predict the ability to run future projects, although they do not prove moral merit.

What is AI Internet-Meritocracy?

AI Internet-Meritocracy is an app in which artificial intelligence evaluates documented scientific, technical, and other contributions and distributes treasury funds according to estimated usefulness. Human governance would supervise sanctions, disputes, and risks.

Is AIIM a form of socialism?

It has similarities to socialism because payments are not determined solely through market exchange. However, it does not require a central plan assigning people their work. Individuals remain free to choose projects, while an evaluation system rewards completed contributions. It is better understood as a proposed hybrid economic institution.

Could AIIM replace capitalism?

AIIM is more plausibly a complement to markets than a complete replacement. Markets could continue allocating ordinary private goods, while AIIM finances public goods and useful contributions that cannot be adequately monetized.

What happens to money after superintelligence?

No one knows with certainty. If superintelligence performs most productive work, money may shift from being primarily a wage for labor toward being a claim on automated production, a social dividend, an investment right, or an instrument of governance.

Why would superintelligent AI need human voters?

Technical intelligence does not by itself establish legitimate social goals. Humans must still determine rights, priorities, acceptable risks, sanctions, and distribution principles. Human voting and appeals can also provide an independent check against correlated failures among automated systems.

More importantly, LLMs are not many and most of them are similar to each other, making “judges” viable to the same attacks as “victims”, so maybe AI Should Not Judge Itself.