Surplus Humans and the Politics of Containment

The post-work problem is not unemployment alone. It is the management of people whose claims remain while their necessity declines.

The state does not announce surplus.

It manages it.

There is no policy document that says the economy no longer needs you. There is no official declaration that your claim on the system has become administratively inconvenient. The signal never arrives directly. What arrives instead is a set of arrangements: transfers, platforms, therapeutic frameworks, attention systems, civic rituals, and algorithmic participation that absorb the presence of people the productive order has learned to operate without.

This is not malice.

It is adaptation.

When the wage weakens as the institution that turns contribution into claim, when markets move their decisive acts upstream into interfaces, when capital captures surplus from systems built on collective inheritance, the political system faces a structural problem it cannot name openly: what to do with the people whose claims remain after their necessity declines.

The surplus human is not someone without needs. She is not someone who has stopped wanting recognition, income, standing, or a role. She is someone whose claims remain intact while the system has quietly learned to operate around her.

Strictly speaking, there are no surplus humans. There are surplus claims: to income, to standing, to recognition, to a role, claims the productive system no longer knows how to redeem through necessity. That is why the phrase is so violent. It locates in the person what actually lives in the classification.

And she still votes.

That is what makes the politics of containment different from simple exclusion. The surplus population is not outside the political system. It is inside it, with full formal membership, with grievances, with ballots, and with an accumulating sense that the system has stopped making good on the promises it extracted compliance to honor.

The system still needs these people politically, morally, and symbolically after needing fewer of them productively.

That is the contradiction.

They may be less central to production, but they remain central to legitimacy. They may be less necessary to output, but they remain necessary to social peace. They may lose leverage inside the economy, but they do not lose their formal claim on the political order.

A society can automate tasks faster than it can automate belonging. It can reduce labor dependency faster than it can reduce the moral claims of the people labor once explained. It can thin the wage without yet building another grammar of membership.

Managing that interval, not resolving it, not explaining it, not eliminating it, is the central political task of the AI economy. And the toolkit for managing it is already taking shape.

The Claim Outlives the Necessity

The old political economy had a convenient alignment: people who were economically necessary also had political weight.

Workers could strike because production depended on them. Professionals could demand status because institutions needed their judgment. Young people could expect absorption because organizations required junior labor to reproduce senior capacity. The worker, the professional, and the apprentice all occupied recognizable positions inside the productive order.

That alignment is loosening.

AI does not need to eliminate labor to change this. It only needs to make large categories of human contribution less scarce. A firm can preserve visible output with fewer juniors. A sector can expand productivity while hiring less. An institution can keep delivering services while reducing the cognitive dependency that once gave workers leverage. The exposure estimate that opened this series, roughly 60 percent of jobs in advanced economies affected, some complemented and others facing substitution, is only the surface. The deeper change is that even workers who remain employed may find that the economy’s dependence on their specific contribution has weakened.

They are still present. They are simply less necessary.

The claims, meanwhile, do not disappear. Votes remain. Political voice remains. Consumption power, however thin, remains. The capacity for disruption remains. The expectation of recognition remains. The sense of entitlement to a recognizable role remains, because modern society spent a century teaching people that work was the bridge between existence and belonging.

That bridge does not dissolve the moment the productive system learns to cross it differently.

The political system cannot simply ignore the people left on the wrong side of that crossing. It cannot explain them away. It has to manage them. And because they vote, management has to be sophisticated enough to maintain at least the appearance of recognition while the underlying condition persists.

This is why the post-work problem is not idleness.

It is recognition.

People do not merely need income. They need a socially recognized reason to believe their presence matters. A transfer can prevent destitution. A platform can produce engagement. A credential can extend preparation. A wellness system can help manage distress. But none of these automatically restores the role that wage society once promised: a place where effort, status, income, discipline, and belonging appeared to form one coherent arc.

When that arc breaks, people do not become passive.

They become difficult to explain.

The Containment Toolkit

The welfare state was built around the necessary worker. Its premise was simple: the people who labored near the center of production had to be protected, educated, insured, and represented, because the economy needed them and the polity depended on them.

What is being assembled now starts from a darker premise. It manages people whose claims remain after their necessity declines. It does not integrate them into production. It stabilizes them outside the old grammar of usefulness: transfers without centrality, participation without leverage, therapy without explanation, credentials without absorption.

This successor arrangement has no name, because naming it would require admitting what it is for. So it borrows the welfare state’s vocabulary instead. Containment does not announce itself as containment. It announces itself as support, opportunity, wellness, engagement, empowerment, and care. The language is always that of provision. The structure is that of management.

The toolkit has five components.

The first is income stabilization: transfers, credits, supplements, UBI pilots, and platform-mediated earning that prevent destitution while doing little to restore the leverage or recognition that wages once conferred. These systems are genuinely necessary. They prevent acute suffering. They also decouple income from standing. This is welfare as maintenance: the material problem addressed, the legitimacy problem left intact.

The second is platform participation: the conversion of surplus presence into engagement metrics, content creation, digital identity performance, and algorithmic circulation. Platforms discovered before governments did that people who are no longer economically central still have enormous amounts of time, attention, and need for recognition, and that discovery became a business model. The creator economy, the gig economy, and gamified productivity tools all provide systems that feel like contribution without conferring the standing that wage labor once produced.

The third is therapeutic normalization: the expansion of wellness systems, resilience discourse, and individual self-management practices to absorb what are structurally produced conditions. This does not mean care is false or unnecessary; many people need it, and much of it is real. The political point is different. The vocabulary does the work. Resilience asks the individual to absorb what the institution has offloaded. Burnout describes as a personal energy problem what is often a structural recognition problem. Mindfulness teaches the worker to observe her anxiety without asking what the anxiety is accurately perceiving. Employers expand wellness programs alongside thinning career ladders; universities add counselors as the credential’s redemption value falls. None of this is coordinated. It is the path of least resistance: treating the distress is cheap and scalable. Treating the structure is not for sale.

The fourth is credential extension: the continuous expansion of degree requirements, certification programs, badge systems, and reskilling initiatives that keep people in preparation mode longer. It delays the moment when the gap between preparation and absorption becomes undeniable. It generates revenue for the education and training industries. And it allows institutions to continue demanding performance from people whose prospects they are, structurally, no longer in a position to honor.

The fifth is civic ritual: voting, national service proposals, participatory governance experiments, and digital democracy platforms that maintain the appearance of consequential participation while the decisive acts of allocation move upstream into interfaces, procurement systems, and technical architectures that no democratic process meaningfully governs.

Each component is real, and each provides something. None resolves the underlying condition.

Why Containment Works, For a While

Containment is not stable.

But it is durable.

It persists not because it satisfies people, but because it fragments the conditions that would produce a coherent political response to their situation.

The surplus human’s condition is produced by multiple overlapping systems: compute ownership, market interfaces, AI-mediated hiring, platform distribution, credential inflation, declining wage leverage, and the weakening of the entry-level roles through which institutions once absorbed and socialized the young. No single actor is responsible. No single villain can be identified, and no single reform would address the aggregate.

This fragmentation is politically significant for the reason the tenant condition made visible earlier in this series: the environment has no single landlord. There is no focal antagonist around which a movement can coalesce, only an aggregate condition that no individual actor is responsible for producing.

The surplus human may be angry. She will find it difficult to direct that anger coherently at its structural source. The platform provides content that gives her anger an object. The transfer gives her enough to prevent the desperation that produces organized disruption. The credential system gives her a task that feels productive. The therapeutic framework tells her that her difficulty is partly about mindset. The civic ritual tells her that she participates in the decisions that shape her life.

Together these systems do not eliminate the problem.

They manage it at a level below political crisis, most of the time.

When the System Has No Face, People Attack Its Body

When containment frays, the anger does not remain formless.

It finds targets.

And increasingly, the targets are not abstract systems or dispersed corporate actors. They are the physical infrastructure of the AI economy: the data centers, the energy contracts, the chip foundries, the cloud campuses, and the fiber corridors that make synthetic production possible.

This is already happening. In Virginia, the Prince William Digital Gateway, planned as the largest data center campus in the world, was approved by an outgoing county board in December 2023 after a 27-hour public hearing, voided by a circuit court in August 2025 for defective public notice, and abandoned by the county in April 2026 after years of resident litigation and organizing. By that point, the share of Virginia voters comfortable with a new data center in their community had collapsed from 69 percent to 35 percent in three years, and one industry tracker counted 48 data center projects blocked or delayed across the United States in 2025 alone. In Arizona, the Tucson city council voted unanimously in August 2025 to refuse annexation and water access for Project Blue, a 290-acre data center campus linked to Amazon, after weeks of public mobilization in a drought-stressed desert city. In Congress, members of both parties have moved from rhetoric to bills: the Clean Cloud Act would set emissions standards for data centers, and the bipartisan AI-Related Job Impacts Clarity Act would require companies to report the jobs AI eliminates. [1]

This is not irrational. It is the political economy of visible infrastructure meeting the political economy of invisible displacement. The data center is concrete, local, and photographable. The algorithmic process that filtered a job applicant out of consideration before any human reviewed the resume is not. The cooling tower uses water the community can measure. The model that depreciated a professional skill set operates in a cloud region the affected worker cannot locate on a map.

When economic displacement is structurally diffuse but its infrastructure is physically concentrated, the infrastructure becomes the target.

This dynamic has historical precedent. Luddite machine-breaking was not random vandalism. It was skilled workers attacking the specific capital equipment whose introduction was destroying their trade. The machines were the visible, attackable form of a broader economic transformation whose ultimate authors and beneficiaries were considerably harder to reach. [2]

What makes the advanced-economy version of this dynamic distinctively volatile is something the Global South comparison makes visible.

The condition this series tracks as emerging, a widening gap between the claims people hold and the place the formal economy will grant them, is not new. Billions outside the OECD have lived a version of it for decades, absorbed through informal labor markets, remittance dependence, NGO service delivery, and the particular containment toolkit of societies that never fully industrialized before deindustrialization arrived. These populations were never economically unnecessary. Many were indispensable in practice, through informal markets, household production, and precarious services, while being denied a stable place inside the official grammar of employment, taxation, and protection.

Lagos, Cairo, Dhaka, and Jakarta have long housed populations the formal economy could not absorb at scale and the state could not fully account for. What is different in advanced economies is not the condition but the political infrastructure surrounding it: the ballot, the zoning board, the congressional hearing, the class action, the investigative press.

Populations excluded from formal absorption in the Global South were often managed through scarcity, informality, and weak provision. Their advanced-economy counterparts are managed through abundance and strong institutions. The export is upward, and the destination has more tools for converting grievance into political pressure.

None of this is grounds for optimism. It means the political consequences will be louder, more institutionally disruptive, and harder to contain through the informal mechanisms that absorbed displacement elsewhere.

The contemporary version is not machine-breaking, at least not yet. It is zoning fights, permit battles, legislative restrictions, and the slow accumulation of local political resistance to the physical plant of AI: the same permission politics that now constrains the buildout itself. The political form is democratic and procedural. The underlying logic is the same: attack the thing you can see when the thing that harmed you cannot be named or reached.

The important political implication is that this targeting, however understandable, does not address the underlying legitimacy problem.

Blocking a data center does not restore wage leverage. Restricting AI energy use does not rebuild the junior layer. Passing algorithmic accountability legislation does not answer the question of what grammar should replace the wage as the primary explanation for why people have a recognized place inside the productive order.

Infrastructure resistance is real politics.

It is not a settlement.

The Grammar That Has Not Been Built

The AI economy has produced a bilateral legitimacy crisis at its center.

Labor can no longer say “I made this” with the force it once had. Capital can no longer say “I own this and therefore deserve the full return” without the moral thinness of that claim becoming more apparent as the surplus concentrates in systems built on collective inheritance.

The surplus human lives inside that double failure.

She is the person for whom neither grammar works. The wage no longer explains her income if she has any. The ownership claim does not apply because she does not own the systems she depends on. And no new grammar has been built that can tell her, and tell the political system, why she has a recognized place inside the world that synthetic production is creating.

One direction for that grammar is inheritance: the claim that AI surplus depends on accumulated civilization, on public science, public law, collective data, shared infrastructure, and inherited knowledge, and that membership in that civilization therefore carries a claim on some portion of what it produces.

That grammar does not require treating contribution as meaningless. It requires admitting that contribution has always depended on systems no individual built alone.

This grammar has begun appearing at the edges of mainstream policy. Even some proposals from frontier AI actors now gesture toward public stakes in AI-linked assets, citizen dividends, robot taxes, or social wealth funds. The motive may be strategic positioning as much as principle. But even as positioning, the gesture concedes the structural point: AI surplus cannot be narrated only as private corporate property. The inheritance is too visible. The collective substrate is too large. The concentration is too stark. [3]

But grammars take time to become politically legible. They have to be narrated, contested, institutionalized, and eventually taken for granted before they can organize stable legitimacy. The inheritance grammar is not yet politically usable at scale. It remains too unfamiliar, too weakly institutionalized, and too easily mistaken for either charity or confiscation.

In the meantime, the containment system fills the gap.

Provision Without Membership

The obvious objection to this account is that it is too cynical. Transfers are not containment; they are what a decent society owes people the market has failed. Therapy is not containment; it is care. Education is not containment; it is opportunity. Civic participation is not containment; it is democracy doing what democracy does. On this reading, the toolkit described above is not a management system. It is the achievement of civilized society.

The objection is partly right, and it is worth conceding exactly how. None of these systems is containment by nature. Each becomes containment by substitution. A transfer becomes containment when it replaces a claim on production without creating a new claim on membership. Therapy becomes containment when it treats structurally produced dislocation as an individual adjustment problem. A credential becomes containment when it extends preparation after the destination has narrowed. Civic ritual becomes containment when people are invited into politics after the decisive acts have migrated beyond its reach.

The same institution can be care in one historical setting and containment in another. The difference is whether it restores membership or manages the absence of it.

Provision without membership is management.

Containment Is Not Stability

The error is to mistake containment for equilibrium.

A society can suppress the political consequences of a legitimacy failure for a long time. Transfers keep people fed. Platforms keep people engaged. Credentials keep people occupied. Therapeutic frameworks keep private distress from becoming shared grievance. Civic rituals maintain the grammar of participation. Infrastructure resistance gives anger a local, legal, manageable form.

But suppression is not resolution.

The condition that containment manages accumulates rather than dissolves. As more workers find that AI can do more of what once gave them leverage, as more young people find that the junior layer has thinned before any alternative pathway was built, as more professionals find that their judgment has become supervisory rather than generative, the population whose claims exceed their recognized necessity grows.

The gap between what those people were told to expect, from education, from work, from participation in a productive society, and what the system is now offering becomes harder to paper over with any combination of transfers, platforms, and therapeutic reassurance.

The political science literature on economic anxiety confirms the direction. Autor, Dorn, Hanson, and Majlesi find that trade-induced manufacturing displacement increased political polarization, with stronger effects in more exposed congressional districts. [4] Case and Deaton document rising mortality among working-class Americans without college degrees as manufacturing employment collapsed and the social structure that organized those communities deteriorated. [5]

These are not studies of AI.

They are studies of what happens when the economy stops explaining people.

AI extends that condition more broadly, more unevenly, and faster. Containment does not resolve the surplus human problem. It defers it while the conditions compound.

The political system that manages this condition successfully for a decade may find that it has created larger and more volatile problems in the next one. The anger of people the system has learned to operate around does not disappear because it is absorbed by the attention economy or channeled into zoning disputes. It accumulates grievance, loses specific form, and waits for a political framework capable of giving it direction.

That framework will not necessarily be constructive.

The surplus human is not someone without needs.

She is someone whose claims remain after the system has learned to operate around her. And a civilization that responds to that condition only with management, without explanation, without a grammar that makes her presence intelligible inside the world being built, is not stable.

It is delayed.

More people each year stand at the edge of a productive system that functions without fully needing them, holding claims the system has not yet found a language to honor or a mechanism to redeem.

Containment keeps them standing there.

It does not answer the question of why they belong.

Containment is what a civilization builds when it still needs obedience from people it no longer knows how to need.

Notes

[1] On the Prince William Digital Gateway: the Prince William County Board of Supervisors approved the rezonings in December 2023 after a 27-hour public hearing; Circuit Court Judge Kimberly Irving voided the rezonings in August 2025 for defective public notice; the Virginia Court of Appeals upheld that ruling on March 31, 2026; and the board voted on April 14, 2026 not to appeal further. Prince William Times, “Prince William County drops legal fight for PW Digital Gateway,” April 2026, https://www.princewilliamtimes.com/localnews/breaking-prince-william-county-drops-legal-fight-for-pw-digital-gateway/article_e7152a10-cbf9-4a7a-80ef-55fa9e13770b.html. On collapsing public support: a Washington Post-Schar School poll published in April 2026 found 35 percent of Virginia voters comfortable with a new data center in their community, down from 69 percent in 2023. Data Center Watch, a tracker run by 10a Labs, recorded 48 data center projects blocked or delayed across the United States in 2025, representing roughly $156 billion in planned development. On Tucson: AZ Luminaria, “Tucson City Council rejects Project Blue data center amid intense community pressure,” August 6, 2025, https://azluminaria.org/2025/08/06/tucson-city-council-rejects-project-blue-amid-intense-community-pressure/. On federal legislation: Clean Cloud Act of 2025, S. 1475, 119th Congress, introduced April 10, 2025 by Senators Sheldon Whitehouse and John Fetterman, https://www.congress.gov/bill/119th-congress/senate-bill/1475; AI-Related Job Impacts Clarity Act, S. 3108, 119th Congress, introduced November 2025 by Senators Josh Hawley and Mark Warner, https://www.congress.gov/bill/119th-congress/senate-bill/3108.

[2] The classic account of Luddite machine-breaking as skilled workers’ rational response to capital-sponsored deskilling and displacement is E. P. Thompson, The Making of the English Working Class (1963), Part Two. Thompson establishes that the frame workers were targeting was not machinery in general but the specific equipment being used to undercut their trades. On surplus labor management in the Global South: Mike Davis, Planet of Slums (Verso, 2006) documents how cities across Africa, Asia, and Latin America absorbed populations that formal labor markets could not employ, through informal economies, NGO provision, and the containment logic of urban peripheries. The ILO’s World Employment and Social Outlook reports consistently find that over 60 percent of employment in low-income countries is informal, representing a long-running structural condition of economic non-necessity managed without the institutional infrastructure of advanced welfare states.

[3] For the frontier-actor gesture: Sam Altman, “Moore’s Law for Everything,” March 2021, proposing an American Equity Fund financed by taxes on large companies and land, https://moores.samaltman.com; Cullen O’Keefe et al., “The Windfall Clause: Distributing the Benefits of AI for the Common Good,” Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (2020). The specific mechanisms in the broader family of proposals vary, from social wealth funds to robot taxes to citizen dividends, but their shared premise is that AI-generated surplus cannot be narrated only as private corporate property when the underlying capability depends on public science, collective data, shared infrastructure, and inherited civilization.

[4] David Autor, David Dorn, Gordon Hanson, and Kaveh Majlesi, “Importing Political Polarization? The Electoral Consequences of Rising Trade Exposure,” American Economic Review 110, no. 10 (2020): 3139-3183.

[5] Anne Case and Angus Deaton, Deaths of Despair and the Future of Capitalism (Princeton University Press, 2020).