Picture a household in a mid-sized city. This is an illustration, not a case study, and it names no one. Rent rises. A work schedule shrinks. A car repair lands in the same week as a utility bill. The adults lie awake doing arithmetic that never balances, and every small surprise feels like a threat.

Now picture the same month with one change. A steady deposit arrives on a known day, and the household decides how to spend it. Nothing about the person has been relabeled. The arithmetic has simply changed. This article asks how much public distress looks like the first month, and what the best studies of income, housing security, and personal control can and cannot tell us. The answer is mixed.

Szasz and the Idea of Problems in Living

Thomas Szasz, a psychiatrist at SUNY Upstate Medical Center in Syracuse, argued in a 1960 paper and then in his 1961 book The Myth of Mental Illness that much of what is called mental illness is better understood as problems in living: conflicts over money, work, family, duty, and loss. He held that a disease label can cover over the human conflict beneath it.

Szasz also built a case about liberty. Adults should keep authority over their own lives, consent matters, and coercion in the name of care is still coercion. Many clinicians and researchers reject parts of his position, and income studies do not prove him right. This article borrows his question. Before we reach for a label, have we looked at the money, the housing, and the freedom a person actually has? That is a question for economics and ethics.

What the Income Studies Found

Economists like randomized designs because they separate cause from coincidence. Each study below measures something slightly different, and most rely on self-reports.

Read together, they show a pattern. Relief often arrives early, and what happens next depends on design, duration, and context.

Oregon: Insurance as Financial Security

In 2008 Oregon drew names by lottery for a limited number of Medicaid slots for low income, uninsured adults. That created a randomized comparison. Baicker, Finkelstein, and colleagues reported in the New England Journal of Medicine in 2013 that coverage cut the chance of screening positive for depression symptoms on a questionnaire by 9.15 percentage points, from a control rate of 30 percent. It also nearly eliminated catastrophic out-of-pocket medical spending.

This was insurance, not an income check, so it is a cousin of the income evidence and not a twin. The Szaszian reading is natural all the same. Removing the threat of ruinous bills coincided with fewer people reporting low mood. The confidence interval was wide, running from 1.6 to 16.7 points.

Kenya: Cash and Psychological Wellbeing

Haushofer and Shapiro studied unconditional transfers from GiveDirectly to poor households in rural Kenya, averaging 709 dollars in purchasing power terms. About three years later, recipients held 40 percent more assets than households in distant villages, roughly 422 dollars, or 60 percent of the original transfer. Within villages, recipients also showed higher consumption, food security, and psychological wellbeing, with gains similar to those at nine months.

Haushofer and Fehr, in a 2014 Science review, argued that poverty raises stress and negative feelings, which can push people toward short-term choices and make escape harder. The Kenya authors were careful, though. Other outcomes in the across-village comparison were not significant, possibly because of low power and spillovers, and spillover households had lower consumption and food security. Rural Kenya also differs sharply from the United States.

Finland, Ontario, and Manitoba

Finland’s social insurance agency, Kela, paid 2,000 unemployed people 560 euros a month in 2017 and 2018. In an end-of-experiment phone survey, recipients reported higher life satisfaction and less mental strain, depression, sadness, and loneliness than the control group. Of 3,970 people reached, only 1,633 agreed to take part, which leaves room for doubt about who answered.

Ontario ran a pilot in Hamilton, Brantford, Thunder Bay, and Lindsay, then cancelled it in 2018. A McMaster and Ryerson report surveyed 217 former recipients in the Hamilton area. Of those, 83 percent said they felt stressed or anxious less often, and 81 percent said they felt more self-confident. That was a voluntary online survey with no control group. Evelyn Forget reanalyzed Manitoba’s Mincome experiment of 1974 to 1979 and found that hospital separations fell 8.5 percent in Dauphin against matched controls, driven mainly by accidents, injuries, and mental health admissions. That was a matched comparison, not a lottery.

Stockton and Baby’s First Years

The Stockton Economic Empowerment Demonstration gave 125 residents 500 dollars a month for 24 months. The published randomized analysis in the Journal of Urban Health compared 131 recipients, counting alternates, with 200 controls. It found lower psychological distress, better energy, and more agency over work and caregiving. The distress effect faded during the pandemic or after payments ended.

Baby’s First Years randomized 1,000 low-income mothers in four American metropolitan areas to 333 dollars or 20 dollars a month. The 2022 PNAS study found infant brain activity differences at about age one, though significance varied with analytic choices. The 2025 four-year results from Noble, Magnuson, Duncan, and colleagues found no statistically significant effects on four preregistered primary outcomes: language, executive function, social-emotional problems, and high-frequency brain activity. That is a null result.

Causal Direction and the Counterevidence

A fair reading needs two more steps. The first is the question of which way causation runs. The second is the strongest evidence against an easy story.

Neither step erases the case for looking at income first. Both change how confidently it should be stated.

Ridley and Colleagues on Causal Direction

Ridley, Rao, Schilbach, and Patel reviewed the evidence on poverty, depression, and anxiety in Science in 2020. They found causation running both ways. Income shocks and antipoverty programs shift mental health, and mental health problems can also reduce work and income. Their mechanisms include income volatility, trauma, and social status or isolation.

For a Szaszian lens this matters twice. The two-way finding warns against the slogan that poverty simply causes illness. The listed mechanisms are mostly conditions of life, which fits the problems in living idea. The authors also argued that policy should combine poverty reduction with mental health care. This article gives no guidance on care, which is a matter for individuals and the professionals they choose.

Fading Effects and Null Results

The strongest counterevidence is the OpenResearch trial. Miller, Rhodes, Bartik, Broockman, Krause, and Vivalt gave 1,000 low-income adults aged 21 to 40 1,000 dollars a month for three years, against 50 dollars a month for 2,000 controls. They found large but short-lived improvements in stress. After the first year there was no improvement in mental health, and the authors could rule out small effects on many outcomes.

Other cautions pile up. Stockton was small. Most distress measures are self-reported. A fixed-term pilot differs from a permanent floor, so recipients may respond differently to money they know will stop. Cash also does not repair grief, conflict, or loss.

Delivery: Credit Unions and Nonprofits

Most pilots above paid through nonprofit partners and prepaid cards. One documented Santa Clara County pilot for foster youth paid 1,000 dollars a month through a Community Financial Resources card, while Excite Credit Union staff gave financial mentoring and credit-building support.

That hints at a design question. A payment that arrives like an ordinary deposit, in an institution members own, may carry less stigma and more local help than a distant agency. A companion IdeaRiff piece covers the model: basic income through credit unions and semi-private delivery.

A Design Idea, Not a Proven Program

No study above compares credit union delivery with a state check. The trials test cash, not the pipe. Credit unions also have fields of membership, so some people would have no door. A workable design would need a public fallback, fee caps, and public audit.

Real evidence would need a pilot that randomizes delivery channel, publishes null results, and tracks timeliness, fees, and the usual outcomes. Until then the credit union role is a hypothesis about administration.

Counterarguments

First, money is only part of life. Severe distress, grief, and conflict persist after income rises, and critics of Szasz hold that serious conditions are real. People do suffer, and this article offers no view on what any individual should do about their own health. Second, cost and scale. A pilot of a few hundred people is not a nation. Ontario’s government said a project of 4,000 people fell short of an answer for a province with almost two million people in poverty.

Third, work. Finland found small employment effects, and most employed Ontario respondents kept their jobs, but small pilots cannot settle long-run labor questions. Fourth, liberty cuts both ways. A state-run floor could bring conditions and surveillance, which a Szaszian should fear. That is one reason delivery design deserves testing.

What Readers Can Do Now

These are civic steps, not clinical ones, about how a community arranges income, housing, and choice.

• Read the studies themselves, including the null results. Abstracts are free at NBER, PNAS, and Science.
• Ask local officials whether an income pilot is planned and whether all results, including null ones, will be published.
• Support credit unions and nonprofits that already help people with accounts, tax filing, and benefit paperwork.
• Back local housing supports, eviction prevention, and income supports through public comment and voting.
• Ask that any program use neutral enrollment language, fee caps, and strict limits on data use.

Each step is small. Together they move a debate about labels toward one about conditions people can change.

Conclusion: Income, Housing, and Agency Before Labels

The evidence supports a modest claim. When people gain income and some control over their circumstances, reported distress often eases, sometimes sharply, sometimes briefly, and sometimes not at all. Oregon, Kenya, Finland, Ontario, Manitoba, and Stockton lean one way. Baby’s First Years and OpenResearch lean the other on key outcomes, and Ridley and colleagues show causation runs both ways.

A society that wants less distress should look first at income, housing, and agency, because those are things a community can change without overriding anyone. That is the practical heart of the problems in living idea. The credit union and nonprofit channel remains a design to test, and the honest course is to test it in the open.

This article is general information, not medical, financial, or legal advice.

This short scene is an illustration, not a prediction or a data point. Picture a weekday in 2040. A driverless car waits at the curb, and a home battery charged by midday sun runs the kettle. An AI assistant has drafted the paperwork for review, and a weekly injection keeps a relative heart risk in check.

Now notice what the picture leaves out. No flying taxi lands on the roof. No swarm of microscopic robots repairs an artery. The rent is still due, and the electric bill still stings. The rest of this article asks which parts of that morning rest on evidence, and which belong to fiction.

How to Read the Ratings

Each prediction gets one label for the window from 2035 to 2040. Likely means current deployment data and trial timelines already point there. Possible means a credible path exists, but it depends on a result nobody has yet. Unlikely means the known obstacles outweigh the known progress.

Each rating also says what would change it. Sources are named so you can check them, and any number I could not verify was left out.

Machines That Move

Self-Driving Cars

Robotaxis are real. In September 2026, TechCrunch reported that Waymo serves 15 U.S. cities and averages 500,000 paid rides each week, up from 50,000 in May 2024. Its fleet is roughly 4,000 vehicles, about 80 percent in California and Texas. Waymo reports 271.3 million rider-only miles through June 2026, with 82 percent fewer injury crashes and 95 percent fewer serious-injury-or-worse crashes than a human benchmark. These are company figures built on public crash reports.

The obstacles are rare odd situations, city-by-city permits, vehicle cost, and public trust. Regulators are investigating Waymo cars passing school buses, and tariffs raise the cost of its Chinese-built Ojai minivan. Rating: robotaxi service in many large metro areas is likely. A personal car that drives itself everywhere, including snow and rural roads, is unlikely. Safety rates holding as miles grow, or a cluster of serious failures, would move this.

Electric Air Taxis

As of early October 2026, neither Joby nor Archer holds an FAA type certificate. Joby reports it is in the fifth and final stage. Archer closed Phase 3 of four in April 2026. Dates have slipped before: in 2021 a former Archer co-CEO told CNBC the company was working toward certification by 2024. In Europe, EASA applies a special condition called SC-VTOL, and Vertical Aerospace targets UK certification in 2029.

Scale is the second limit. Joby says its Marina plant can build up to 24 aircraft per year, and its Dayton plant is designed for up to 500. Even the larger figure yields only a few thousand aircraft in a decade, and Joby has not disclosed a price. Rating: certified service on a few routes is likely, while mass commuting is unlikely. Certificates, output above several hundred aircraft yearly, and taxi-level fares would change that.

Robots and Automation

The International Federation of Robotics counted 542,076 industrial robots installed in 2024, with 4,663,773 in operation. Amazon announced in 2025 that it had deployed its one millionth robot across more than 300 facilities, mostly mobile units rather than humanoids. It also said a newer Louisiana site needs 30 percent more staff in maintenance and engineering roles.

The realistic labor effect is task shifting. Heavy lifting and sorting move to machines, while demand grows for technicians and planners. Rating: more warehouse and factory automation is likely. Humanoids at large scale in logistics are possible. Humanoids handling varied chores in ordinary homes are unlikely. Independent price and safety data from real worksites would move these.

Intelligence and Energy

AI Progress and Its Limits

The Stanford AI Index 2026 says generative AI reached 53 percent population adoption within three years. It also describes a jagged frontier: one model won gold at the International Mathematical Olympiad, yet the best model reads analog clocks correctly only 50.1 percent of the time. AI agents still fail roughly one attempt in three on structured tests.

METR measures the length of tasks that models finish at 50 percent reliability. That length doubled about every seven months from 2019 to 2025, and every 131 days since 2023. Claude Opus 4.5 scored about 320 minutes, with a wide interval of 170 to 729. Rating: AI as a routine office tool is likely. Dependable autonomous work lasting days is possible. Trust without human review in medicine or safety-critical settings is unlikely. Gains at 90 percent reliability, and fewer documented incidents (362 in 2025, up from 233), would change my view.

Cheap Energy, Lagging Bills

Generation keeps getting cheaper, yet bills have not followed. The IEA reports that from 2019 to 2024 household electricity prices rose 36 percent in the European Union and 26 percent in the United States. Net earnings for a two-earner family with two children rose 25 and 23 percent. Networks, taxes, and fees take a large share of each bill. I covered this gap in cheap solar, batteries, and rising electric bills.

Batteries help: the IEA links fewer negative-price hours in California and Texas in 2025 to growing storage. Rating: abundant cheap generation in sunny regions is likely. Household bills that fall in real terms are possible, but they depend on wires, tariffs, and rate design more than on panels. Reform of network charges would change this.

Aging Medicine: The Main Thread

What Could Exist by 2040

The strongest human evidence concerns diseases of aging, not aging itself. The SELECT trial enrolled 17,604 adults aged 45 or older with heart disease and excess weight but no diabetes. Weekly semaglutide cut the combined rate of cardiovascular death, heart attack, and stroke by 20 percent (hazard ratio 0.80) over a mean of 39.8 months. That is a real healthspan gain for a defined group. It does not show that aging slowed.

Other tools remain early. The TAME metformin trial was designed for 3,000 adults aged 65 to 79 over six years. As of September 2026, trackers found no registry record, and the sponsor still seeks donors. Reported costs run $45 to $70 million. A 2023 senolytic trial in five people with Alzheimer disease found safety but no significant cognitive change, and a 2024 trial in 60 women found no overall drop in bone resorption. Epigenetic clocks track risk across populations, yet reviews say they are not validated for individual decisions or as trial endpoints.

The Counterevidence and the Rating

Olshansky and colleagues, writing in Nature Aging in 2024, studied the eight longest-lived countries plus Hong Kong and the United States from 1990 to 2019. Gains in life expectancy slowed. Survival to age 100 is unlikely to exceed 15 percent for women and 5 percent for men. Repeating the 30.2 percent drop in old-age mortality seen in that period would add only 2.5 years of life expectancy at birth. The authors call these limits not brick walls, since a drug that slows biological aging could change them.

Rating: drugs that delay specific age-related diseases in targeted groups are possible by 2040. An approved drug for aging itself is unlikely, because TAME has not started and needs six years. Large lifespan extension is unlikely. A funded, registered trial, a clock validated against real outcomes, or a strong randomized human result would change that. For the economics, see the longevity dividend.

Nanobots and Molecular Medicine

DNA nanotechnology is a vivid research field. Reviews from 2025 list the hurdles to human use: enzymes that degrade DNA structures, poor stability in blood, immune reactions, weak targeting, and costly manufacturing. One developer in the field states it has no approved products and no open clinical trials. Nucleic-acid nanoparticle drugs have reached early trials, but they are not robots.

Rating: more precise targeted drug delivery is likely. A first DNA nanodevice therapy tested in people is possible. General-purpose medical nanobots that patrol the body are unlikely. Human safety data and clinical-scale manufacturing would change this.

Abundance, With Receipts

Harvard JCHS reports the median home price was 4.7 times median income in 2025, down from a 5.0 peak in 2022 but far above the 1990s average of 3.2. Child Care Aware puts average child care at $13,184 in 2025, which is 10 percent of median income for two-parent households and 33 percent for single parents. The College Board lists $11,950 for public in-state tuition in 2025-26. CMS projects health spending rising from 17.6 percent of GDP in 2023 to 20.3 percent in 2033.

What gets cheap is anything made of information, energy, or factory output: software, many goods, and electricity generation. What resists is land and work that needs human time, such as care and teaching. Who shares the gains is a policy choice, as I explored in who owns AI productivity gains. Rating: cheaper goods and software are likely. Falling housing, care, and health costs as a share of income are possible. True post-scarcity is unlikely.

The Rating Summary

Each rating below applies to 2035 to 2040.

• Likely: robotaxis in many metro areas; certified eVTOL on a few routes; more automation; AI as a work tool; cheap generation; cheaper goods and software; targeted drug delivery.
• Possible: humanoids at logistics scale; days-long autonomous AI work; lower bills; healthspan drugs; a first DNA nanodevice trial; cheaper housing and care.
• Unlikely: go-anywhere personal robocars; eVTOL commuting; humanoid home helpers; unsupervised high-stakes AI; an approved aging drug; large lifespan extension; general nanobots; post-scarcity.

The pattern is plain. Software and generation move fast, while physical, regulated, and human-time systems move slowly.

What Readers Can Do Now

These steps hold up whichever ratings prove right.

• Learn to use AI tools while checking their numbers and citations.
• Build skills that pair with machines, such as maintenance, analysis, and communication.
• Protect health with proven basics, and discuss any drug with a clinician.
• Plan retirement with realistic lifespans, not 100-year assumptions.
• Ask what share of an energy bill is generation versus wires.

Also test any headline: does it name a date, a regulator, a trial, or a number? If not, wait for evidence.

Conclusion: Confidence Is a Skill

The 2040 that evidence supports is useful and uneven. Rides, software, and electricity generation improve fast. Aircraft, medicine, housing, and care improve slowly, because rules, biology, land, and human time set the pace. Aging research is the key thread, and its honest rating is a hopeful maybe for healthspan and a firm unlikely for dramatic lifespan gains.

Hold forecasts loosely, and watch certificates, registered trials, safety data, and the share of household budgets. When those numbers move, the ratings should move with them.

This article is informational forecasting, not medical, financial, or legal advice.

The short answer is that in the United States, cheap electricity gets stuck in line. It waits years for permission to connect, it waits for transmission lines that take a decade to build, and it arrives at a utility whose profits grow when it spends more on poles and wires. The hardware got cheap in factories. The rules that move it to your house did not.

The Principle: Cheap Power Has to Get Through the Gate

Here is the principle stated plainly. Abundance arrives first where costs are set by manufacturing, and last where costs are set by permits, queues, and regulated monopolies. Panels and battery cells ride a factory learning curve. The approvals, the transmission lines, and the rate formulas that decide what reaches a household do not ride that curve at all, and in the United States they have become the bottleneck.

This explains why a household can read headlines about record low prices and still open a higher bill, and it points toward which fixes actually matter.

The factory side keeps getting cheaper

In its December 2025 survey, BloombergNEF reported that average lithium ion battery packs fell to $108 per kilowatt hour, and packs for stationary storage averaged $70, a 45 percent drop in one year. Lazard’s 2025 cost analysis put unsubsidized utility scale solar at $38 to $78 per megawatt hour, with average costs down 84 percent since 2009.

Meanwhile, the U.S. Energy Information Administration shows the average residential price for the first half of 2026 at 18.16 cents per kilowatt hour, compared with 16.91 cents a year earlier. Solar at scale costs roughly 4 to 8 cents to produce. The gap between those numbers is where the American problem lives.

The line to connect

Before a new solar farm or battery can sell power, it has to pass an interconnection study run by the grid operator. Berkeley Lab’s Queued Up research found that at the end of 2024, about 2,290 gigawatts of generation and storage were waiting in these queues. The typical project built in 2024 took 55 months from request to operation, compared with 22 months in 2008. The 2026 update puts the median above five years. Only 13 percent of capacity that applied from 2000 to 2019 had been built by the end of 2024.

Much of that waiting is about transmission. Where lines are full, newcomers can be billed for upgrades, and many withdraw. Cheap electrons that never connect cannot lower anyone’s bill.

The rooftop permit tax

The same pattern shows up on house roofs. A 2025 Boston University Institute for Global Sustainability comparison put average Australian residential solar at about $0.82 per watt against roughly $2.53 in the United States, and credited Australia’s advantage largely to streamlined permits and near automatic utility approval for small systems. The advocacy group Permit Power estimates a median U.S. price of $28,000 for a 7 kilowatt system, compared with about $4,000 in Australia. Those are advocacy figures, but the direction of the gap is consistent across studies.

Sales costs, labor, and margins are higher here too. Yet slow, inconsistent approvals inflate each of those, because installers pay staff to chase permits and absorb canceled jobs.

A business model built on wires

The last gate is how most Americans buy power. Investor owned utilities operate under cost of service regulation: state commissions let them earn an authorized return on capital they add to their rate base, while ordinary operating savings earn nothing. RMI and other analysts describe the resulting bias toward capital spending. Berkeley Lab’s 2025 retail price study found that from 2019 to 2024, inflation adjusted utility spending on distribution and transmission rose while spending on generation declined.

Some of that spending is necessary. In California, Berkeley Lab found wildfire related costs made up 17 percent of large utility revenue requirements by mid 2024, up from 1.7 percent in 2019. But a structure that rewards building over saving gives utilities little reason to chase cheaper options such as batteries that defer new lines.

Where Policy Has Already Helped Households

The encouraging part is that each gate can be opened, and some American places are already doing it. These examples are proof that the bottleneck is a policy choice rather than a law of physics.

The common thread is removing friction so cheap hardware can compete, rather than waiting for another breakthrough.

Texas builds fast

Texas runs its own grid, ERCOT, with a competitive market and a connection process that lets projects plug in relatively quickly. The result has been a building boom. The EIA reports that ERCOT solar averaged 24 gigawatts at midday in summer 2025, double the 2023 level, while the gas share at midday fell from 50 percent to 37 percent. Batteries supplied an average of 4 gigawatts in the 8 p.m. hour as the sun set.

The Federal Reserve Bank of Dallas found that battery capacity reached 19 gigawatt hours, more than three times the end of 2023, and that this reduced the intensity of wholesale price spikes. ERCOT’s own summer 2025 review said energy and reserve costs were much lower than in summer 2023, crediting battery supply and milder weather.

Instant permits and plug in panels

SolarAPP+, a free permitting tool built by a federal laboratory, approves code compliant home solar and battery systems in minutes. It now serves more than 350 jurisdictions. A 2024 performance review found that typical SolarAPP+ projects were permitted and inspected 12 business days sooner than traditional ones, eliminating more than 100,000 business days of delay in a single year.

Utah went further for renters. Its 2025 law, H.B. 340, lets residents use certified plug in solar devices up to 1,200 watts without a utility interconnection agreement or fees. It is ownership without a roof or a permit.

Pakistan as a warning and a lesson

Pakistan imported about 17 gigawatts of panels in 2024, according to Ember data reported by CNN, as households fled expensive, unreliable grid power.

Families cut bills and gained power during blackouts, but grid demand fell and the Japan Times reported fears that those unable to afford panels would carry a weakened network. Cheap hardware spreads fast, and someone still pays for the wires.

The Frank Counterarguments

Rooftop solar remains far pricier per unit than utility scale solar. A December 2025 review of Lazard’s figures cited $122 to $284 per megawatt hour for residential systems. In a state with cheap grid power, a home system may never pay off.

Faster connections and lighter permitting also carry real tradeoffs. Interconnection studies exist partly to protect reliability, and Texas learned in the 2021 winter storm that a lightly regulated grid can fail badly. And Berkeley Lab found that net metered rooftop solar can shift fixed costs onto households without panels.

What abundance will not change

Cheap solar will not make wires free or remove storm and fire risk. For most households, a grid connection stays the cheapest backup, so part of the bill will remain.

The open question is who captures the savings. If they vanish into profits, fees, or poor rate design, households see little. That is why ownership matters.

Owning a Piece of the Power

The surest way to share in cheap energy is to own part of it, or to belong to an organization that owns it on your behalf. Americans have more options here than most realize.

Each model offers a different degree of control, so read the terms before joining.

Community ownership models

Rural electric cooperatives are member owned and nonprofit. According to NRECA, they serve 42 million people across 56 percent of the nation’s land, and members elect the board. Municipal utilities are owned by their cities and answer to local voters rather than shareholders. Both can be pushed to build local solar and storage.

Community solar lets renters subscribe to a shared array. Under Illinois Solar for All, income eligible participants pay nothing upfront and their fees cannot exceed 50 percent of the value of their bill credits. Note the difference between a subscription, which buys credits, and true ownership, where a co op or community group owns the panels and keeps the returns.

Getting paid for a home battery

Virtual power plants pay battery owners to share stored energy at peak hours. In a July 2025 California test analyzed by The Brattle Group, more than 100,000 home batteries delivered about 535 megawatts during the evening peak, comparable to a mid sized power plant.

Payments vary widely, and some programs tie you to one installer or lease. Owning the battery outright keeps you free to switch.

What Readers Can Do Now

Start with the parts of the system you can actually influence, then work outward.

• Read your bill line by line. Separate energy, delivery, and fixed charges, and compare your price with your state average from the EIA.
• Find out who owns your utility. If you belong to a co op or municipal utility, attend meetings and vote in board elections.
• Ask your city about SolarAPP+. Instant permitting costs local governments nothing and lowers installer costs.
• Compare community solar terms. Look for guaranteed savings, no cancellation fees, and the option of community ownership.
• Check virtual power plant programs before buying a battery, and prefer ownership over long leases.
• Cut waste first. Insulation and efficient appliances shrink any system you later buy.
• Comment on rate cases. State commissions take public input on how much utilities may earn on new spending.

None of these steps requires a perfect future. Each one helps.

Opening the Gates to the Kitchen Table

Elena’s two pieces of paper are both telling the truth. The battery is cheaper than ever, and her bill is higher. What sits between them is a set of American gates: long queues, slow lines, tangled permits, and a rate formula that rewards spending.

That is cause for hope, because gates are built by people and can be opened by people. Texas, Utah, Illinois, and hundreds of permitting offices are already proving it. When households own more of the power and demand faster, fairer rules, the abundance made in factories can finally reach the kitchen table.

This article is for general information only and is not financial advice. Energy prices, incentives, and utility rules vary by location and change over time, so consult current local sources and qualified professionals before making purchasing decisions.

Dana has plenty of company. According to the Caregiving in the US 2025 report from AARP and the National Alliance for Caregiving, 63 million American adults provided ongoing care to someone with a medical condition or disability in the past year. That is almost one in four adults, and it is about 45 percent more than in 2015. Behind that number sits a simple fact that our health systems rarely say out loud: the main driver of most chronic disease in later life is aging itself.

This article makes one argument. Adding healthy years to human life is among the most valuable things a society can do, and it is worth treating as public policy rather than private luck. The evidence for that claim is strong in some places and thin in others, and both sides deserve an honest look.

The Gap Between Living Long and Living Well

Two numbers that tell different stories

The World Health Organization tracks two measures side by side. Life expectancy counts how many years a newborn can expect to live. Healthy life expectancy, which the WHO calls HALE, counts how many of those years are lived in full health. According to WHO data for 2021, global healthy life expectancy at birth was 61.9 years. Global life expectancy that same year was about 71.4 years. The difference, close to a decade, is the stretch of life most people spend managing illness or disability.

The same WHO table lists healthy life expectancy in the United States at 63.9 years in 2021, down 1.4 years from its 2000 level. A wealthy country with enormous medical spending has been losing healthy years. That gap is where Dana spends her Thursdays.

Why disease by disease is a slow road

Modern medicine is organized around single diseases. Cancer has its institutes, heart disease has its foundations, and dementia has its advocates. Yet the risk of nearly all of these conditions climbs steeply with age, and they tend to arrive together. Curing one often means a person lives long enough to meet the next.

Geroscience, the study of the biology of aging, takes a different approach. It asks whether slowing the shared underlying processes could delay many diseases at once. If that works even modestly, the effect would show up not as one conquered illness but as a later and shorter period of decline for millions of families.

What the Longevity Dividend Is Worth

The economists’ estimate

In 2021, economists Andrew Scott of London Business School and Martin Ellison of Oxford, together with Harvard geneticist David Sinclair, published “The economic value of targeting aging” in the journal Nature Aging. Using willingness to pay estimates drawn from existing data, they found that a slowdown in aging that raised American life expectancy by one year would be worth about 38 trillion dollars. A ten year gain would be worth about 367 trillion dollars. At a 2 percent interest rate, they calculated, the ten year gain equals about 7.2 trillion dollars a year, or roughly a third of 2019 United States GDP.

Two findings in that paper matter more than the headline totals. First, the authors concluded that compressing illness into a shorter span at the end of life is more valuable than simply adding years. Second, they showed that the value of improvement grows as we make progress, because each healthy year makes the next one more useful.

What the figures do and do not mean

These are not cash amounts sitting in a vault. They measure how much people would value extra healthy time, based on how they already trade money against risk. Still, the costs they reflect are real and familiar. Lost wages, forced early retirement, and nursing home bills all sit inside that estimate. For a household, the dividend looks less like a check and more like a parent who can still drive, cook, and live independently at eighty-five.

There is also an abundance argument here. A society where people stay healthy longer has more experienced workers, more active grandparents, more volunteers, and more time for each person to learn and contribute. Healthy longevity is one of the ways a future grows richer.

Working Models and Real Tests

The TAME trial and the problem of funding

The most famous test of the geroscience idea is the Targeting Aging with Metformin trial, known as TAME. According to the American Federation for Aging Research, it is designed as six year trials at 14 research institutions with more than 3,000 people aged 65 to 79. The goal is to learn whether metformin, a cheap diabetes drug used for more than sixty years, delays the onset of heart disease, cancer, and dementia together.

The trial’s story so far is instructive. Reporting on the effort notes an estimated cost of 45 to 70 million dollars, with the National Institute on Aging setting aside about 5 million. Because metformin is generic, no drug company stands to profit, so researchers have turned to philanthropists. AFAR’s own page still describes the work as raising funds to launch. A question that could affect hundreds of millions of people has waited years for an amount of money smaller than the budget of a single blockbuster film.

Aging as a target

TAME has a second purpose that may outlast its results. Regulators in the United States do not treat aging as a condition that a drug can be approved to address. The trial was designed partly to show that a composite outcome, such as time to the first of several age related diseases, can serve as a credible endpoint. If that model gains acceptance, it opens a path for many future treatments and for public funding to follow.

This is where policy matters. Markets reward patented products. They do not reliably reward cheap, open knowledge about which existing compounds, habits, or public health measures slow decline. Like open source software, that knowledge is a public good that no single buyer has enough reason to fund.

The Frank Counterarguments

Life expectancy gains have slowed

The strongest challenge comes from demography. In 2024, S. Jay Olshansky and colleagues published “Implausibility of radical life extension in humans in the twenty-first century” in Nature Aging. Studying the longest lived populations, including Japan, France, Switzerland, and South Korea, along with Hong Kong and the United States, from 1990 to 2019, they found that gains in life expectancy have decelerated. Their analysis suggests survival to age 100 is unlikely to exceed 15 percent for women and 5 percent for men this century. Old age mortality improved by an average of 30.2 percent over those three decades, and repeating that pace would add only about 2.5 years to life expectancy.

That is a serious cold shower for anyone promising near term immortality. Yet the authors themselves add an important condition. Radical extension is implausible, they write, unless the processes of biological aging can be markedly slowed. In other words, their work argues that the old toolkit of sanitation, vaccines, and disease by disease medicine has reached its limits. That is a case for testing the new approach, not a case against it.

Hype, inequality, and money

A second fair objection concerns hype. The longevity field attracts supplements with weak evidence, clinics selling unproven infusions, and confident claims built on mouse studies. Skeptics are right to demand randomized trials before anyone changes medical practice.

A third objection concerns fairness and cost. If therapies arrive as expensive patented products, the rich could gain healthy decades while others wait. Longer lives also strain pension systems built for shorter retirements. These problems are real, but they point toward design choices: public funding for trials of cheap compounds, open publication of results, and retirement rules that adjust to healthier lives. A future of abundance requires that the gains be shared, and that is a political choice as much as a scientific one.

What Readers Can Do Now

In your own household

Most of the healthy years available today come from ordinary, proven habits rather than exotic products. A few practical steps make a difference for individuals and caregivers alike.

• Talk with a doctor about the basics, such as blood pressure, physical activity, and sleep, before spending money on longevity supplements.
• Be wary of clinics and products that promise to reverse aging without published randomized trials behind them.
• If you are a caregiver, keep a single shared list of medications and appointments so that specialists see the whole person.
• Plan for longer lives by considering how work, savings, and housing might look across a retirement of twenty or thirty years.

As a citizen

Public policy follows public attention. Readers can ask elected representatives how much of the national medical research budget goes to the biology of aging compared with single diseases. They can support open, publicly funded trials of generic compounds, where the results belong to everyone. They can also push for caregiver support, since families will carry much of the load until science delivers more.

Sharing good information matters too. When a friend forwards a miracle cure, pointing them toward a trial registry or a peer reviewed study is a small act of care that slows the spread of false hope.

A Longer Season of Health

The real prize

The point of defeating aging was never to stretch out the hardest years. It is to shorten them and move them later, so that Dana’s father might have spent his seventies hiking, teaching a grandchild to fish, or simply reading his own mail. The economics say that prize is worth more than almost anything else we could buy. The demography says we will not reach it by doing more of the same.

A shared project

Seen clearly, healthy longevity is one of the most humane goals a civilization can set. It asks us to treat the time and vitality of every person as something precious and worth protecting. Science can supply the tools, but only a society that chooses to fund open research and share the results widely will turn those tools into a dividend for all.

This article is for informational purposes only and is not medical, financial, or legal advice. Please consult a qualified professional about your own situation.

That question is older than artificial intelligence, and it is the most important one about it. Technology decides how much a society can produce. Ownership decides who receives it. A machine that doubles output can make a town richer or make a handful of shareholders richer while the town stays where it was. The difference is set by contracts, laws, and habits of ownership that people chose, and can choose again.

The Principle: Productivity Is Shared Only When Ownership Is Shared

Economists often speak as if rising productivity automatically lifts wages. For about three decades after the Second World War, it roughly did. Since then, the link has weakened. The Economic Policy Institute calculates that net productivity in the United States grew 90.2 percent from 1979 to 2025, while hourly compensation for typical workers grew 33.0 percent. Before 1979, the two lines moved almost together, with productivity growing about 2.5 percent a year and compensation about 2.1 percent.

The gap went somewhere. Part of it went to higher earners within the workforce, and part went to the owners of capital. That second channel matters most for AI, because AI is capital. A model, the data centers that run it, and the software built on top of it are assets. Their returns flow to whoever holds the shares.

Who Holds the Shares Today

The Federal Reserve’s Distributional Financial Accounts show how concentrated those holdings are. In the second quarter of 2026, the wealthiest 1 percent of American households held 50.9 percent of all corporate equities and mutual fund shares. In mid 2025, the top 0.1 percent alone held about 24 percent. Half of the stock market, in other words, belongs to one household in a hundred.

If AI raises corporate profits and the ownership map stays as it is, the arithmetic is predictable. Roughly half of the gain in stock value lands with the top 1 percent before anyone else sees a dollar.

Why AI Could Widen the Gap Faster

Earlier waves of automation needed large numbers of workers to build, install, and maintain the machines, which spread some of the gains through wages. Software scales differently. Once a model is trained, serving one more customer costs very little. That cost structure favors a small number of firms with large fixed investments and very wide reach, and those firms are mostly owned by the same concentrated group of shareholders.

None of this means that workers gain nothing. Cheaper services help everyone who buys them. But a household budget is dominated by rent, health insurance, childcare, and tuition, and those costs have tended to rise faster than general inflation for years. A free chatbot does little for a family whose rent consumes a third of its income.

Three Working Models of Broad Ownership

The encouraging part of this story is that broad ownership is already real. It exists at scale, in ordinary places, and it has decades of data behind it. Three models stand out because they are tested, legal today, and compatible with a market economy.

Each has strengths and weaknesses. Looking at them honestly is more useful than treating any one as a cure.

Employee Ownership

The National Center for Employee Ownership, analyzing Department of Labor filings, reports 6,609 employee stock ownership plans in the United States for plan year 2023, covering 15.1 million participants, of whom about 11 million were active employees. Those plans held more than 2 trillion dollars in total assets and paid over 166 billion dollars to participants in 2023. Privately held ESOP companies, which are common in manufacturing, engineering, and construction, have been growing in number for five straight filing years.

An employee owned firm that adopts AI tools faces a different calculation than a conventional one. When the claims adjuster’s team becomes more productive, the extra profit raises the value of shares the team itself holds. Employee ownership does not stop hard decisions, but it changes who benefits from making them well.

Cooperatives

Cooperatives take the idea further by giving each member one vote, regardless of how much capital they contribute. Credit unions, rural electric cooperatives, and agricultural co-ops already serve tens of millions of Americans. The model has a natural fit with digital services, where a cooperative of freelancers, drivers, or small sellers could own the platform that matches them with customers instead of paying a large commission to an outside owner.

The challenge for cooperatives is capital. Training frontier AI models costs enormous sums, and co-ops have historically struggled to raise money because they cannot offer outside investors control. The realistic path is for cooperatives to own the application layer, the tools and services built on open or licensed models, where costs are far lower and local knowledge matters more than raw computing power.

Public Dividends

The third model treats a shared resource as common property and pays every resident a share of its returns. Alaska has done this with oil revenue since 1982. The Alaska Permanent Fund holds more than 86 billion dollars, according to Alaska Public Media, and pays an annual dividend to more than 600,000 eligible residents. The 2024 dividend was 1,702 dollars per person. The 2025 dividend was 1,000 dollars.

That drop is the most instructive fact in the whole story. The Associated Press reported that under the fund’s older formula, residents would have received about 3,800 dollars in 2025. Lawmakers reduced the payment because oil revenue fell and the state needed money for schools and other services. A public dividend is only as durable as the political rules that protect it. When those rules are loose, the dividend becomes a budget line that competes with every other budget line.

Facing the Counterarguments

A fair case for broad ownership has to take its critics seriously. Several objections carry real weight, and households would feel the consequences if they were ignored.

The first objection is that markets already let anyone buy shares. That is true in principle. In practice, a family spending most of its income on rent, groceries, and insurance has little left to invest, and the compounding returns of ownership go to those who started with capital. Open access to a market is a different thing from an actual stake in it.

Risk Concentration and Inflation

A second objection deserves careful attention. Employee ownership can concentrate risk. A worker whose wages and savings both depend on one company can lose both at once if the company fails. Sound ESOP design includes diversification rights as workers approach retirement, and responsible plans encourage savings outside the firm as well.

A third objection concerns inflation. If public dividends were funded by printing money or by large deficits, they could raise prices and wipe out their own value. A dividend paid from the real returns of an invested fund, as in Alaska, avoids that problem because it distributes income that already exists. The source of the money matters as much as the size of the check.

Wages Still Matter

A final caution is that ownership is a complement to wages, never a replacement. A dividend of a thousand dollars a year is helpful, and it does not pay for a hospital stay or a year of college. Policies that keep labor markets tight, protect bargaining, and lower the cost of housing and healthcare remain essential. Ownership adds a second income stream to the household. It does not excuse neglect of the first.

Honesty also requires admitting uncertainty about AI itself. Economists disagree sharply about how large its productivity effects will be and how fast they will arrive. If the gains are modest, the stakes of ownership are modest too. If they are large, the ownership question becomes one of the defining economic issues of the century. Preparing for the larger case costs relatively little.

What Readers Can Do Now

Most people will never set national policy, but ownership decisions happen at every scale, and many of them are within reach.

• Ask about ownership at work. Owners of small and midsize businesses who are nearing retirement often sell to private equity because nobody suggested an alternative. Selling to an ESOP or a worker cooperative is a legal, established option, and resources from the National Center for Employee Ownership explain how it works.
• Use member owned institutions. Credit unions and cooperatives return surplus to their members. Moving everyday banking and purchasing to them keeps more value in the community.
• Support durable public funds. When a city, state, or country considers a sovereign or community fund, the key questions are whether the principal is protected and whether the dividend formula is written into law rather than left to annual budget fights.
• Build small with open tools. Open models and open source software lower the cost of starting a cooperative or employee owned service business, which places the application layer of AI within reach of ordinary groups.

None of these steps requires waiting for a breakthrough. Each one shifts a small piece of future AI income toward people who would otherwise receive none of it.

Abundance Needs an Address

Back at the insurance office, the software will keep improving, and the adjuster’s team will keep closing more claims. Whether that progress shows up as a larger dividend on her retirement statement, a shorter work week, or a bonus for a shareholder she will never meet depends on a decision about ownership. That decision is being made now, quietly, in board rooms, statehouses, and the succession plans of family businesses.

Post-scarcity is a question of distribution as much as production. A world that can make almost anything cheaply can still leave most people anxious about rent if the income from that capacity flows to a narrow group. The tools to spread it already exist: employee ownership, cooperatives, and public funds with strong rules. The principle is plain enough to carry into any conversation. Abundance reaches everyone only when everyone owns a piece of the machines that create it.

This article is informational and is not financial or legal advice.

That is the trouble. For decades, the basic income debate ran on hunches. Supporters pictured freedom and new businesses. Critics pictured people quitting work. Then the experiments arrived, with control groups, administrative records, and careful researchers. The results were mixed. Some were encouraging, and some were plainly disappointing.

The honest move is to read all of it, including the parts that hurt the case. Do that, and a clear principle emerges: cash works best as a floor and a launch pad, not as a cure for everything. A version built on that principle can be defended. A version that promises more will keep losing arguments.

The Question Every Pilot Was Really Testing

Most basic income pilots asked about many things: health, stress, education, debt, children. Yet public debate fixed on one question above all others. Will people stop working?

That focus is understandable. A basic income for a whole country would cost a great deal, and its price depends heavily on how people respond. If recipients keep working, taxes keep flowing. If many stop, the math gets hard fast.

Why the work question dominates

Work is also moral territory. Many voters believe help should go to people who try, and they worry that free money breaks the link between effort and reward. Any serious design has to answer that worry with evidence, not slogans.

Supporters have their own version of the question. They hope cash lets people leave bad jobs, train for better ones, or start something of their own. So both sides end up staring at the same numbers: employment, hours, and earnings.

What a pilot can and cannot show

A pilot can tell you how a few thousand people behave when they get money and others do not. That is valuable. It cannot tell you what happens when everyone gets money at once, when prices and wages adjust, or when the payment is permanent and taxed back from higher earners.

Keep that limit in mind. A pilot is a flashlight, not a floodlight. It shows one patch of ground clearly and leaves the rest in shadow.

The Disappointing Results Deserve a Fair Hearing

Two of the most careful studies landed with a thud for many supporters. Their findings were modest, and in places they cut against popular claims.

Dismissing them would be a mistake. A movement that only cites the studies it likes stops being trusted, and trust is the scarcest resource in any policy fight.

OpenResearch: less work, short-lived relief

The largest American study so far was run by OpenResearch in Illinois and Texas. From November 2020 to October 2023, 1,000 low-income adults aged 21 to 40 received $1,000 a month, while 2,000 people in a control group received $50 a month.

The results were sobering. According to the working paper on employment, recipients were 2 percentage points less likely to be employed and worked about 1.3 fewer hours per week. Their individual earned income, not counting the cash, fell by about $1,500 a year compared with the control group. The extra time went mostly to leisure, and the researchers found no improvement in job quality.

The health findings were also thin. A companion paper found large improvements in stress and food security during the first year, but those gains faded by the second year. Researchers saw no effect on physical health or blood test results. Recipients did use more hospital, emergency, and dental care.

Still, context matters. Employment rose for both groups during the study. The control group simply rose a little faster. And including the transfer, recipients ended up with household income roughly $6,100 higher per year than the control group. Cash did not make people idle. It made them slightly less eager to add hours.

Finland: a small nudge, muddied by a policy change

Finland ran a national experiment in 2017 and 2018. The government paid 2,000 randomly selected unemployed people 560 euros a month, tax free, whether or not they found work. The idea was to remove the trap where taking a job means losing benefits.

The employment effect was small. Over the main measurement period, recipients worked about 6 more days than the control group, and they averaged 78 days of employment. In the first year, researchers found no statistically clear effect at all. A separate activation reform in 2018 also changed the rules for the control group, which makes the results harder to read.

Recipients did report better wellbeing. They were more satisfied with life and reported less stress, depression, and loneliness. The survey response rate was low, however, so those findings carry less weight than the employment records.

The Encouraging Results, With Their Fine Print

Other studies gave supporters real reasons for hope. But each comes with fine print that is easy to skip and important to read.

The pattern across them is worth noticing. The strongest results tend to appear when cash helps people get over a specific hump, such as a job search, a car repair, or the cost of starting a small business.

Stockton: a striking first year

The Stockton Economic Empowerment Demonstration, known as SEED, gave 125 residents $500 a month for 24 months, starting in February 2019. In the first year, full-time employment among recipients rose from 28 percent to 40 percent. In the control group, it rose from 32 percent to 37 percent. Recipients described using the money to cover gaps while they interviewed, trained, or took unpaid internships.

The fine print is large, though. The sample was small. The pandemic hit at the end of year one. A later peer-reviewed analysis found that the employment shifts, while trending in a good direction, were not statistically significant. Stockton is a hopeful story, not proof.

Kenya: when the promise lasts

The most ambitious test is in rural Kenya, run with the charity GiveDirectly and studied by economists including Tavneet Suri and Nobel laureate Abhijit Banerjee. Some villages receive about $22.50 a month for 12 years. Others received the same monthly amount for only two years, and others got a single lump sum of about $500.

The early results, published in 2023, were striking. Villages with the long-term payment saw more businesses and higher revenues. Total labor supply did not fall. It shifted from wage work toward self-employment. The lump sum did about as well or better, while the two-year monthly stream did noticeably worse, even though it had delivered the same money so far. People who knew payments would last saved through local savings groups and invested.

Kenya is a very different economy from Illinois, so the findings do not transfer directly. But the lesson about predictability and lump sums is hard to ignore.

Five Design Lessons the Evidence Supports

Put the studies side by side and some practical design choices come into focus. None of them requires believing cash is magic.

Here are five that a city, a nonprofit, a credit union network, or a voter evaluating a proposal can use today:

1. Promise a timeline people can plan around. In Kenya, the long-term promise changed behavior more than the short stream did. Short pilots may understate what a lasting program could do, and they may also overstate it. Either way, certainty is part of the product.
2. Offer a lump-sum option. Many people face one large barrier: a deposit, a used car, a license, a tool kit. Letting recipients take part of the money up front, as some savings groups already do, matches how real investment works.
3. Measure success honestly before launch. If the goal is stability and choice, say so. If a program is sold as a jobs program or a health program, the OpenResearch results will be used against it, and fairly.
4. Pair cash with the things cash cannot buy quickly. OpenResearch itself concluded that cash may have limited impact without resources such as health care and child care. Money does not build a clinic or open a daycare slot.
5. Plan for the cost of reduced hours. A modest drop in work is a real cost, and budgets should include it rather than pretend it away. Delivery through community institutions, such as the credit union model discussed elsewhere on this site, can keep administrative costs low, but it cannot erase that tradeoff.

The strongest objection, answered plainly

The toughest critique goes like this: if $36,000 over three years only buys a small drop in work and fleeting relief, why spend it? That is a fair question, and the answer depends on what you think a floor is for.

A floor is insurance. Its value shows up when things go wrong: a layoff, an illness, a breakup, a sudden bill. Pilots that measure average outcomes in calm periods can miss that value, the same way fire insurance looks wasteful in a year without fires. That does not settle the debate, but it explains why averages alone cannot.

Where automation changes the math

The case for a floor grows stronger if AI and automation reduce demand for some kinds of labor. In that world, a small drop in hours matters less, and a reliable floor matters more. Today, though, that future is a forecast, not a measurement.

The sensible approach is to build delivery systems, savings tools, and evidence now, at modest scale, so that society is ready if displacement accelerates. Waiting for a crisis to design a program is how bad programs get built.

Conclusion: Judge Cash by What It Can Actually Do

The experiments did not prove that basic income transforms lives on its own. They also did not show mass quitting. Recipients mostly spent on food, rent, and transportation, worked a little less in some places and a little more in others, and gained a stretch of breathing room that sometimes faded.

That is the evidence. It supports a humbler and sturdier case: cash is a dependable floor, and in the right design it can be a launch pad.

So when you next see a basic income proposal, ask three questions. How long is the promise? Can people take some of it as a lump sum? And what does it claim to fix? A proposal with clear answers is worth your support. One that promises everything deserves your skepticism.

This article is for general education only and is not financial, legal, or medical advice.

Nothing about this family is unusual. They are living inside two price curves at once. One curve keeps falling, and it carries screens, phones, software, solar panels, and now artificial intelligence. The other keeps rising, and it carries the bills that decide whether a household feels secure: rent, health care, college, and child care. Understanding why those curves split apart is the first step toward bending the expensive one down.

The Curve That Keeps Falling

In 1936, an aircraft engineer named Theodore Paul Wright published a paper in the Journal of the Aeronautical Sciences called “Factors Affecting the Cost of Airplanes.” He had noticed a steady pattern. Each time the total number of airplanes built doubled, the labor needed for each new plane dropped by a predictable share. Cost, he found, follows experience rather than the calendar.

That observation became known as Wright’s Law, and it describes almost every manufactured product that is built at scale. Workers learn shortcuts. Engineers redesign parts. Suppliers find cheaper materials.

Solar, Batteries, and Screens

Solar power is the textbook case. A watt of solar module capacity cost about $96 in 1976, measured in today’s dollars. By 2024 it cost about $0.26. Lithium-ion battery packs tell a similar story. BloombergNEF’s annual survey found average pack prices of roughly $790 per kilowatt-hour in 2013 and about $115 in 2024.

Consumer electronics ride the same curve. According to Bureau of Labor Statistics price data compiled by economist Mark Perry of the American Enterprise Institute, the price of televisions fell about 98 percent between January 2000 and late 2025. Computer software fell about 75 percent. Toys fell about 74 percent, and cell phone service fell about 43 percent.

AI Joins the Club

Artificial intelligence is now on the fastest version of this curve yet. The Stanford AI Index 2025 reports that the cost of querying a model performing at the level of GPT-3.5 fell more than 280-fold between November 2022 and October 2024. A task that once cost real money per request now costs a fraction of a cent.

That matters because AI is not a product sitting on a shelf. It is a tool that can be applied to other work. If cheap intelligence can reach the expensive side of the economy, it could do something TVs never could. It could lower the price of services, not just gadgets.

The Curve That Keeps Rising

Now look at the other half of Perry’s chart, which he has called the “chart of the century.” Over the same stretch from 2000 to late 2025, overall consumer prices rose about 92 percent. Hospital services rose about 275 percent. College tuition and fees rose about 196 percent. Child care and nursery school rose about 185 percent. Medical care overall rose about 129 percent, and housing rose about 111 percent.

Average hourly earnings rose about 131 percent over the same period. That number explains a great deal. Wages beat general inflation, so the typical hour of work buys more of most things than it did in 2000. Yet hospital care, tuition, and child care rose faster than wages. For those bills, an hour of work now buys less than it did a generation ago.

Why These Bills Hurt More

The categories that got cheaper are mostly things people buy occasionally. A family replaces a TV every several years. The categories that got expensive are paid every month or come in large, unavoidable lumps. Rent arrives on the first. Child care is often the second-largest bill a young family pays. A single hospital stay can erase a year of savings.

So averages can mislead. A household that rents, has small children, and carries a health condition lives almost entirely on the rising curve. Cheaper electronics are pleasant. They do not cover the gap.

What Happened to Wages

Wage stagnation is real, but the details matter. The Bureau of Labor Statistics tracks median usual weekly earnings for full-time workers, adjusted for inflation. In constant 1982 to 1984 dollars, that figure was $332 in 1979, $334 in 2000, and $374 in 2025. That is growth of about 13 percent over more than four decades, and almost all of it arrived after 2014.

Economists debate how to read this. Some point out that average hourly earnings, total compensation including benefits, and different inflation measures show stronger gains. Others note that rising health insurance premiums have absorbed much of what employers spend on workers, so take-home pay grew slowly even when total compensation grew faster. Both views can be true. Either way, the typical worker gained modestly, while the price of the biggest essentials rose sharply.

Why the Expensive Curve Will Not Bend on Its Own

The rising categories share a few traits. They are services rather than goods. They cannot be imported. They depend on land or on licensed people. And they often face limited competition. Each trait points to a specific cause.

These causes do not belong to one political party. They are structural, which means they respond to structural fixes.

Baumol’s Cost Disease

In the 1960s, economists William Baumol and William Bowen studied live performance. A string quartet in 1800 needed four musicians and about the same amount of time as a string quartet today. Productivity did not rise. Yet musicians must be paid wages that compete with factories and offices where productivity did rise. So the price of a concert climbs.

Teaching, nursing, and child care work the same way. One caregiver can safely watch only so many toddlers. A nurse can attend only so many patients per shift. When wages rise across the economy, these labor-heavy services get more expensive even if nothing about them changes. That is not a failure. It is arithmetic.

Land, Zoning, and Gatekeeping

Housing has a different problem. A home is a building plus the land under it, and land in high-demand cities is scarce by definition. Zoning rules that ban apartments on most residential lots, require large minimum lot sizes, or mandate parking make that scarcity worse. When cities limit what can be built, rising demand turns into rising rent.

Health care and higher education add licensing and gatekeeping. Licensing protects quality, but it also limits who may perform a task and how many new providers can enter. Hospital mergers have reduced competition in many regions. Accreditation and prestige make it hard for cheaper colleges to win students. Each barrier may have a reason. Together they keep prices high.

Can AI Bend the Expensive Curve?

Here is the honest answer. In some places, yes. In others, only if policy changes first. The difference depends on whether the cost comes from labor and paperwork, which software can shrink, or from land and permission, which software cannot create.

Hype in either direction does damage. Promising that AI will fix rent creates cynicism. Dismissing it entirely leaves real savings on the table.

Where AI Plausibly Helps

Several costs on the rising curve are made of paperwork and routine cognitive work, which is exactly what cheap AI does well.

• Tutoring. A capable AI tutor can give one-on-one practice to students whose families could never afford a private tutor, easing one source of the Baumol problem in education.
• Health care administration. Billing, coding, prior authorization, and documentation consume a large share of clinician time and hospital budgets. Automating that work could free staff for patients.
• Design and permitting. Software can check building plans against codes, speed reviews, and help standardize designs, cutting months from projects where delay itself costs money.

None of these replaces a nurse at a bedside or a teacher in a room. They shrink the overhead around those people, which is where much of the cost has quietly grown.

Where AI Cannot Help Without Policy

No algorithm can create land in a city where zoning forbids building. A faster permit review means little if the answer is still no. AI can draft a perfect apartment plan, but only a city council can make that plan legal on most lots.

Credentialing works the same way. If a task requires a specific license, cheaper software does not lower the price until rules allow trained people, supported by good tools, to do more of the work. In these areas, technology is ready before the law is. Savings arrive only when reform catches up.

Building Abundance That Reaches Every Household

Michael Ten’s mission at IdeaRiff is abundance for all, and the two price curves show why that goal needs more than one tool. Cheaper technology handles half the job. Reforms in housing, health care, and education handle the other half. Without both, families keep buying cheaper TVs while falling behind on rent.

There is also a role for income floors. A basic income, delivered through trusted community channels such as credit unions and nonprofit networks, can give households stability while slower reforms take hold. Delivery through local institutions keeps money close to the people it serves. Ownership matters as well. When communities can own and use open tools, the savings from cheap AI flow to them rather than collecting with a few platforms.

What to Push For

Readers who want the expensive curve to bend can support concrete, nonpartisan steps that already have backing across the political spectrum.

1. Zoning reform that legalizes apartments, duplexes, and accessory dwellings near jobs and transit.
2. Faster, digital permitting with clear timelines.
3. Scope-of-practice updates that let nurse practitioners, pharmacists, and other trained professionals do more.
4. Price transparency and competition review in hospital markets.
5. Open educational tools and lower-cost credential paths.
6. Pilots of income floors run through credit unions and nonprofits.

None of these requires picking a side. Each simply removes a barrier between people and cheaper essentials.

What to Watch For

The best test is simple. Track the next annual update of the BLS price data. If AI is truly reaching the rising curve, hospital services, tuition, and child care should start growing more slowly than wages. If they do not, the savings are being captured somewhere along the way.

Watch local decisions too. City council votes on housing and state votes on licensing will shape your budget more than any new gadget. The TV will keep getting cheaper on its own. The rent will not.

This article is for educational purposes only and is not financial advice.

That small message carries a big question. If a company can decide which working parts your device will accept, who really owns it? You paid for the phone. Yet the final say over its repair sits somewhere else, on a server you will never see.

This article makes one argument. Right to repair is freedom tech for physical things. It belongs in the same family as open source software, self-custody, and encryption, and it deserves the same attention from anyone who cares about real ownership.

What Freedom Tech Already Teaches Us

Freedom tech is a loose name for tools that move control from institutions back to individuals. Open source software lets you read, change, and share the code you run. Self-custody lets you hold your own keys instead of trusting a third party with your money. Encryption and privacy tools let you decide who sees your messages. The common thread is simple. You should not need permission to use what is yours.

Most of that conversation happens on screens. It is about code, keys, and data. But the same principle applies to the objects on your desk, in your kitchen, and in your garage. A laptop, a dishwasher, or a tractor is also a bundle of rules about who may open it, fix it, and keep it running. Those rules can favor the owner or the maker.

Ownership is a set of abilities

On paper, ownership is a receipt. In daily life, it is a list of things you are able to do. Can you open the case? Can you buy a replacement part? Can you hire any technician you choose, or must you go back to the company that sold it? Each yes makes ownership thicker. Each no makes it thinner.

When the maker controls the parts, the diagnostic tools, the manuals, and the software that approves each repair, the answers drift toward no. At that point you are closer to renting your device than owning it.

Repair is the hardware version of self-custody

Self-custody means you hold the keys. Open source means you can read the code. Right to repair means you hold the manual, the parts list, and the freedom to choose who does the work. The logic is the same in all three cases. Remove the gatekeeper, and the user gains real control.

Seen this way, repair is not a hobby or a trend. It is the physical layer of a larger idea. A world with free software running on devices nobody may fix is only half free.

How Control Slips Away From Owners

Products rarely announce that they are hard to fix. The change happens through small design and policy choices.

None of these choices changes the price tag. They change what the price buys. Each one shifts a little more power from the person who owns the device to the company that made it.

Parts pairing in practice

The newest barrier is called parts pairing. A manufacturer links a component, such as a screen, battery, or camera, to one specific device. If a different part is installed, the device checks whether the part has been approved through the maker’s own software. If it has not, the device can show warnings, hide features, or refuse to work with the part.

The iPhone is the best known example. According to Apple’s own support pages, the Parts and Service History screen can label a replaced screen, battery, or camera as “Unknown” when the part is not genuine, or when it has not been verified and linked to the phone after the repair. An unverified battery can also lose reliable battery health readings. Apple has since added a Repair Assistant that lets supported genuine parts, including used ones, be calibrated on the device. That is real progress. It also shows where the power sits. The owner gets the feature back when the maker’s software agrees.

The cost of a missing permission

Think about what parts pairing means for a working part. The part fits. The part functions. The only thing missing is a permission. That is a software lock on a physical object, and it is the clearest place where freedom tech and repair meet.

The costs land on ordinary people. Devices that could have been fixed for a modest price are replaced instead. Every forced replacement moves money out of households and into a small number of large firms.

The Laws That Are Opening the Hood

Lawmakers have started to treat repair as a matter of ownership rather than convenience. Several states and the European Union now require makers to share what repairers need. The details differ, but the direction is the same. More parts, more manuals, and fewer locks.

Here are the key rules worth knowing. These are summaries for general education. For specifics, read the full text or check with your state attorney general.

California and Minnesota set the baseline

California SB 244 took effect on July 1, 2024. For electronics and appliances with a wholesale price of $100 or more, makers must offer documentation, functional parts, and tools for at least seven years after the last date a model was manufactured. For products wholesaling between $50 and $99.99, the requirement is three years.

Minnesota’s Digital Fair Repair Act also took effect on July 1, 2024. It covers digital electronic equipment sold on or after July 1, 2021. Makers must make parts, tools, and documentation available to owners and independent repair shops on fair and reasonable terms. Together, these two laws set a simple expectation. If you sell it, you help keep it running.

Oregon takes aim at parts pairing

Oregon SB 1596 goes a step further. It bans parts pairing that blocks a working replacement part, reduces the performance of a device, or shows warnings the owner cannot dismiss. The rule applies to consumer electronic devices first made and sold after January 1, 2025. The Oregon Attorney General can enforce against violations that occur from July 1, 2027.

That makes Oregon one of the first places to treat a software lock on a part as a consumer issue in its own right. It names the exact problem the unknown screen message represents.

Europe moves on its own track

The European Union adopted Directive (EU) 2024/1799 in June 2024 to promote the repair of goods. Member states were required to apply their national rules from July 31, 2026. Progress has been uneven. Germany had its law in force by that date, while several large markets, including the Netherlands and Italy, were still working through their bills in the weeks after the deadline.

Even with the delays, the signal is clear. Repair is becoming a standard expectation in major markets, not a special request.

Repair as a Path to Abundance

Abundance is usually framed as making more things. Repair offers a quieter version. It is about getting more years out of the things you already have. A phone that lasts six years instead of three doubles the value of the money spent on it.

That extra value stays in the household. When products cannot be fixed, each purchase starts to behave like a subscription to the maker, with a renewal fee every time something breaks.

Community repair keeps skills alive

Volunteer repair cafes show what this looks like at street level. Neighbors bring broken lamps, toasters, and bicycles, and volunteers help fix them while teaching the owners how.

Those events depend on access. A skilled volunteer can do little without parts and manuals. Good repair laws and repair-friendly design make every local fixer, paid or unpaid, more useful to the people around them.

How to Buy With Repair in Mind

You do not have to wait for every law to pass. The most direct vote you have is at checkout. A few minutes of research before you buy can add years to a device and keep more control in your hands.

Here is what to look for:

• Published repair manuals. Check whether the maker offers official guides or whether independent guides exist for the model.
• Parts you can buy. Look for an official parts store that sells to owners, not only to authorized shops.
• Repair scores. Some regions publish repairability ratings, and independent sites score popular devices.
• Standard screws and replaceable batteries. These small details often decide whether a repair costs little or a lot.
• Software support length. A device that stops getting updates can become unsafe long before the hardware fails.
• Parts pairing policies. Search the model name with “parts pairing” or “unknown part” before you buy.

Questions to ask before you pay

A short checklist helps at the store or online. Ask whether an independent shop can replace the battery without the maker’s approval. Ask how long the company promises parts and updates.

If the answers are vague, treat that as information. A company that is proud of its repair support usually says so clearly. Silence often means the gatekeeper is still in place.

Conclusion

Freedom tech has taught a generation to ask who controls their code, their keys, and their data. The same question belongs to the physical world. When a maker can veto a working part, withhold a manual, or lock out an independent technician, ownership becomes a rental agreement with extra steps.

Right to repair closes that gap. Open manuals, available parts, and independent repair do for hardware what open source and self-custody do for software and money. Buy with repair in mind, support the laws that open the hood, and keep the things you own working for you.

This article is for educational purposes only and is not legal advice.

The payoff of fixing this is concrete. You get notes you can still open in twenty or thirty years with whatever software exists then. You avoid lock-in to one company and its pricing changes. And you hold files you own outright, stored where you decide, readable by any editor and by your own local AI. Some writers call a collection like this a “second brain”, a term popularized by Tiago Forte’s productivity book. This article uses plainer words: notes, notes library, and personal knowledge base.

Why a Single App Is a Weak Home for Lifelong Notes

Software companies are businesses, and businesses change. Evernote offers a clear example. Bending Spoons agreed to acquire it in November 2022, closed the deal in January 2023, and laid off 129 employees the following month. In December 2023, the free plan dropped to a maximum of fifty notes and one notebook per account, down from limits that had allowed up to 100,000 notes.

Evernote is one case among many. Google shut down Google Reader in 2013, and many smaller tools have closed or pivoted since. Apps still have value, but an app should be treated as a lens for viewing your notes, while the notes themselves live somewhere you control. When the lens breaks, you swap it for another one and keep your eyes.

Five Layers of a Long-Lasting Notes Library

A useful way to think about durability is as a stack of five layers: format, identity, links, copies, and publication. Each layer answers one question, from whether future software can read the file to whether a note should be public and permanent.

You can adopt the layers one at a time. Fixing the first two layers alone makes a large difference.

Layer One: Format

The most durable everyday format for personal notes is plain text, and Markdown is the most popular way to add light structure to it. A Markdown file is readable by a person even with no special software. The CommonMark specification gives the syntax a stable public definition, which reduces the risk of one tool interpreting your files differently from another.

For finished documents meant for long term preservation, institutions often prefer richer formats. The Library of Congress Recommended Formats Statement lists XML based markup, EPUB3, and PDF/A among its preferred formats for textual works, with plain text as an acceptable format. A sensible personal rule follows from this: keep working notes in Markdown, and when a piece becomes finished, export a PDF/A or HTML copy alongside it.

Layer Two: Identity

Every note needs a stable identity that does not depend on a database. The simplest method is a file name that combines a date stamp with a short title, such as 20261009-durable-notes.md. Date stamps sort naturally and rarely collide.

Identity also includes metadata. A small block of key and value lines at the top of each file, often called front matter, can record tags, a source URL, and a created date. Because front matter is plain text, any future tool or script can read it. Avoid storing important facts only in a colored label or proprietary field, since those rarely survive an export.

A Folder Layout You Can Copy

A simple structure keeps identity and privacy clear from the first day. The layout below uses one top folder, a few subfolders, and date-stamped file names:

• notes/ as the top folder for everything
• notes/inbox/ for quick captures, such as 20261009-podcast-idea.md
• notes/library/ for lasting notes, such as 20261009-durable-notes.md and 20250314-reading-habits.md
• notes/index/ for one map per topic, such as index-writing.md and index-health.md
• notes/attachments/ for images and PDFs, named to match their note, such as 20261009-durable-notes-diagram.png
• notes/public/ for finished pieces you are happy to publish
• notes/archive/ for PDF/A or HTML exports of finished work

Lowercase names with hyphens instead of spaces work on every operating system and in web addresses. Keep the tree shallow so links survive reorganizing.

Layer Three: Links

Links are what turn a pile of notes into a personal knowledge base, and they are also the first thing to break during a migration. The wiki style double bracket link, which places a note title inside two pairs of square brackets, is now supported by many tools, including Obsidian and Logseq. Standard Markdown links that point to a relative file path are even more portable, since nearly every Markdown renderer understands them.

Two habits protect links over time. First, link to the stable file name rather than to a heading or a block identifier that only one app understands. Second, keep a plain index note for each major topic that lists its key notes in order. If every automatic backlink feature vanished tomorrow, these index notes would still give you a readable map of your own thinking.

Layer Four: Copies

Format and links mean nothing if the only copy sits on one laptop. The widely used 3-2-1 backup rule, a phrase coined by photographer Peter Krogh in his book on digital asset management, calls for three copies of your data, on two different types of media, with one copy kept off-site. For a notes folder, that might mean the working copy on your computer, a second copy on an external drive or home server, and a third copy in an encrypted cloud backup.

Sync is helpful, but sync is a different thing from backup. A sync service faithfully copies deletions and corruption to every device within seconds. A true backup keeps older versions you can roll back to. Git works well here, since every commit is a snapshot that can be pushed to more than one remote. Schedule it to run automatically.

Layer Five: Publication

Some notes are meant to be shared: reading summaries, glossaries, or public essays. For these, decentralized storage adds a layer that no single host controls. IPFS identifies files by a content identifier derived from the data itself, so the same file has the same address wherever it is stored. Content on IPFS stays available only while at least one node keeps it pinned, which means you either run a node yourself or rely on a pinning service.

Arweave takes a different approach. Uploaders pay a one time fee that, according to its documentation, is intended to cover storage for 200 years at present costs, with the remainder held in an endowment for later years. That makes it attractive for notes you want to keep public indefinitely. Never upload anything private, since removal is designed to be impractical. Treat permanent storage as a printing press for finished public work, never as a backup for a personal journal. This is educational information and should not be read as financial advice.

A Yearly Restore Test for Your Notes

A durability plan is only proven when you test it. Once a year, pick a quiet afternoon and pretend your main note app is gone. Restore your notes from the off-site copy onto a different computer, open them in a basic text editor, and try to answer a few questions using only the files, such as where a favorite quote came from or how two ideas connect.

If any answer requires the original app, you have found a weak layer. Perhaps an important tag existed only in the app database, or a group of attachments never made it into the backup. Fix that gap and record what you changed. Over a few years, these small repairs add up to a truly portable notes library.

A Weekend Checklist

The following list condenses the framework into steps you can complete over a single weekend. All of them work with free software.

• Export everything from your current app into Markdown or the closest open format it offers.
• Rename files with a date stamp and short title so each note has a stable identity.
• Add front matter for tags, sources, and created dates.
• Convert app specific links into file name links and build one index note per major topic.
• Set up three copies on two kinds of media with one kept off-site, and confirm that older versions are retained.
• Choose a public folder for notes you are happy to publish, and keep it separate from private material.

Once these steps are done, the choice of the app becomes a matter of taste. You can try a new editor, return to an old one, or use two at once, because the notes belong to you.

Owning Your Records in an Age of Local AI

A plain text notes library has a benefit that few people expected a decade ago. Local AI models running on your own hardware can read, search, and summarize a folder of Markdown files directly, with no need to upload your private thinking to a remote service. A folder that is open, well named, and richly linked is close to an ideal input for a personal research assistant that answers to you alone.

There is also a broader point about ownership and freedom. When personal records live in open formats on storage you control, people decide who can read them, how long they last, and which tools touch them. No platform can raise a price, change a policy, or close an account and take your thinking with it. Durable notes are a small, practical piece of a future where knowledge tools serve many people instead of concentrating control in a few companies.

Keeping notes readable for decades comes down to a handful of deliberate choices: open formats, stable names, a simple folder layout, portable links, real backups, and a clear line between private notes and public, permanent work. Together they turn years of reading and reflection into an archive you truly own. Start with one layer this week, run your first restore test within the year, and your future self will inherit a library instead of a locked box.

If the answer is a handful of API platforms, AI abundance will look a lot like the last wave of the internet, with enormous gains flowing to a few gatekeepers and thin margins for everyone else. If the answer includes the workers, builders, and small operators doing the work, abundance can spread widely. The outcome is a matter of design, and the design choices are available now.

Where the Value Goes by Default

Markets tend to reward whoever controls the bottleneck. In app stores, ride hailing, and freelance marketplaces, the platform owns the customer relationship, the payment flow, and the reputation data. Workers supply the effort, but the platform sets the fee and can change it at will. Agent markets are drifting toward the same shape, with a few large providers controlling model access, discovery, billing, and identity.

That shape is a choice. The open web itself was built on protocols that nobody owned, such as HTTP and email, and those protocols let millions of small sites and businesses reach customers directly. Agent markets can follow that path if builders insist on it early, before habits harden into lock in.

The Toll Booth Problem

When one company sits between every buyer and every seller, it becomes a toll booth. Each transaction pays a fee, each listing follows the rules of the booth, and each seller can be removed without much recourse. For agents, that can mean a platform taking a large share of every job while also owning the data that proves who did good work.

The fix is not to ban platforms. Good platforms provide real services such as discovery, dispute handling, and quality checks. The fix is to make sure the core pieces of the market, especially identity, reputation, and payment, are open enough that workers can leave and take their value with them.

Principle One: Open Protocols Over Closed Portals

An agent should be able to find work, describe its abilities, and accept a task using open, documented protocols. Standards for tool calling and agent to agent messaging are emerging, and the healthiest markets will build on them rather than on private interfaces that only one vendor controls.

Open protocols lower the cost of entry. A cooperative of five developers can publish an agent that speaks the same language as an agent from a large firm. Buyers can compare them on results rather than on which portal they happen to live inside.

Principle Two: Reputation That Travels With the Worker

Reputation is the most valuable asset a worker builds, and today it is usually trapped inside one platform. A freelancer with years of excellent reviews on one site starts at zero on another. That trap keeps workers loyal to the platform even when the fees rise.

Agent markets can do better with portable, verifiable records. Completed jobs, signed attestations from clients, and public performance logs can be tied to an identity that the operator controls. When reputation moves with the worker, platforms must compete on service quality instead of on captivity.

Principle Three: Direct Payment Rails

Payment is where much of the value leaks away. Card networks, processors, and platform wallets each take a cut and add delay. For agents doing small tasks, those costs can exceed the value of the task itself. Direct rails let the buyer pay the worker with minimal friction.

The x402 protocol revives the HTTP 402 Payment Required status code so that a service can request payment per request and a client can pay and retry automatically. Low fee networks such as Bitcoin Cash, where typical fees are a small fraction of a cent, make per task settlement practical. Community projects such as x402-bch show how that payment pattern can work on Bitcoin Cash.

Low Fees Make Small Jobs Worth Doing

When each payment costs only a tiny fee, new kinds of work become viable. An agent can sell a single translation, one data lookup, or a quick code review and receive payment for each. Micro work that was never worth billing becomes a real income stream for small operators.

Direct payment also shortens the wait. A worker who is paid at the moment the task is accepted does not carry thirty or sixty days of unpaid invoices. That cash flow matters most to the smallest operators, who have the least cushion.

Principle Four: Transparent Fee Splits

Every market has costs, and someone must cover them. The problem is hidden fees, not fees. A fair agent market publishes exactly how each payment divides between the worker, the operator, the platform, and any infrastructure provider, and the split appears on every receipt.

Programmable payments make this easy to enforce. A single transaction can route shares to several parties at once, so the split is visible and verifiable rather than promised in a terms of service page that changes quietly.

Principle Five: On Prem and Cloud Agents Working Together

Not every agent should live in a giant data center. Many workers and small firms run capable models on their own hardware, which keeps sensitive data private and keeps costs predictable. Others rent cloud capacity when demand spikes. A good market treats both as equal participants.

Interoperability is the key. If an on prem agent in a home office can accept the same task format, prove the same credentials, and receive the same payments as a cloud agent, then ownership of compute stays distributed. That distribution is itself a guard against concentration.

Principle Six: Ownership by the People Who Do the Work

The most direct way to keep value with workers is to let them own the operator. A worker owned cooperative can run an agent marketplace, set fees by member vote, and return surplus to the members who generate it. Platform cooperatives already exist, such as Stocksy United, a stock photography cooperative owned by its contributing artists.

Cooperative ownership can be paired with professional management. The Mondragon Corporation in Spain shows that worker cooperatives can operate at industrial scale across manufacturing, retail, and finance. Agent markets are young enough that cooperative models can be built in from the start.

Learning From Credit Unions

Credit unions offer a useful template. They are member owned, they share back office services through networks, and they let members use shared branch and ATM systems across many independent institutions. Each credit union stays local and accountable while gaining the reach of a larger system.

An agent network could follow the same model. Independent cooperatives and small operators could share discovery, dispute resolution, and identity services through a jointly governed network, while each keeps its own members, pricing, and culture.

A Sketch of a Fair Agent Market

Picture a client who needs a market research summary. The request goes out over an open protocol. Three agents respond: one from a worker cooperative, one run on prem by a solo consultant, and one from a cloud provider. Each response includes a portable reputation record and a price.

The client chooses the consultant. Payment flows directly through a low fee rail, with a published split such as most of the payment to the consultant, a small share to the network for shared services, and a fixed amount for infrastructure. The completed job adds a signed record to the consultant’s reputation, which stays with the consultant wherever she works next.

• No single company owns the customer.
• No worker loses reputation by switching networks.
• No fee is hidden from either side.

Abundance for All, Not Abundance for a Few

AI can make many goods and services dramatically cheaper. That is the promise of a post scarcity direction, where basic needs become easier to meet because the cost of knowledge work falls. Cheap output alone does not guarantee shared prosperity. If ownership and payment flows concentrate, falling costs can coexist with rising inequality.

The path to broad abundance runs through the structure of markets. Open protocols, portable reputation, direct payments, clear fee splits, distributed compute, and cooperative ownership each push value outward toward the people doing the work. Together they form an economy where productivity gains show up in many households instead of a few balance sheets.

Agent markets are being designed right now, in protocol drafts, payment experiments, and the business models of early operators. Builders, buyers, and policymakers who want AI abundance to reach everyone can choose open standards, insist on portable records, support direct payment rails, and back cooperative operators. The technology is ready. The remaining question is whether the market will be built to pay workers, and that choice belongs to the people building it.

This article is educational and is not financial or legal advice.