Healing America

A five-year plan built on what is already measured

Written under the Reality Engine discipline: every number traces to a source, the plan states what would prove it wrong, and the limits section is honest about what the author has and has not done.

The premise, which is not a hope

The usual argument for renewal begins by asserting that people are good, and proceeds on faith. That is not necessary here, because the claim is measured.

Garrett Hardin argued that shared resources are inevitably destroyed. Elinor Ostrom documented more than eight hundred cases where they were not — irrigation systems, fisheries, forests and pastures governed sustainably for centuries, without privatisation and without a state. She received the Nobel Prize in Economics in 2009.

The disaster sociologists went looking for social breakdown under catastrophe and could not find it. Quarantelli concluded in 1954 that panic is rare compared with other reactions. What the literature does document is the opposite failure: elite panic, institutions acting on the assumption that the public will break down.

That is the premise. Americans are not the problem to be solved, and a plan that treats them as one will fail for reasons that have nothing to do with its budget.

What the plan changes

Four tracks. Each rests on a measured result rather than a projection, and each changes who owns an asset or who holds a decision, because a programme can be defunded in one budget cycle and an owned asset cannot.

Heart: ownership, so that staying is rational

None of these is a transfer payment and none requires a forecast to be right.

Homes: where the promise is actually decided

Housing is not adjacent to the question of whether work frees anyone. It is the mechanism. Black tenants won Gautreaux in 1969, and the 1973 remedy designed a regional housing mobility programme that became the template for the federal demonstration that followed. Illinois did not merely join that experiment; it invented it. Five decades later, housing in Chicago remains segregated.

The experiment's finding is unusually specific: children who moved before the age of thirteen saw a 31 percent increase in annual income by their mid-twenties, and the return decays with every year of childhood spent elsewhere. That single sentence contains a queue-ordering rule worth more than most housing budgets.

Health: build for the network that exists

Nearly 96 percent of Canadian homes had high-speed internet access by the end of 2023, while only about 24 percent of households in Indigenous communities met the basic service standard, and every community in Nunavut is served by satellite that does not meet it at all. A system designed against the national average is designed for somebody else.

The architecture that follows is specific: fully functional with the network unplugged; retrieval with provenance rather than generation, because composition is where fabrication lives; calibrated refusal as a frequent and first-class output; and a named clinician who decides, always. On an 8 GB consumer graphics card, across three runs in August 2026, a 7-billion-parameter open-weight model scored 100 percent on the local task battery at about 1.1 seconds per task, under a licence that imposes no field-of-use restriction. The hardware is not the obstacle.

Governance: the part that keeps the other three honest

Twelve governance modules were derived from human institutions and living systems in a companion paper: separating who proposes from who approves, a second key for irreversible acts, self-shutdown with an outside backstop, a check on the checker, tamper-evident records, no-blame incident investigation, independent rotated audit, consent and prior review, compartments, authority that expires unless renewed on evidence, a memory of what failed, and early warning. Of those, two are built in the author's own system, five are partial and five are missing. The missing ones are also the ones absent from the current rules for advanced technology, which suggests they are hard rather than overlooked.

The five-year sequence

The ordering is a dependency, not a wish list. Each year ends with a measurement that decides whether the next year proceeds.

Year one — remove the legal obstacles and prove one case of each. Ordinary state legislation, none of it requiring a constitutional amendment or anyone to become poorer: end the preemption of municipal decisions on wages, housing and procurement; permit community benefit as a scoring criterion in public procurement rather than treating it as favouritism; make employee-ownership conversion the default path for a retiring owner, with pre-approved structures and subsidised valuation; prioritise housing vouchers by the age of the youngest child, which is a queue-ordering rule that costs nothing. In parallel, one anchor institution redirects a measurable share of spending locally, and one nursing station or rural clinic runs the offline design with a clinician who can overrule it. Test to proceed: the procurement share moved, and the clinician says the tool saved time rather than cost it.

Year two — capital, so the pilots can be repeated. A state public bank capitalised to lend where community enterprises and land trusts currently fail, which is at the credit stage rather than the idea stage. The existing state-owned bank's sixteen consecutive record years is the evidence that this is solvent rather than charitable. Mobility counselling funded; source-of-income protections enforced with real penalties; payment standards set by neighbourhood rather than metropolitan average. Test to proceed: loans issued and performing, and voucher holders placing in neighbourhoods they previously could not.

Year three — scale what survived, and only that. Conversions move from exceptional to routine. The health design extends from one station to a region, still with local corpus governance and still with refusal as a first-class output. Anchor procurement extends to hospitals and universities across the region. Test to proceed: retention and productivity in converted firms match the published ranges, or the difference is explained.

Year four — measure honestly, including the failures. Publish what did not work with the same prominence as what did, through no-blame investigation and protected reporting of near-misses. This is the year the plan is most likely to be quietly abandoned, because it is the year the disappointing numbers arrive. Test to proceed: failures published, and at least one element formally retired.

Year five — hand it over, and let the authority expire. Ownership sits with the community rather than the programme. Every authority created in year one reaches its sunset and must be re-justified on evidence, not renewed by formality. The people who bear the consequences hold the power to remove those who run it. Test of success: the thing survives a change of administration.

Who does what

This is not a federal plan, and that is deliberate. Preemption is state law and can be repealed by the same legislature that passed it. Procurement rules are municipal and county instruments. Employee-ownership conversion is tax and corporate structure. Housing mobility is an administrative queue. Anchor institutions are hospitals and universities that already spend the money somewhere. The federal role is mostly to stop preventing it.

The individual role is smaller than the rhetoric of renewal usually suggests, and more real: buy locally where a local supplier exists, ask a retiring owner whether they have considered selling to their employees, and serve on the board of the land trust.

What would falsify this

The plan should be abandoned, or substantially revised, if any of the following turns out to be true.

• The measured results do not transfer. Preston is a British city with a different procurement law, and the employee-ownership figures come from firms that chose conversion, which is a selection effect rather than a controlled comparison.
• Credit is not the binding constraint. If community enterprises fail for reasons other than access to capital, a public bank is an expensive answer to the wrong question.
• Refusal is not useful in the clinic. If a system that frequently says it does not know costs more time than it saves, the health track is wrong at its root. This has not been measured.
• Communities do not want it. The most likely and most legitimate failure. Nothing here entitles anyone to a hearing, and a plan that arrives from outside with its own metrics has already failed the seventh design principle it claims to respect.
• Ownership does not produce the political durability claimed. The argument that an owned asset is harder to take than a programme is historically plausible and not experimentally established.

Limits

This is a synthesis of four prior papers and not a new empirical result. No element of it has been piloted by the author, whose own system has helped zero real users to date, a number stated here because a plan about honest institutions should start by being honest about its source. The costings are deliberately absent: each measure is either ordinary legislation, which is nearly free, or capitalisation, which depends on a state balance sheet the author has not examined. Several results are cited from single jurisdictions. The five-year framing is a discipline for sequencing and measurement, not a forecast, and anyone who tells you they can forecast five years of American politics is selling something.

What the plan does claim is narrower and, I think, defensible: that these four bodies of evidence exist, that they point the same way, that the legal obstacles are ordinary legislation rather than constitutional barriers, and that the sequence above is testable at every stage by someone who wants to prove it wrong.

Licence: SAGL-1.0. Byline: SomaSoft Research. Companion papers: on community wealth and the two-hundred-year horizon; on labour, housing and ownership; on clinical decision support at the edge of the network; and on governance modules.