The US
ProgramEight plans, checked against the numbers
Energy, housing, healthcare, education, work, government, money. Every number carries a source. Every plan names what would prove it wrong.
What this is
Eight plans for fixing the United States, written so they can be checked, with a public backend that lets anyone check them.
Most national plans are written so they cannot be checked. They name a direction, avoid a rate, and put the number somewhere the reader will not multiply it. These do the multiplication in public.
Each plan states where we actually are with sourced figures, what has to change, the strategies with numeric targets attached, and the specific observations that would prove the plan wrong. Each one also carries a section called How this fails, because the strongest objection belongs in the document rather than in the replies.
They are not eight separate documents. They share a workforce, a permitting system, and a balance sheet, and they fail together if any of three of them fail.
None of those six numbers is contested. They come from EIA, OECD, NAGB, GAO, Census and CBO. The argument is about what to do, and that argument is easier to have when everyone is looking at the same table.
Eight plans
Each one stands alone and each one depends on the others. Click through to the detail.
The Buildout
10× the energy, rebuild the machines that build things, train 50 million adults — paid for by AI capital and recovered corporate tax base.
Housing Abundance
Build 15 million homes in ten years by making it legal to build them.
Healthcare
Pay 2.5× the developed world for 2.7 fewer years of life. Close that gap and it frees roughly $1.5 trillion a year.
Education for Human 3.0
45% of twelfth graders can't do basic math. Fix the floor first, then build the thing worth building.
Work & Income
The jobs are going. The work isn't. Move 170 million people from holding jobs to owning capability.
Government
$186B in improper payments, four-year permits, 79% of IT spend keeping old systems alive. Every other plan runs through this one.
Fiscal Solvency
Interest hits $1 trillion this year and $2.1 trillion by 2036. Everything else is unaffordable if this fails.
Shared Identity
Unify the country around a positive shared identity based on the Bill of Rights. The plan that makes the other seven possible.
How they connect
- Government is the enabling plan. Nothing else executes if permits take four years and $186B a year leaks. Buildout, housing and healthcare all inherit its timelines.
- Healthcare is the fiscal plan. Health spending at 17.2% of GDP is the largest recoverable pool in American public life. Fiscal solvency depends on it more than on any tax change.
- Education and Work are the same pipeline at two ages. The buildout needs 100,000 electricians a year and housing needs 250,000 construction trades. Neither exists without them.
- The buildout is the physical layer. Cheap abundant energy is the growth lever under the fiscal plan and the input to everything else.
Energy, autonomy, capability
10× American energy capacity, a domestic drone and robotics industry, and free training for 50 million adults — paid for without touching households.
- S1Climb the rate ladder. 86 GW/yr today to a sustained 250 GW/yr, 1,860 GW added over ten years. Target: ~3,160 GW installed by 2036.
- S2Cell independence. Underwrite the 56 GW of announced-but-unbuilt cell plants with offtake guarantees. Target: 60 GW/yr domestic cells by 2030.
- S3Transformer surge. Lead times are 128 weeks, up to four years for high-capacity units, with prices up 77% since 2019. Target: under 52 weeks by 2030.
- S4Queue reform. 2,061 GW is already financed, sited and designed. Process converts it at near-zero capital cost. Target: median request-to-operation under 24 months.
- S5One million electricians. DOL projects 81,000 openings a year as replacement demand alone. Target: 100,000 new electricians and line workers a year.
Build 15 million homes
The shortage is a permitting artifact, not a construction-capacity one.
- S1Legalize the housing. By-right approval on objective standards, conditioned into federal formula funds — the only federal lever over local zoning that has ever worked. Target: median permit decision under 60 days.
- S2Build what is already legal. A federal construction-lending facility for the 750,000 entitled-but-unbuilt units. Approvals sunk, sites entitled, design done. Target: backlog under 100,000.
- S3Industrialize the build. Housing productivity per worker has not risen in fifty years. Target: 30% of new units factory-built by 2036.
- S4Make the trades a career again. Target: 250,000 construction-trade completions a year.
- S5Stop subsidizing the shortage. Demand subsidies in a supply-constrained market raise prices. Phase them down as supply arrives. Unpopular and correct.
Costs what it's worth
America pays 2.5× the developed world and lives 2.7 fewer years. Closing that gap is the largest pool of recoverable money in American public life.
- S1One price list, published. A market cannot clear on secret prices and neither can a regulator. Target: 100% transparency compliance with real penalties by 2028.
- S2Reference pricing on everything the government buys. Federal programs are roughly half of American healthcare. Target: benchmarked to a basket of peer-country prices.
- S3Kill the administrative layer with software. One claims standard, one eligibility API, one prior-auth schema. Target: under $400 per person, from over $1,000.
- S4Pay for prevention like it's a procedure. Metabolic disease drives the outcome gap more than access does. Target: measurable reduction in age-adjusted prevalence.
- S5Publish the outcomes. Price transparency without outcome transparency is a race to the cheapest. Target: full risk-adjusted outcome publication by 2030.
Education for Human 3.0
Record spending and record-low outcomes, in the same year. Any plan that opens with "spend more" has not read the table.
| Cohort | Reading | Math | Note |
|---|---|---|---|
| 4th grade | 40% | 24% | Largest reading share since 2002 |
| 8th grade | 34% | 39% | Largest in NAEP's 32 years of reading testing |
| 12th grade | 32% | 45% | Highest math share ever recorded |
| Spending per pupil, FY2024 | $17,619 | +6.6% | A historic high, in the same year |
- S1Mastery gates, not grade levels. No advancement past a foundational gate without demonstrating it, with unlimited attempts and no stigma. Target: below-basic under 10% by 2036.
- S2One tutor per student. The two-sigma result has been known since 1984 and was never affordable. It is now. Target: every student has an AI tutor with a human teacher supervising by 2030.
- S3Teach how the world actually works. Physical law, evolution, economics, statistics, incentives, propaganda, and how to evaluate a claim. Target: required sequence in every state by 2032.
- S4Make building things a graduation requirement. Every student ships something real that someone outside the school saw. Target: 100% of graduates with a portfolio, not just a transcript.
- S5Publish everything. Per-school, per-dollar, risk-adjusted, standard schema. Target: full outcome transparency by 2030.
Per-pupil spending is held flat in real terms in this plan on purpose. If outcomes only move by adding money, the plan is wrong and should be replaced rather than funded.
Through the AI transition
The country is going to have a labor shortage and a displacement crisis at the same time, in different places. Any plan that treats that as one problem gets it wrong.
192,000 net displacements a year is about 0.1% of a 170-million labor force. But it is not spread evenly: it lands almost entirely on people entering the workforce, in exactly the roles that used to be how a career started. The ladder is losing its bottom rungs while the building is still standing.
- S1Fund the person, not the job. Training, credentials, tooling and startup capital attach to the individual and travel with them. Target: every adult has a portable capability account by 2030.
- S2Rebuild the entry rung deliberately. If firms no longer hire juniors, the training that used to happen inside firms has to happen somewhere. Target: 1 million funded apprenticeships a year by 2032.
- S3Point the displaced at the shortage. The country is short 81,000 electricians and 250,000 construction trades while automating claims processing. Nothing currently connects those people. Target: placement within 12 months.
- S4Make owning a business the default option. Most people have never been shown how. Target: 5 million new independent businesses by 2036.
- S5Measure it honestly, monthly. Net displacement by occupation, entry-level postings, time-to-placement, formation rate, on a fixed public schema. The policy has to change when the rate changes.
A government that runs on understanding
Not how big the state is — how well it does the things it has already decided to do. Every other plan here runs through this one.
- S1State, Policy, Questions, Actions. Rebuild program administration on live state, explicit policy, continuous evaluation, and defined action — instead of procedure manuals executed by people. Target: the ten largest benefit programs on SPQA by 2034.
- S2Pay when you can verify, not when the clock runs out. Not more auditors — verification against live authoritative state at the moment of payment. Target: improper payments under $40B, recovering ~$150B a year.
- S3Shot clocks with defaults. A deadline without a default outcome is a suggestion. Target: median EIS under 18 months, median environmental assessment under 6.
- S4Fund replacement, not maintenance. Cap operations budgets on a declining schedule so agencies must modernize to keep operating. Target: maintenance under 50% of IT spend.
- S5Publish state continuously. Not an annual report — a live feed on a public schema. Target: every major program by 2032.
Solvency
Interest is the fastest-growing line in the budget and it buys nothing. No road, no grid, no student, no cure.
- S1Treat healthcare as the fiscal plan. Getting health spending to 12% of GDP releases ~$1.6 trillion a year. The federal share is the largest deficit reduction available anywhere. Target: federal health spending growing slower than GDP by 2030.
- S2Collect what is already owed. $186B a year in improper payments, plus $233–521B in estimated annual fraud losses. Not a tax and not a cut. Target: ~$150B/yr recovered.
- S3Restore the corporate base. Receipts fell 14.7% in FY2025 to $452B against $264B in annual corporate tax expenditures. Target: +$300B/yr.
- S4Grow the denominator. Cheap energy, affordable housing near jobs, a workforce that can do the work, a state that permits in months. The highest ceiling and the least controllable lever — no responsible version of this plan assumes it.
- S5Term out the debt while you can. A rate shock against a short average maturity turns a manageable interest line into an unmanageable one inside two years. The cheapest insurance in the plan and the least discussed.
Where it comes from
Price, waste, and base. Not households, and not new deficit.
| Source | Annual | Where it comes from |
|---|---|---|
| Health price and administration | ~$1.6T | 17.2% → 12% of GDP, still above every peer country (PL-00004) |
| Corporate tax expenditure repeal | $264B | JCT total corporate tax expenditures, 2025 (PL-00002, PL-00008) |
| Improper payments and fraud | ~$150B | Of $186B/yr misspent across 64 programs (PL-00007) |
| AI infrastructure levy | $109B | 15% of $725B in 2026 hyperscaler capex (PL-00002) |
| Profit-shifting recovery | $100B | $369B was booked in havens in 2022 (PL-00002, PL-00008) |
| Total recoverable | ~$2.2T/yr | Against a program costing well under a quarter of it |
The conservative case
The buildout's funding stack is the one costed to the dollar, and it is close enough that four inputs can break it — blended capex per GW, the integration adder, the profit-shifting estimate, and the fact that $127B of the expenditure repeal comes from credits that subsidize this program's own buildout. Run all four conservatively and the stack is $306B/yr instead of $473B/yr, which funds about 65% of the energy ladder and forces its last three years down.
That number belongs on this page as much as the headline does.
How the claims get validated
The plans are not documents. They are a graph of files in a public repository, and every node in it can be attacked separately.
Substrate is an open framework for structured knowledge about problems and progress. Problems, Solutions, Claims, Funding-Sources, Risks and Data are separate objects that reference each other by ID. A plan is not prose containing assertions; it is a spine that points at objects, each of which stands or falls on its own.
That structure is what makes a policy plan checkable. You do not have to accept or reject the whole thing. You can go after one node.
The five validation rules
- 01Every number carries a primary source. Not a summary of a source, and not a number recalled from memory. If a figure appears in a plan, its origin is a link to EIA, BLS, CBO, JCT, GAO, OECD, Census, NAGB, LBNL, IFR, SEIA, Freddie Mac, or a named academic paper.
- 02Derived numbers are labeled derived. The ~1,300 GW energy baseline is arithmetic on top of EIA's published 1,189,492 MW, and the plan says so every time it uses it. A derived number presented as a measured one is the most common way policy documents lie without lying.
- 03Every claim names what would falsify it. Each file ends with the specific observations that would prove it wrong — a single year of US capacity additions above 500 GW, health administration failing to fall after claims standardization, entry-level postings recovering on their own. If you cannot say what would change your mind, you have written a slogan.
- 04Each plan argues against itself in public. Every plan carries a section called How this fails. The 10× target is the author's own and the plan states what it would demand. The funding section states the conservative case that breaks its own match.
- 05Every object is separately attackable. Reject the profit-shifting estimate and the stack drops to $373B/yr and the energy ladder shortens — the rest survives. Contributions arrive as pull requests against a single file, so disagreement is a diff rather than an argument.
The point of the backend is not that the plans are right. It is that they are checkable, and that checking them is cheap enough that people actually will.
How you would know it is working
Every baseline here is a published figure from an agency that will keep publishing it, so progress can be checked annually without the program reporting on itself.
| Metric | Plan | Baseline | 2030 | 2036 | Source of truth |
|---|---|---|---|---|---|
| Annual electric capacity additions | PL-00002 | 86 GW | 160 GW | 250 GW | EIA |
| Domestic solar cell capacity | PL-00002 | 3 GW/yr | 60 GW/yr | 80 GW/yr | SEIA |
| Median interconnection time | PL-00002 | >48 mo | 24 mo | 18 mo | LBNL |
| Robot density, per 10,000 | PL-00002 | 307 | 600 | 1,220 | IFR |
| Homes completed per year | PL-00003 | ~1.4M | 2.0M | 2.2M | Census |
| Price-to-income ratio | PL-00003 | 5.08 | 4.3 | ≤3.5 | JCHS |
| Health spending, % of GDP | PL-00004 | 17.2% | 15% | 12% | OECD |
| Life expectancy | PL-00004 | 78.4 | 80.0 | ≥81.1 | OECD |
| Admin cost per person | PL-00004 | >$1,000 | $600 | <$400 | OECD / PGPF |
| 8th grade below basic, math | PL-00005 | 39% | 25% | <10% | NAEP |
| 12th grade below basic, math | PL-00005 | 45% | 28% | <10% | NAEP |
| Entry-level postings, YoY | PL-00006 | −15% | 0% | positive | Hiring Lab |
| Funded apprenticeships per year | PL-00006 | — | 500,000 | 1,000,000 | DOL |
| New independent businesses, cumulative | PL-00006 | 0 | 1.5M | 5M | Census BFS |
| Improper payments, annual | PL-00007 | $186B | $100B | <$40B | GAO |
| Federal IT spend on maintenance | PL-00007 | 79% | 65% | <50% | GAO |
| Median EIS duration | PL-00007 | >48 mo | 24 mo | 18 mo | CEQ / RFF |
| Debt held by the public, % of GDP | PL-00008 | 101% | 103% | ≤100% | CBO |
| Net interest, % of GDP | PL-00008 | 3.3% | 3.1% | <3.0% | CBO |
| Deficit, % of GDP | PL-00008 | 5.8% | 4.0% | <3.0% | CBO |
The three leading indicators that matter most
- Cost per installed gigawatt. If it rises faster than the buildrate ladder, the program is being absorbed as inflation rather than converted into capacity. Every large industrial appropriation fails this way, and it shows up here a year or two before anywhere else.
- Health spending growth versus GDP growth. The fiscal plan is downstream of this single ratio. If it does not cross, nothing else closes.
- Entry-level job postings. The cheapest early read on whether the AI transition is a reallocation or a collapse.
Every number, traceable
If a figure appears anywhere above, it came from one of these.
Energy capacity & generation
EIA — Electricity generation, capacity and sales
EIA — Record generating capacity additions in 2026
EIA — Construction costs for solar, wind and gas generators
World installed power capacity
IRENA via Earth.org — 692 GW of renewables added in 2025
Grid constraints
LBNL — Queued Up 2025
POWER — Transformers in 2026
pv magazine — Transformer lead times
SEIA — Module capacity and the cell gap
Labor
CEWD — Energy workforce fast facts (USEER 2025, DOL)
Goldman Sachs Research — workforce per 300 GW
NREL — JEDI labor modeling
BLS — Employment Situation
US civilian labor force
Indeed Hiring Lab — workforce, AI and reallocation
AI displacement statistics (Goldman Sachs, Apr 2026)
AI demand & capital
LBNL / DOE — Data center demand through 2028
Statista — Big Tech AI capex, $725B in 2026
Hyperscaler capex by company
CNBC — AI spending approaches $700B
Housing
Freddie Mac — Housing supply shortage
White House CEA — 10 million home shortage
Multifamily production and the entitled-but-unbuilt backlog
Home price-to-income ratio by metro
Income needed to buy a home, 2026
Harvard JCHS — price-to-income at a record high
Healthcare
OECD — Health at a Glance 2025, United States
PGPF — US health system vs peers
Peterson-KFF Health System Tracker
Commonwealth Fund — Global perspective 2026
Education
NAGB — Nation's Report Card, reading and math
NAGB — 12th grade declines
Hechinger — NAEP 2024 analysis
Chalkbeat — 12th grade NAEP scores
Census — Per-pupil spending, FY2024
Government
GAO — Improper payments, FY2025
CRFB — $186B in improper payments
GAO — Fraud and improper payments
GAO — Modernizing decades-old legacy systems
RFF — NEPA timelines for clean energy projects
Carlton Fields — NEPA permitting reviews
Fiscal & tax
CBO — Budget and Economic Outlook, 2026 to 2036
CBO — Revenues in fiscal year 2025
CRFB — Net interest costs will double again
BEA — GDP, year 2025
PGPF citing JCT — six largest corporate tax expenditures
CRFB — JCT projects $2.3T in tax expenditures
JCT — full repeal only reaches a 28% rate
Wier & Zucman — Global profit shifting, 1975–2019
CEPR — Profits in tax havens
Autonomy
IFR — Robot density
IFR — US robot installations
CEPA — China's drone ecosystem
DJI market share statistics
DroneLife — DoD Drone Dominance
Training
TCF — Reimagining WIOA
CRS — WIOA and the one-stop delivery system