Methodology

Canonical parameters

Every parameter displayed on this site originates in one of two canonical sources: the canonical parameters document (published in full at /canonical-parameters, v10.7 current canon, 2026-07-03) and the fiscal workbook (NAA_Fiscal_Projection_v10.xlsx, 18 tabs, 1,745 formulas). When sources disagree, the workbook wins. Every AI-assisted response carries a version footer (— Based on NAA DNA v10.7) so drift between site canon and AI responses can be detected.

Lock note. Prior documents — including DNA v17, v19, v20, v21 drafts — are superseded. They remain on file as architectural-exploration archives only. See /canonical-parameters →.

Master tables. Every revenue and expenditure category, every year 2030–2079, in dollars and as % of GDP — computed live from the fiscal engine with formula and assumptions annotated per line, and asserted against the site's charts at every build. See /methodology/master-tables →.

What changed in v10.10 — plain English

Until July 2026, the model's estate-and-wealth revenue used the tax base the IRS can currently see — about $9T of reported wealth above the $10M threshold. The Accord's registry (annual filings with agreed valuations) makes the actual base collectible, and the actual base — from Federal Reserve household data — is four to five times larger (~$50T). In v10.10 we replaced every hand-seeded wealth figure with a model computed from that base: who holds what, who dies when, what the top-up mechanics collect. Result: wealth-transfer revenue at Year 10 is ~$734B/yr, not the ~$164B previously booked, and the Year-10 deployable balance moved from ~$0.79T to ~$1.5T. Derived debt retirement moves decades earlier; the published retirement claim is being restated as a corridor (“retired no later than Year 55 from enactment”) pending the offsets/rates ruling. Skeptical of the registry? The conservative grading toggle re-runs the model at 1× or 2× the IRS-visible base (wealth_base_grading = 0.25 / 0.5); a constitutional-injunction scenario (prepayment enjoined Years 1–5, severability fallback active) is also scored (prepayment_enjoined_years = 5). See /constitutional-memo.

Three scenarios

The fiscal engine simulates three scenarios. All three run with the Debt Sunset Governor ON — the published mode — so what varies is the revenue path and the rate steps the governor takes:

• Optimistic (aspirational): Distributed Healthcare basis $5,550B; revenue runs above central in the transition years. Assumes 25% admin savings and drug pricing held at 120% OECD median for 15 years. Retirement ~2052.
• Central: Distributed Healthcare basis $5,900B. Retirement ~2079 (Year 50, counting 2030 as Year 1). Year-10 deployable surplus ~$0.78T.
• Conservative: same $5,900B healthcare basis, with revenue stressed below central in the transition years (no upward revenue multiplier — the stress is honest, not offset). The governor steps rates up within the corridors; retirement holds inside the 50-year window.

In every scenario, federal debt retires within 50 years by architectural construction (2030 Year 1 through 2079 Year 50).

Known sensitivities (published; pending calibration)

• Wealth & institutional excise erosion: European experience shows 30–40% behavioral erosion of estate-prepayment bases within five years via valuation disputes and mobility. The conservative scenario will apply a 35% haircut in the next model revision (central and optimistic keep the current assumption pending empirical data).
• VAT base vs. Pre-bate cost: Pre-bate of $100/adult + $50/child per month gives an envelope of ≈ $362B/yr (population × per-person caps); the prebate paid is min(gross VAT, envelope), so early years are fully rebated and the Treasury share grows from Year 3. Gross VAT near $1.6–1.8T implies a consumption base that requires validation against CBO 2039 projections.
• Carbon revenue decay: The fee escalator reaches its statutory ceiling and emissions fall over the horizon. By the late 2040s, carbon receipts fund the Energy Stipend and little else. Carbon money is not used to backfill permanent spending — the surplus through the 2030s accrues to the Climate Adaptation Trust, not the General Fund.
• Interest-rate path: Chapter 4 uses a payroll tax corridor but does not yet publish the rate path. If rates stay higher for longer, central retirement slips 6–8 years. The full path and debt-retirement sensitivity band are staged for the next publication.

Dynamic-scoring aspiration

CBO pre-scoring is being pursued through three external channels: the Penn-Wharton Budget Model (PWBM), Resources for the Future (RFF) for carbon and externality components, and the Brookings Hamilton Project for distributional analysis. Until dynamic scoring is complete, certain revenue lines are displayed as sensitivity bands rather than point estimates. The Alliance Incentive is not one of them: it is scored at $0 net — its Layer-A border adjustments are already scored inside the carbon border adjustments, and Layer-B compact revenue is negotiated, never pre-scored.

What is not yet scored

• Alliance Incentive tariff revenue — scored at $0 net; Layer-B compact revenue is negotiated, never pre-scored (Layer-A border adjustments are already inside the carbon border adjustments)
• Specific state-level rollout schedules
• Bottleneck Workforce Programs envelope
• Several externality rates pending EPA cost-benefit review
• Behavioral elasticity for household carbon response (ε = 20–80% band, see Climate tab)

Sensitivity bands vs. point estimates

Every numeric claim on the site carries one of five labels:

• [ARCHITECTURAL] — architectural commitment, not a point estimate (e.g., debt retires within 50 years; central ~2079)
• [POINT] — canonical point estimate from DNA v10.4 (payroll tax 28%, Universal Child Allowance beginning at $800/mo, carbon fee $80/ton Year 1)
• [BAND] — central estimate within a published range (household carbon-response elasticity ε = 20–80%)
• [PLACEHOLDER] — awaits CBO / EPA / National Statistics Board scoring
• [HISTORICAL] — based on a specific historical datum (CBO 2025 LTBO, NHE 2024)

Two composite scores: an empirical architecture, not a claim

COMPASS splits quality-of-life measurement into two composite scores because they are hypothesized to respond to different remedy types at different speeds. Individual scores measure outcomes that travel with people — health, education, income, victimization. Place-based scores measure structural conditions of the tract — infrastructure proximity, housing stock, environmental exposure, local crime rate. A third Balanced composite is the geometric mean across all eight domains, preserved for continuity.

The split is a litmus test for action guidance. If a tract's individual score is low but its place-based score is moderate, remedies that deliver cash and services to individuals (Universal Child Allowance, Pre-bate, Skills Wallet, SUD treatment) are expected to lift that score faster. If place-based score is low but individual score is moderate, remedies that build infrastructure (FQHC construction, mobile health, broadband, school investment) are expected to lift faster.

The academic twin of this architecture is Chetty's Opportunity Atlas; Chetty, Hendren & Katz (AER 2016) showed that place itself changes children's outcomes — the evidentiary basis for tract-level triggers — and Lost Einsteins (QJE 2019) prices what unfound capability costs. Three metric additions are adopted with credit: Chetty's economic-connectedness index, a Case–Deaton despair-mortality sentinel (CDC WONDER), and a Cost-of-Thriving companion (Cass).

This architecture is validated only to the extent that the two scores do diverge, and that different remedies actually produce the predicted differential lifts. Quarterly regeneration publishes the measurements; annual analysis (starting Year 2) tests whether elasticity coefficients match the two-composite hypothesis. If the composites move in lockstep regardless of remedy type across two years of data, the architecture collapses to a single score and the site simplifies.

Elasticity model — first-cut and how it updates

All lift estimates on this site (the fields carrying a gold ⁂) come from a v10.1 first-cut elasticity model derived from pre-launch parameters. Empirical coefficients replace these estimates after the first annual analysis, published Year 2 of deployment. The analysis regresses quarterly score changes on intervention amounts per tract, separately for individual and place-based composites, and publishes the delta between predicted and observed lift per remedy type. The model changelog — date, what changed, and why — lives on this page.

Domain weights (v10.1 first-cut)

Each domain contributes to both composites with different weights; weights sum to 1.0 per domain.

• Health & Longevity: 55 individual / 45 place-based
• Education & Skills: 45 / 55
• Economic Security: 65 / 35
• Housing & Infrastructure: 30 / 70
• Safety & Justice: 40 / 60 (individual-side data is sparse; see note below)
• Environment & Climate: 25 / 75
• Civic Engagement: 55 / 45
• Child & Family Wellbeing: 50 / 50

Safety & Justice data gap

Safety & Justice is the most data-challenged of the eight for this split. Place-based crime data (FBI UCR, state crime registries) is available at tract or small-area resolution. Individual-side crime data (victimization and incarceration by resident) is sparse — NCVS reports regionally, not locally. Tract pages in this domain display a data-quality indicator (🟢 full / 🟡 partial / 🔴 gap). The National Statistics Board roadmap commits to improving individual-side data collection as part of the measurement architecture buildout.

How the two place-based metrics are computed

The Accord displays two tract-level metrics, both pre-computed to the tract centroid and regenerated quarterly. QOL Composite (0–10) collapses the eight COMPASS domains (Health, Education, Economic, Housing, Safety, Environment, Civic, Child/Family — see Ch 15) via geometric mean. Equal domain weighting; geometric mean penalizes weak links so that one catastrophic domain is not masked by a strong one.

Critical Access Proximity (0–10) measures drive-time to six life-threatening-service facility types: Level II+ trauma, OB-GYN, specialty care (cardiology/oncology/neurology), dialysis, mental-health crisis, and FQHC. Each type scored 0–10 against published clinical thresholds (e.g., Level II: 10 at ≤30 min, 0 at ≥120 min). Six sub-scores combined via geometric mean.

Why two, not thirteen. Earlier drafts carried a separate layer per facility type. Two composites communicate the structure better: one tells you what the tract is like to live in, the other tells you what the tract is like to have a heart attack in. Sub-scores remain visible on tract click-through.

Pre-computation is from tract centroid (not county centroid) to the nearest facility point. ~73,000 tracts × 6 facility types ≈ 438,000 distance computations, haversine with urbanicity-adjusted drive-time multipliers. Regenerated quarterly by scripts/proximity_compute.py.

Three-tier delivery model

Healthcare access is delivered through three tiers with different clinical capabilities and deployment ramps. The map reports Critical Access in both Physical (brick-only drive-time) and Effective (best across all available tiers) forms.

• Tier 1 · Telehealth (Y1-3, universal). Post Office 2.0 booth at every site. Capital ~$1.55B (50K/booth × 31K sites), operating $155M/yr. No substitute for acute surgery, labor & delivery, or dialysis sessions.
• Tier 2 · Mobile (Y1-5, targeted). Scheduled mobile units for OB-GYN, dialysis, specialty care, crisis response, FQHC circuits. ~$1.1B capital over 5 years; $550M/yr operating at maturity; ~18M residents served.
• Tier 3 · Brick-and-mortar (Y3-20, plausibility-gated). VHA expansion > hospital upgrade > integrated-care anchor > county-seat addition > greenfield. Only tier for 24/7 surgical and inpatient acute care.

Tier capability matrix

Not every facility type can be delivered at every tier. Tele-psychiatry is standard of care (75% of brick value); dialysis via telehealth is clinically impossible (0%). The matrix:

Effective vs physical Critical Access

The map reports both versions. Physical captures time-to-brick-and-mortar — the residual gap Tier 3 must close over 20 years. Effective captures best available time-to-care across whichever tiers are deployed in that tract at that year. For acute trauma, physical is what matters. For mental-health crisis, specialty consult, routine primary care — effective is what matters.

Tier 3 plausibility gate + site preference

The 20-year schedule ranks tracts by composite deficit (drive-time excess × log-population × adjacency bonus × access-urgency) and selects each facility's site through the plausibility gate: VHA expansion > hospital upgrade (Critical Access Hospital → Level III → Level II) > integrated-care anchor (Kaiser-model) > county-seat addition > greenfield (only if catchment ≥150K and no upgradable anchor within 60-120 min). Capacity constraints reflect workforce pipelines, not capital — Level II trauma is limited to ~7 new facilities/year by the 5-year-lag surgeon residency pipeline.

Operating subsidy for security capacity. Per Blueprint Ch 20, Level II trauma centers, OB-GYN outreach, specialty care, and mental-health crisis facilities in small catchments receive permanent operating subsidy above fee revenue. Their value is in standing capability, not throughput — like fire stations. This is a permanent line in Distributed Healthcare's budget, not a transition cost.

Replication package

Everything needed to reproduce the headline table is published as a single JSON bundle: every canonical parameter (the full CFG export), scenario definitions, calibration constants and phase curves, the avoidance-haircut citation list, and the current headline figures. The bundle is regenerated from the fiscal engine at every build.

Download replication.json →  To verify: clone the repository, run npm install, then npx tsx scripts/generate-derived-fiscal.ts. Your output must match the bundle's scenarioHeadlines block exactly.

Red-team process

The document has been red-teamed continuously via structured conversations with multiple AI systems (Claude, ChatGPT). External validator engagement with the institutions listed above is ongoing. When errors are identified, corrections are posted publicly to the updates log with the prior value, the correction, and the source. The standing invitation — find a flaw we haven't listed and we canonize the fix and credit you — lives at /red-team, seeded with the flaws already found and fixed.

Authored by Bill Hersman, LiDAR Space LLC · DNA v10.7 · July 2026