A result should survive
independent scrutiny.
Method version ${METHOD_VERSION}. Older 0.1.0 input files can be imported and are recalculated using the current method. All calculations run locally; optional public-data retrieval is separate from the calculation engine. Unknown results remain null with warnings.
Five evidence classes
- Observed: a raw dated observation from a cited source; not a model inference.
- Derived: deterministic arithmetic from disclosed observations or user inputs.
- Scenario: conditional arithmetic using hypothetical or substituted inputs.
- Hypothesis: a proposed mechanism that has not been demonstrated here.
- Validated forecast: requires a documented fit, holdout evaluation, calibration, error analysis and appropriate external validation. None is included in this release.
Input origins and missing data
Exports distinguish synthetic examples, manual assumptions and public-source observations. Locally uploaded data retains a source record and should be reviewed before classification. Empty fields, undefined values and null are not zero. Nonfinite numbers, inconsistent units and impossible dates fail validation. Counts cannot be negative; signed fiscal changes remain permitted where the formula allows them.
Computation versus empirical validity
Independent fixtures check conservation, hand-calculated values, edge cases and rejected invalid inputs. They do not establish that an input is accurate, an effect is causal or an approach works in the real world. A simple minimum-capacity model supports a single-path series; a branching or feedback network needs a separate flow model. Linear sensitivity supports a declared expression, not arbitrary causal relationships.
Provenance and audit
Every metric retains its input sources and transformation equations. JSON exports include method version, full inputs, output, assumptions and SHA-256 checksums. Imports recalculate from the inputs. Hashes establish content equality, not truth, identity or source authenticity. Editing source metadata remains a human responsibility.
Comparison and authority
Declare equal resources, budgets, staffing, release dates, evaluation periods and criteria. Material declared differences return not_comparable. No country multiplier, general safety score or war probability exists. A blocked or pending authority gate stops a transaction-dependent evaluation. Supplied approval is never independently certified by the software.
What still requires research?
Domain-specific specializations are marked core-backed exercise or planned. Forecasting, causal claims and recommendations require independent datasets and validation beyond these calculations. No live API import updates a scenario automatically. The original research hypotheses and Scenario Studio remain separate and unchanged.
- NIST: AI Risk Management Framework — Voluntary governance background; not validation of MAR. Reviewed 2026-11-09. Reference, not endorsement or validation.
- NIST: Cybersecurity Framework 2.0 — Outcome taxonomy, not a calculated national risk score. Reviewed 2026-11-09. Reference, not endorsement or validation.
- U.S. GAO: Cost Estimating and Assessment Guide — Published 12 March 2020. Estimation and sensitivity guidance; no endorsement of this implementation. Reviewed 2026-11-09. Reference, not endorsement or validation.
- CBO: Processes and methodology — Published government estimates differ from user-input arithmetic. Reviewed 2026-11-09. Reference, not endorsement or validation.
- U.S. Census Bureau: Data API developer guidance — Keep dataset year, geography, margins of error, and privacy limits visible. Reviewed 2026-11-09. Reference, not endorsement or validation.
- BLS: Public Data API v2 — Series definitions, frequencies, revisions, and access limits vary. Reviewed 2026-11-09. Reference, not endorsement or validation.
- EIA: API v2 technical documentation — Requires a key; no private key is embedded. Automatic access needs a separate approved design. Reviewed 2026-11-09. Reference, not endorsement or validation.
- U.S. Treasury: USAspending API — Obligations and outlays differ; reporting lag and coverage require review. Reviewed 2026-11-09. Reference, not endorsement or validation.
- FEC: OpenFEC documentation — Aggregate research only. No donor-level profiles or private credentials. Reviewed 2026-11-09. Reference, not endorsement or validation.
- World Bank: Indicators API v2 — Optional keyless adapter for selected population/current-USD GDP observations; missing values stay missing. Reviewed 2026-11-09. Reference, not endorsement or validation.
- UNDRR: Sendai Framework monitoring — Country-reported indicators have coverage limits; no causal program effect is inferred. Reviewed 2026-11-09. Reference, not endorsement or validation.
- OECD: Policy coherence for sustainable development — Policy lens, not an empirical spillover multiplier. Reviewed 2026-11-09. Reference, not endorsement or validation.
- FEMA: Official disaster-risk resources — Automated retrieval blocked. Product/version validation remains planned; no dataset integration is claimed. Reviewed 2026-11-09. Reference, not endorsement or validation.
- WTO: Trade statistics and methods — Year and series definitions required; observations and forecasts remain separate. Reviewed 2026-11-09. Reference, not endorsement or validation.
- UNCTAD: Trade and maritime context — Automated retrieval blocked; no live maritime feed is claimed. Reviewed 2026-11-09. Reference, not endorsement or validation.
- U.S. Treasury / OFAC: Sanctions programs and guidance — Competent current legal review is required; no clearance or workaround is generated. Reviewed 2026-11-09. Reference, not endorsement or validation.
- NIST: Contingency Planning Guide for Federal Information Systems (SP 800-34 Rev. 1, updated) — Background for recovery time and recovery point objectives. Reviewed 2026-11-11. Reference, not endorsement or validation.
- U.S. Energy Information Administration: Measuring electricity — Distinguishes power in kilowatts from energy in kilowatt-hours. Reviewed 2026-11-11. Reference, not endorsement or validation.
- U.S. Department of Justice: Herfindahl–Hirschman Index — Defines the market concentration calculation. This exercise draws no legal conclusion and requires a declared quantity scope. Reviewed 2026-11-12. Reference, not endorsement or validation.
- OECD: OECD Supply Chain Resilience Review: Navigating Risks — Supply continuity and shared dependency context for the fictional planning exercises. Reviewed 2026-11-11. Reference, not endorsement or validation.