By search — a continuing sweep of federal, academic, and foundation publishing — and by contribution. If you make grants, or hold them, the reports you commission belong in the common record. Submission is free; inclusion is earned on the rubric below.
For grant-making entities, grantees, agencies, and researchers. Every submission is fetched, read, and graded before it enters the registry — the same treatment as every existing source.
The corpus covers the United States (all fifty states, tribal nations, and territories) as primary geography, Canada as secondary, and Mexico where continental supply chains cross the border. Global sources appear only where they define North American baselines. Coverage spans seventeen domains, from bioregions and production through processing, distribution, consumption, waste, externalities, economics, policy, labor, health, and the documented record of failures and successes.
Every entry is graded: A — peer-reviewed journal article, National Academies report, or primary federal dataset; B — state land-grant extension, state agency, foundation-commissioned peer-reviewed work, credentialed think tank; C — foundation grantee report or established nonprofit research with disclosed methodology; D — journalism, used for framing and contemporary case detail only. A fifth category, X, is excluded outright: advocacy without methodology, industry PR without primary data, unverified aggregations. Every quantitative claim on this site traces to an A or B source; C sources support qualitative case framing; no D source ever anchors a number.
The rule that makes the registry trustworthy is procedural, not editorial: every value in every entry was extracted from a page fetched during compilation. Nothing was carried in from memory, and no URL was constructed rather than followed. Where a fetched page did not state something, the entry says "n.a." rather than guessing. The corpus does not privilege any food-systems ideology — regenerative, conventional, or agroecological — and it is not tied to any place-based project.
The Phase 1 build (20 August 2026) holds 229 verified entries with 223 unique URLs, 100% populated. A stratified 45-URL live check returned 84% clean; of the seven flags, two were real 404s (both with documented replacements pending), three were anti-bot 403s on pages that served normally to the compiling agent, one was a habitually slow federal PDF, and one was fixed during the registry build. Expected URL rot for a federal-heavy corpus is roughly 5% a year — a scheduled re-fetch is the first item on the Phase 2 list. A same-day addendum added Domain M: 25 meta-resource entries documenting the databases, dashboards and platforms that already measure the food system — bringing the registry to 254 sources from 132 organisations (69% Grade A).
The grade audit stands at 161 A (70%), 32 B (14%), 34 C (15%), and 2 D (1%) — the two journalism entries are each paired with primary sources for their factual anchors. The per-capita models carry a file-by-file honesty log: which values are n.a. and why, where independent estimates converge (bottom-up and top-down land footprints agree within 1.5%), and where they diverge and by how much (two published GHG bases differ by 60%; the corpus documents its choice). The full report ships with the corpus repository.
Phase 1 — done. Scoping, taxonomy, rubric; three domain bibliographies; machine-readable registry; five per-capita models; verification report. Phase 2 — next. URL re-fetch and repair; the PDFs behind landing-page-only entries; a foundation-grantee retrieval sweep; a ten-state agency deepening; the energy-model refresh; six thematic dossiers; long-form case studies. Phase 3 — ongoing. Annual refresh on the Census of Agriculture cycle, each Farm Bill cycle for policy, ad hoc for foundation reports — plus everything the submission pipeline brings in.
Let's Begin Here: Food Systems is Volume Three of the Let's Begin Here series of research commons — Volume One covers rural revitalization; Volume Two, climate & rural impacts. It is a not-for-profit project by Re:Vision & The Rural Institute, ©2026. The corpus and registry are built for long-term reuse: stable IDs, database-ready exports, and a public methodology. Registration, when it arrives, will gate the workspace tools (saving, collections, analysis, PDF export) — the reading, searching, and citing will stay open.