Any single report tells you what its authors measured. Read 254 of them against each other and patterns emerge that none states alone. Six of those patterns, each traceable back to the registry — select any passage to highlight it, and it will be waiting in your saved workspace.
The North American food system produces more calories than its population needs, on less land than it used a generation ago, with fewer and larger farms every year it is measured. Its failures are not failures of production. They are failures of structure — of who owns, who processes, who moves, who eats — and the evidence for that claim is scattered across federal datasets, National Academies reports, and foundation evaluations that rarely cite one another. That is the gap this commons exists to close.
Every per-capita accounting in the corpus converges on the same asymmetry. Beef is about 5% of the food an American eats by mass, yet it accounts for 77% of the cropland footprint, roughly a third of the water footprint, and 40% of diet greenhouse gases. Nothing else in the model is even the same order of magnitude: packaging, food miles, and retail waste are all real and all small by comparison. The corpus is ideologically agnostic on what to do with this fact — it documents regenerative grazing claims and feedlot efficiency gains with equal skepticism — but the arithmetic is not agnostic. Any strategy that ignores the protein portfolio is tuning the radio while the engine burns oil.
Evidence: the per-capita model · beef in the registry · GHG accounting
The 2022 Census of Agriculture counted 1,900,487 farms — a 6.9% decline in five years — on 880 million acres, with the largest sales classes gaining share while every middle class shrank. USDA ERS's own three-decade consolidation series shows the same ratchet in almost every commodity, and it tightens downstream: four firms slaughter most of the nation's beef, a fact that moved from advocacy claim to federal antitrust concern within the corpus's own citation window. Consolidation is not a trend in the data; it is the trend the rest of the data happens inside. Land tenure follows the same slope — farmland values, heirs'-property losses, and beginning-farmer access all read as chapters of one enclosure story.
Evidence: consolidation · meatpacking · land tenure · heirs' property
Between the 1,000-acre commodity operation and the farmers-market stall, the infrastructure of the middle — regional processing, aggregation, cold chain, mid-scale distribution — has thinned to the point of fragility. The corpus's Domain B reads like a coroner's report on it: small meat plants that vanished, grain-milling capacity that centralized, food hubs asked to rebuild in a decade what took a century to dismantle. The success cases that do exist (values-based supply chains, farm-to-institution programs, the cooperative models catalogued in Domain C) share a signature: they are organizationally clever and chronically undercapitalized. For funders, this is the clearest structural gap in the entire registry — the interventions are known, documented, and starved.
Evidence: food hubs · aggregation · values-based supply chains · processing capacity
Domain C closes with eighteen documented failures and supply-chain shocks, and COVID-19 is the load test that exposed what the other seventeen only hinted at. The same consolidation that delivers efficiency in normal times delivered synchronized fragility under stress: meatpacking outbreaks that closed double-digit percentages of national capacity in weeks, milk dumped while food banks queued, and a food-insecurity spike that SNAP's design absorbed better than any emergency program improvised alongside it. The pattern across all eighteen cases is consistent — the system's failures are correlated, not independent, because its capacity is concentrated. Resilience, in this corpus, is not a virtue word; it is a measurable property the system currently lacks.
Evidence: COVID-19 shock · supply-chain disruption · food insecurity
A quieter finding, and for researchers maybe the most important: the federal statistical apparatus that makes a corpus like this possible is itself eroding. The national pesticide-use series ends in 2012. The definitive food-system energy accounting rests on 2001–2012 data with no successor. Broiler-industry structure was last examined in 2014. Aquaculture is split between two agencies whose definitions cannot be reconciled. Where the numbers stop, claims float free — and both advocacy and industry are happy to fill the vacuum. A research commons can flag the rot; only renewed public measurement can fix it. Until then, every synthesis (including this one) inherits those blind spots, and says so.
Evidence: the full gap inventory · pesticides · food-system energy · the platforms that already measure
Run the corpus forward along the current climate trajectory and three constraints tighten together. Water first: the irrigation economy of the arid West is drawing down aquifers against a shrinking snowpack, while the per-capita model shows food already claiming three times the water of all household use combined. Land second: the diet's 1.92-acre footprint meets sea-level, drought, and development pressure on exactly the acres that are easiest to farm. Heat third: the labor the system depends on works outdoors. The corpus's climate-vulnerability assessments, ecoregional baselines, and diet-shift scenarios point to the same conclusion from three directions — the future of food is not a technology question first. It is a question of whether the protein portfolio, the middle infrastructure, and the water ledger adjust on purpose or by force. The sources documenting deliberate adjustment — tribal food sovereignty programs, bioregional planning, urban agriculture at municipal scale — are the weak signals in this registry most worth watching.
Evidence: climate · irrigation & water · tribal food sovereignty · urban agriculture · bioregions
Ranked by how far each moves the system per unit of effort, on the corpus's own evidence: first, the protein portfolio — the only lever with double-digit percentage reach across land, water, and emissions at once; second, middle infrastructure — documented interventions, known operators, a capital gap rather than a knowledge gap; third, land tenure — slow, compounding, and the precondition for every new-entrant and equity outcome downstream; fourth, public measurement — cheap by any standard, and the thing every other lever depends on to be steered at all; fifth, demand-side policy — SNAP and institutional procurement, the two demand instruments the record shows actually working at scale.
This synthesis is regenerated as the corpus grows. Disagree with a reading? Every claim links back to its sources — and the registry accepts submissions.