Show the reasoning
The interval is a percentile band over the Monte Carlo draws — both the loss factors and the mixing pool sizes are resampled from their recorded lo/mode/hi. The evidence filter drops loss stages weaker than the grade you pick, which shows how much of the answer depends on figures we could not source.
Where the uncertainty in the requirement comes from
Each bar swings one stage across its own lo/hi with every other stage held at its mode. The longest bar is where more research would pay off most. Mass-only stages register zero — they cannot move a count however uncertain they are. This chart is about the requirement; the panel below asks a different question, by a different method, about the number on the plate.
What drives the number on the plate
A different question from the tornado above. That chart swings one loss stage at a time and reports individuals required; this one varies every mixing input at once and reports each one's share of the variance in distinct. It has to: the mixing model is not a product of factors, so its inputs interact, and one-at-a-time would miss exactly the interaction that matters.
Bars are first-order (this input alone) and total-order (including everything it interacts with). Total above first is interaction. Whiskers are a bootstrap over the same draws. Every input here is estimate-grade — none is a sourced measurement — so the whole chart renders in one colour, and that is the point rather than a fault.
Distribution of the answer
How the floor becomes the requirement
Every count-affecting stage, in order, compounding on the one before it. This walk uses each stage's mode, so it lands slightly below the headline figure above, which is the mean of the Monte Carlo draws. The two differ because the lo/mode/hi bands are asymmetric — several stages have more room to go wrong than to go right, which drags the mean above the mode. That gap is a real feature of the data, not a rounding artefact.
Evidence behind the count
Counting only stages that can move the count. Mass-only stages are excluded because their evidence quality is irrelevant to this answer.
The mixing simulator
Drag the pool size. This is the whole argument of the project in one control: six chickens is only reachable if you cut up six chickens yourself. Add any pooling at all and the count runs for the ceiling.
Average broiler live weight by state
Ohio runs 4.6 lb birds. North Carolina runs 8.4 lb birds. Same species, same year, different businesses — small-bird programs feed fast food, big-bird programs feed deboning. States the survey suppresses under its disclosure rules fill in gray from the census — presence, not weight.
Production by country
Different countries publish different things, and the gaps are not symmetrical. Pick one to see what its own statistics can and cannot answer — including, where it applies, that they cannot answer this project's headline question.
National series
Subnational
Sources for this country
Does size make it better?
Every species is graded on something, but not on the same thing, and the answer does not carry across. Pick one to see its own axis — and, where the corpus cannot answer yet, to see that stated rather than filled in.
—
The axis
What more of it costs you
Nutrition
Per 100 g from USDA FoodData Central, with per-piece values derived from segment mass rather than stored separately.
Resource footprint
Two numbers per metric, and the gap between them is the point. The naive figure charges whole birds to your order. The allocated figure charges only this product's share of the bird, because the breast and thighs fed other people.
Who got paid
Year over year
Birds get heavier and more feed-efficient every year — and they also die more often than they did in 2013.
Does bird weight have a season?
Every state, and whether its swing means anything
Sorted by how much the shape of the year can be trusted, not by the size of the swing — because sorting by size puts the least trustworthy states first. Hover a verdict for the reasoning.
What this does not say
—
Arrow keys or swipe to move between cards.
Where the sources disagree
Some of the most honest things in this dataset are the numbers that do not agree. Where two credible sources give different figures for the same quantity, averaging them would invent a precision neither supports and picking one would hide a real uncertainty — so the disagreement is recorded as data, with both sides cited, and neither figure is fed into the model.
Every number here has a citation
Where a figure could not be sourced it is marked as an estimate rather than dressed up as a fact. That distinction is the point. The full standard this site publishes against is written down in the SwampLink data policy.