Automation does the repetition. People own the judgement. Every published record states the highest verification tier it has actually reached, and never a tier above it.
We can show you exactly where this number came from.
The original publisher, the exact document, and the page, table or dataset identifier are recorded. Aggregators, news write-ups and reposts are never accepted as the source of record. A record stops at this tier when the underlying document exists but has not yet been checked line by line.
We have checked the number against an archived copy of that document.
The value, its definition, its reference period and its stated method are read out of an archived copy of the authoritative document. The archive is fixed with a SHA-256 checksum, so a later silent edit upstream is detectable. Extraction from tables and maps is checked independently rather than trusted to a single pass.
We recalculated it ourselves and got the same answer.
Where the source releases usable microdata, the estimate is recomputed with the focus’s declared instrument — complex-sample survey estimation, panel models, gridded climate data — and compared against the published figure. Disagreements are published, not resolved quietly in our favour.
Collection can run unattended. Publication cannot. Each focus declares the gate a record must pass before it becomes public, and the most sensitive focus is never fully automatic.
| Focus | Gate | Cadence | Instrument |
|---|---|---|---|
| Economy & Markets | Methodologist gate | Weekly plus official release days | statsmodels |
| Public Money & Governance | Two-person gate | Weekly and monthly allocation releases | Pandera |
| Health & Nutrition | Methodologist gate | Wednesday plus surveillance releases | R survey |
| Education & Youth | Methodologist gate | Thursday plus term and annual releases | R survey |
| People & Society | Two-person gate | Saturday with separate survey labelling | R survey |
| Infrastructure & Services | Methodologist gate | Friday plus regulator releases | statsmodels |
| Food, Agriculture & Environment | Methodologist gate | Weekly with seasonal monitoring | xarray |
| Safety & Civic Life | Two-person gate | Release-led; never fully automatic | statsmodels |
A correction is appended to the record, dated and explained. The original figure stays visible. Statistones does not silently edit a published number, because a reader who cited last month's figure needs to know what changed and when.
If you believe a published record is wrong, write to corrections@statistones.com with the record URL and the evidence. Corrections that hold up are published whether or not they are convenient.
It is not a source. Every number here belongs to the body that produced it, and the citation on each record names that body first. It is not a forecaster: no value on this site is modelled, projected or interpolated between observations. And it is not a scoreboard — a high rank is not a good result or a bad one until you have read the indicator's definition.