Methodology

How a fact gets in

We do not store profiles. We store statements — one fact, about one subject, with one source. A record is just the set of statements currently attached to it.

A statement cannot exist without a source. This is enforced in the software, not by policy: the function that writes a statement refuses any call that does not carry one. If research produces a fact we cannot source, we discard the fact. Missing beats unsourced.

What confidence means

Every statement carries a confidence between 0 and 1, set by how the source states it — not by how much we believe it.

• 1.0 — quoted directly and unambiguously from the source.
• around 0.6 — the source says it, but vaguely (“about 250 staff”, “around 20 offices”).
• 0.3 or below — only implied. These are held for review and are not shown as facts.

Anything below 0.7 is labelled reported, not verified on the page where it appears. Right now 24 of 447 published facts carry that label.

What we verify, and what we do not

Most facts here are reported, not field-verified. We have read the source and confirmed it says what we claim it says. We have not visited the site, counted the trees, or audited the accounts.

Two figures deserve particular scepticism, ours included:

• Survival rates. The most abused number in this sector. We refuse to record one without a stated method and year — if a source gives a rate with no method, it goes to review rather than onto the page.
• Planting totals. Frequently self-reported, rarely independently checked, and often counted more than once when several organisations claim the same site. Treat any total here as a claim by its source, not as a measurement.

Who gets published

Organisations, companies, government bodies and officials acting in a public role are published as a matter of course — these are business and public facts.

A private individual is never published automatically. Their record stays in draft, invisible to the public, until they claim it themselves. This is enforced twice: once in the application, and once as a database rule that refuses the change even if the application is bypassed.

We record nothing about a private individual beyond their tree and climate work — no home address, no date of birth, no family, health, religion or politics.

Corrections and removal

Anyone described here can correct their own record. Corrections do not overwrite anything: the old statement stays, marked as superseded, and both remain visible in the record’s history. Nothing on this site is silently edited.

Removal requests are honoured the same day, with no review step and no questions. A removed record has its values redacted and is withdrawn from public view immediately.

Known gaps

• Coverage is currently concentrated in Pakistan. It is not a world record yet.
• Nurseries and small growers are badly under-represented. Most operate informally and have no web presence, and some have real regulatory reasons not to want one. Reaching them needs field contact, not research.
• Records sourced only from encyclopedias are weaker than those sourced from an organisation’s own publications. Where that is the case, the source line says so.
• Absence from this record means nothing. It is far more likely we have not got to someone than that they do not exist.

Licence

Facts and sources are published under CC BY 4.0 — free to reuse with attribution. Being reusable is how a record becomes infrastructure.