How we rate confidence and on-track status.
Confidence and on-track verdicts are not opinions; they are defined by the framework below. If you find a claim where the framework was misapplied, that is an editorial error and we want to know.
Confidence bands
High confidence — The claim is supported by two or more peer institutions, or is observed (e.g., historical population data). Replaced only when sources update.
Medium confidence — The claim is supported by one primary institutional source (IPCC, IEA, UN, OECD, IMF, WHO, NASA, BNEF, PwC, Goldman, McKinsey, Lancet GBD or equivalent). Subject to revision when the source revises.
Low confidence — Sources disagree, the claim is extrapolated beyond cited sources, or the claim depends on assumptions outside the source\'s scope.
On-track chips
✓ On track The current trajectory is consistent with the projection. Verified by observed data or by a primary source\'s recent progress assessment.
⚠ Slipping The projection is still possible but the current trajectory falls short by at least one credible measurement.
✗ Off track The current trajectory is inconsistent with the projection. Reaching the projected outcome would require a discrete policy or technology change.
The 5-question verification checklist
Every claim is checked against the following before publication:
- Is the source a primary institutional report (not a press release summary)?
- Is the scenario tag present and correct (SSP, NZE, ETS, baseline, etc.)?
- Is the time horizon explicit (2030, 2050, etc.)?
- Is at least one corroborating source available, or are we explicit about single-source reliance?
- Has the source been updated since publication? If yes, the latest revision is used.
Pathway vs projection
IPCC SSP scenarios, IEA Net Zero pathways and BNEF Net Zero Scenarios are pathways, not forecasts. They describe what would be required for a given outcome (1.5°C, 2°C). BNEF\'s Economic Transition Scenario and PwC\'s World in 2050 are projections: what is most likely given current dynamics. We always state which one we are quoting.
Refuted vs unsupported
A claim goes on the Refuted canon when (a) it is widely repeated in public discourse, (b) it lacks any primary institutional support, and (c) at least one credible institutional source provides a counter-finding. "No source" alone does not earn refuted status; "widely cited despite no source" does.
Update cadence
Major sources are reviewed quarterly. Pillar updates triggered by source revisions are published within four weeks. The "last revised" date at the top of each pillar reflects the most recent editorial change, not just the source date.