How we work
A calm explanation, a working model and every source, so you can understand it properly and check us.
Sourcing
Every number links to a primary source: the agency, dataset or paper it came from, not a news story about it. Each explainer ends with its full source list, the definitions and baselines we used, and the date it was last updated.
Checking
Before an explainer goes live it has to pass five questions:
- Would we be proud to show it to the person whose data we used?
- Could a skeptical reader look for the weak spot and not find it?
- Does a stranger understand the main idea within three minutes?
- Does every interaction teach something that reading alone would not?
- Is there anything the piece would be better without?
Data and models
Each explainer says which figures are measured data and which are a model's simplification. A model is never presented as a forecast, and you can reset it to what actually happened with one tap.
What we don't know
Every explainer has its own section on the open questions and the evidence that would settle them. We would rather say "researchers don't yet agree" than guess.
How we use AI
We use AI tools to help build and draft. Anything a model produced is checked against its source before it is published. Realistic AI-generated images and video are labelled as such.
Independence
Nobody pays us to cover a topic or to change what a piece says. There are no ads or sponsored placements on this site. If that ever changes, this page will say so, and anything sponsored will be clearly marked.
What we won't do
- Write headlines, thumbnails or hooks that promise what the piece doesn't deliver.
- Use fear or outrage to hold your attention.
- Take sides on contested politics. We show the positions and the evidence for each.
- Give medical, legal or financial advice.
When we get it wrong
We correct in public and keep a record. See Corrections for how to report a mistake and every change we have made.