- Primary sources only
- Claims are traced to the organisation that made them - Google's own documentation for Chrome behaviour, Mozilla's release notes for Firefox, the published paper for a research finding. We do not cite blog posts summarising a source when the source itself is public, and we never cite ourselves as evidence for a fact.
- Every source is published
- Articles carry a reference list at the end, and each entry names the specific claim it supports rather than sitting there as decoration. If you think we have read a source wrong, you can check it in one click.
- The build enforces it
- A check runs before the site can be deployed. An article with no sources fails it, a source we cannot categorise fails it, and citing this domain as evidence for a factual claim fails it. Articles that legitimately need no sources have to say so explicitly rather than pass by omission.
- Corrections are dated, not silent
- When an article changes materially it gets an updated date shown at the top of the page, and the change is described in the text rather than edited in quietly. Where we removed a figure because we could no longer stand behind it, the article says that we removed it.
- No invented numbers, no AI-generated imagery
- Statistics are quoted from a named source or they are not published. Diagrams, screenshots and cards are generated from the real extension and from real data by scripts in this repository, so a screenshot of the product is a screenshot of the product.
- Photography is licensed, and it is only decoration
- The photographs on this site - hero backgrounds like the one on this page - are stock photography licensed through Envato Elements. They are used as decoration, never as evidence: no photo on this site depicts our team, our office, or a test we ran.