| Unit of analysis | Each individual claim inside the specific video or episode you're watching, scored 0-100 | The source overall — outlet, show, or podcast — scored for bias and reliability |
| Coverage of unrated creators | Works on any link, including anonymous TikTokers and channels nobody has ever rated | Covers its rated catalog (3,400+ sources, ~900 podcasts, ~900 TV/video programs); a random creator won't be on the chart |
| Rating method | AI transcription plus automated web cross-referencing — fast, but no human panel | 3+ trained human analysts per piece (left, center, and right-leaning), scored independently then reconciled live; 94,600+ pieces rated to date |
| Political bias measurement | None — BSmeter scores factual accuracy and flags fallacies, not left-right lean | The core product: a published, auditable left-right bias axis refined since 2016 |
| Speed on brand-new content | An episode posted an hour ago can be checked right now, claim by claim | Human panel ratings take time by design; AI-assisted daily ratings exist for business customers |
| Evidence you can click | Linked sources attached to each individual claim, plus deepfake/AI-media signals | A methodology-backed score for the source, rather than per-claim citations for whatever you're watching |