FAQ

Any project that classifies outlets as left, centre, or right is making judgements, and we should own that. The classifications are our own, based on editorial track record, ownership structure, and existing media-bias research. Reasonable people disagree about where individual outlets sit, and we do not claim our classification is the only defensible one.

What we can say is that the Bunting Rating itself is designed to measure divergence between framings, not to identify which framing is correct. A high rating says: these outlets are telling a very different story about the same event. It does not say one side is lying and the other is telling the truth. The AI framing summaries carry their own risks, which we document openly on the Methodology page.

Partly. The framing summaries and synthesis copy in each analysis are written by Claude (Anthropic's AI). That is the step where reading hundreds of articles by hand is not feasible, so we use AI for it and disclose it clearly.

The lean classifications, the outlet selection, the Bunting Rating formula, and the verbatim headline quotes pulled for each analysis are all deterministic or human-authored. The AI writes the framing prose; it does not decide which outlets are left or right, and it does not select which headlines appear. The full breakdown of which steps use AI and which do not is on the How It Works page.

Classifications draw on existing media-bias scholarship, ownership structure, and editorial track record across a body of coverage. We use the standard three-bucket system that journalists and researchers already widely recognise, rather than inventing a proprietary scale. Within each lean, outlets are further classified as corporate or independent based on ownership and revenue model.

Classifications are set by hand, reviewed periodically, and not changed by the AI model at any point in the pipeline. You can see the full list on the Outlets page. If you think a classification is wrong, we are genuinely open to that conversation via the contact page.

The goal is breadth across both lean and type, not exhaustive coverage of every outlet in existence. The list includes corporate giants alongside independent podcasters and newsletters, because some of the most politically distinctive framing comes from smaller publishers that large-outlet-only analyses miss.

Independents are deliberately oversampled relative to their audience share. A list dominated by large corporate outlets would produce a rating that reflects corporate-media consensus as much as genuine political divergence. Including independent voices pulls the analysis toward what politically-engaged readers on each side are actually consuming, not just what reaches the broadest audience.

Yes. Story tips are very welcome. Keep in mind that a story needs to have been covered by enough outlets across the spectrum to produce a meaningful rating: if only one lean is paying attention to it, there is no divergence to measure. But if you have spotted a story that is clearly being framed very differently depending on where you read it, we want to hear about it.

Get in touch via the contact page.