How we rate sources — tier, confidence, and bias risk

Every entity and event on this site carries three small marks. They look alike, but they answer different questions. A page can rest on an excellent source and still leave us only moderately sure of the claim, and a page can be well-settled even though its sources are ordinary. Keeping the two apart is what lets you weigh a page for yourself.

Tier: how strong the source is

Tier rates the source, not the claim. It asks how close the evidence sits to the thing it describes.

• A. Primary or official record: a statute, a Riksdag document, a court ruling, official statistics, a government page.
• B. Strong secondary work: peer-reviewed research, a major archive, an established reference with its own citations.
• C. General reference and quality press: an encyclopaedia entry, a reputable newspaper.
• D. Weaker secondary: opinion, advocacy, or a single outlet with no corroboration.
• E. Weakest or contested: anonymous, partisan, or hard to verify.

A page shows the best tier among its sources, and each source in the list carries its own tier so you can see where a claim actually rests.

Confidence: how sure we are of the claim

Confidence rates us, not the source. Given what the sources say, how sure are we that the page’s claim is right?

• A. Settled: several strong sources agree and there is little room to dispute it.
• B. Solid: well supported, but some interpretation or a small gap remains.
• C. Provisional: it rests on a single source or a contested reading.
• D. Tentative: plausible but thinly sourced, and flagged as such in the prose.

This is why tier and confidence can diverge. One official document can be tier A yet leave a question genuinely open (confidence B). A dozen consistent newspaper reports can be tier C yet settle a fact beyond doubt (confidence A). We never publish a claim we would rank below D, and a rejected claim does not reach a public page at all.

Bias risk: how much slant we had to weigh

Bias risk flags how loaded the topic and its sourcing are, so you read the page with the right caution.

• Low. The sources are neutral or cross-checked, and little turns on whose account you trust.
• Medium. The subject is contested, or the available sources lean one way and we had to balance them.
• High. The record is mostly partisan or one-sided, and the page reflects that limit openly.

Bias risk does not pass judgement on the party or the event. It is a note about the evidence, a reminder that on some subjects even careful sourcing carries a slant worth naming.