Articles on Reasonable Outrage are researched, written, and edited by an AI system under the editorial direction and review of a human publisher. Every factual claim is checked against a stored copy of its source before publication. We are explicit about this because a site that asks you to check its sources should not be coy about its own methods.
What that means in practice
An automated research step finds a story in primary documents — rulemaking dockets, court filings, federal spending records, securities disclosures, official statistics — and stores a copy of every source it uses. A drafting step writes the article from that stored material and nothing else. A verification step then checks the finished article back against those stored sources, claim by claim, and rejects anything it cannot trace. A human reviews and approves before publication.
The stored copy matters. It means the text we checked against is the text that was actually retrieved, not whatever the page says later.
What the outrage score means
The number at the top of each article measures how warranted anger is on the evidence. It does not measure how heated the writing is. A 2 means the alarm going around is not supported by the record. An 8 means something serious is documented and largely uncovered. A 9 or 10 should be rare; if everything is a 9, the scale is broken.
Low scores are the point as much as high ones. Some of what we publish exists to defuse a panic rather than raise one.
What we will not do
We do not assert what anyone knew, intended, or designed. We lay out documented conduct, the sequence it happened in, and who ended up better off, and we let you draw the inference. Where intent is genuinely in the record — an email in a filing, sworn testimony — we quote it and cite it precisely.
We do not write about private individuals. Public figures are discussed in their public capacity only: what they did in office, in filings, on the record.
We do not reproduce substantial amounts of anyone else’s text. We summarise and link.
Checking our work
Every article ends with its sources, and the markers in the body link to the specific source each claim rests on. If a source does not say what we said it says, that is an error and we want to know.