Where these facts come from, and how they are checked.

Everything on this site is built from the public record: the rule an agency published, the docket it answered, the opinion a court issued, the release the Federal Reserve put out. This page explains where those documents come from, what is checked before an article publishes, and what we do when we get something wrong.

The document is the source.

A story here starts from a primary document, not from someone else's article about one. In practice that means the Federal Register notice, the docket the agency opened, the text of a final rule, a court's opinion, a report from an inspector general, or a statement from the Federal Open Market Committee. Those are the records that prove an event happened, and they are what our outbound links point at.

This is possible because of one property of the beat. Works of the United States Government carry no copyright, so a rule, a filing, an opinion or a Fed statement can be quoted at length and restructured into something a non-specialist can follow. That licence covers the government's document. It has never covered another outlet's article about the document, which is why you will not find one reproduced here.

We do read widely. Coverage from other outlets is used the way a newsroom uses a wire scan: to notice that something happened, to tell a consequential document apart from a routine one, and to corroborate a fact before it is stated. It is input to the decision about what to cover. It is never the output, and we do not send readers to it.

The vocabulary these documents use

What has to be true before anything publishes.

The unit of publication is the event, not the article that mentioned it. Five outlets writing about one rulemaking is one story here, and collapsing them is the step where most errors surface: when two accounts disagree about what a rule requires, the disagreement is visible, and the document settles it. Where it cannot be settled, the article says so rather than picking the more confident version.

Dates get particular attention because they are the most commonly mangled fact in federal coverage. A proposal is not a requirement, a comment deadline is not an effective date, and a rule that publishes today may not bind anyone for another year. Each of those is a different date in the same document, and an article that blurs them is wrong even when every word in it is accurate.

We publish on a quality bar rather than a schedule. An event earns an article when a primary document exists or when independent sources corroborate it, which means a quiet day stays quiet. A daily quota is precisely the pressure that manufactures thin pages, and a thin page costs more than a missed one.

Four questions asked of every article.

None of these is exotic. They are the checks a careful desk editor would apply, written down so they are applied the same way every time. Anything that fails one of them does not publish, and the glossary explains any term an article leans on.

Is there a document?

A named, retrievable record - a Federal Register citation, a docket number, a case number, a Fed release. If the only evidence is another outlet's summary, it is not ready.

Does the document say it?

The claim is checked against the text, not against the headline the text was reported under. Preambles are read, because that is where an agency explains what it rejected and why.

Which date is it?

Proposed, published, effective, and comment-close are four different dates. The article names the one that matters and does not imply the others.

What changed?

An article that cannot say what is different afterwards is describing process, not news. If nothing changed, it is a notice, and we treat it as one.

What this method cannot tell you.

Working from documents means we are as current as the record and no more. A decision made in a meeting is invisible here until it produces a filing, and the gap between the two can be weeks. Where an outcome is obviously coming but nothing has published yet, the honest position is silence, and that is the position we take.

It also means we cover what generates paperwork, which is not the same as what matters. An agency that quietly stops enforcing a rule creates no docket and no Federal Register notice, so a change with real consequences can leave almost no trace. Coverage anchored to documents is verifiable and incomplete at the same time, and pretending otherwise would be the more comfortable lie.

Finally, we do not know what a rule will do. An article can tell you what an agency decided, what it said its reasons were, and what changes on the effective date. Whether the thing works is a question for the people it lands on, and answering it is reporting we have not done. Where that boundary matters, the standards page says plainly what this publication does and does not claim.

Reader questions, including how to send a correction

What the federal government did, once a day.

One email each weekday: the rules, filings, votes and Fed decisions that actually changed something, each one linked to the document it came from.

SubscribeRead today's front page