The Correction

A wrong fact in a published story does not sit still. News sites copy stories from each other, so the mistake gets copied too — even after you know it is wrong. Fixing your own page is only the start.

A real correction has three parts. It says what the story got wrong. It says what is actually true. And it appears where readers will see it, as soon as possible — because every hour of waiting means more copies.

A correction is a published notice that repairs an error in the record, and it has three properties. Prompt: posted as soon as the error is confirmed, because outlets and aggregators copy each other and every hour multiplies the copies. Prominent: placed where the original audience will actually encounter it — on the story itself, and through the same channels that carried the error. Specific: it states what was wrong, what is true, and how the mistake happened.

Notice what a correction is not. It is not silently rewriting the paragraph and hoping nobody kept a copy. And it is not a vague line of regret — 'we regret the error' repairs nothing if the reader cannot tell which fact to replace in memory.

The propagation problem is structural. Wire services, aggregators, and rival outlets republish each other's reporting on their own schedules, and none of them re-verify a routine fact. Once your figure has been picked up, you no longer control it: each copy cites you, and readers of the copy never visit your corrected page.

That is why correction practice is codified. The note stays attached to the story permanently; it names the wrong fact and the right one; it runs through every channel the error ran through, including a note to outlets known to have picked the story up. It also states how the error happened, because the cause tells readers whether to expect it again — a misread budget line is one kind of failure, an unchecked source another.

You told your friend the party is on Saturday. That was wrong. It is on Sunday. Your friend already told two more kids, and they told others. The wrong day is traveling without you.

To fix it, you must do three things. Tell everyone fast. Say it where everyone can see. And say exactly what was wrong: not Saturday — Sunday.

Research on misinformation gives corrections a sobering baseline: the continued influence effect. Readers who see a corrected claim often go on reasoning with the original error, especially when the correction supplies no alternative account. Corrections that work state the true fact and explain how the false one entered circulation — filling the gap the retraction leaves rather than merely deleting the claim.

Practice also distinguishes instruments. A correction fixes an error of fact. A clarification repairs a true but misleading statement. An editor's note addresses failures of process — an undisclosed conflict, a fabricated source. Silent revision, editing the story with no note, is treated as a breach in itself, because it falsifies the public record of what was published.

Two literatures meet here. Cognitive work on the continued influence effect (Lewandowsky and colleagues) finds retractions decay poorly unless a causal alternative is supplied, and that repeating the myth inside the correction can re-strengthen its familiarity — an effect later replications soften but do not erase. Institutional work treats corrections as version control for the public record: persistent errata, visible revision histories, archived prior versions.

There is also a legal register. In many U.S. states, retraction statutes limit recoverable damages in defamation when a prompt and conspicuous correction runs — quietly aligning the newsroom's commercial incentives with its epistemic duties. The lesson's distinction between accountability and reputation management is thus partly institutionalized: what is prompt, prominent, and specific is also what courts and readers each read as good faith.