Every news outlet says it values accuracy. The phrase appears on about pages and in marketing copy across the industry, right next to claims of independence and fearlessness, and it costs nothing to print. What separates an empty claim from a working standard is the machinery behind it: what happens inside a newsroom before a story is published, and what happens after when something goes wrong.
This is a description of that machinery at Digital News Network — what we check, how fast we move, what we do when we get it wrong, and how you can tell whether any of it is real.
What Accuracy Actually Requires
Accuracy is often treated as a single thing, but it decomposes into several disciplines, each of which fails differently.
Getting the facts right. Names, titles, dates, figures, locations and direct quotes have to be exactly correct. These are checkable facts, and there is no partial credit. A misspelled name or a transposed number is not a small error to the person named or the institution affected — it is a signal that nothing else in the story was checked either.
Getting the attribution right. Much of journalism is reporting what people said, and accuracy here means representing their statements faithfully: in context, without trimming a quote until it changes meaning, and without upgrading speculation into assertion. “The official said the decision was under review” and “the decision will be reversed” are not the same sentence, and the distance between them is where most misleading reporting lives.
Getting the proportion right. A story can be factually correct in every sentence and still be inaccurate as a whole. If a minor detail is elevated into the lead, or one side of a contested issue gets three paragraphs and the other gets a clause, the story misinforms without technically lying. Editors carry the main responsibility for catching this, because reporters close to a story frequently cannot see it.
Saying what is not known. Accurate reporting includes accurately describing uncertainty. When casualty figures are preliminary, when an investigation is incomplete, when a claim comes from a single source, the story says so. Confidence beyond the evidence is inaccuracy even when the guess turns out right.
How We Check
Verification at DNN runs along a small set of fixed rules.
Two-source standard for contested claims. Any significant claim that is disputed or unverifiable from documents requires corroboration from at least two sources with independent knowledge. One source can be honestly wrong. Two independent sources wrong in the same way is far less likely.
Documents over recollection. Where a story touches contracts, official decisions, court matters or money, we seek the document. Paper trails are dated, quotable and do not adjust their account when a second call comes in.
Visual verification. Images and video are checked for origin and date before use. Recirculated old footage presented as current events is among the most common forms of misinformation in this region, particularly during protests and severe weather, and the check takes minutes.
Right of reply. Anyone criticised in a story is told what will be said about them and given a fair chance to respond before publication. Their response goes in the story. This is not politeness — it routinely surfaces errors before they become corrections.
A second set of eyes, always. Every article is edited by someone who did not report it. The editor checks facts against source material, challenges attribution, and reads as a reader would: what does this actually say, and is it supported?
Pre-publication review for high-risk stories. Stories involving defamation exposure, active court proceedings, minors, or vulnerable individuals get an additional review. Publishing fast is not a defence available in court, and it is not one we accept internally either.
What Timely Means — and What It Does Not
Timeliness is the second half of the standard, and it needs defining, because it is routinely abused.
Timely reporting means publishing news while it is still useful to the reader: election results as they are declared, weather warnings before the weather arrives, official decisions when they are made rather than days later. It means our publishing cycle is continuous, not tied to a nightly deadline.
Timely does not mean first at any cost. The economics of digital news reward the outlet that publishes ninety seconds ahead of the field, which produces a permanent industry-wide temptation to publish before verification is complete. The result is visible daily: dramatic claims that shrink by evening, casualty figures that triple and halve, stories that are quietly rewritten until they say the opposite of what they said at noon.
Our position on this is fixed. We publish as fast as verification allows, and no faster. Where that means a competitor runs something before we do, we accept it. A story marked as developing carries only what is confirmed at that moment, attributed to its source, and is updated openly as more is verified — with timestamps, so the reader can see the story’s own history.
The hardest version of this discipline is the breaking event where early reports are dramatic, plausible, widely shared — and unconfirmed. Refusing to repeat them costs traffic in the short term. It is also the entire basis of the trust that makes the traffic worth having.
When We Get It Wrong
No verification system catches everything, and a newsroom that claims otherwise is telling you it does not check its own record. What matters is the response.
Corrections are visible. When we fix a factual error, a note on the article records what was wrong, what it now says, and when the change was made. Significant errors are flagged prominently rather than at the foot of the page.
We do not silently delete. Removing a story without explanation erases the record. If an article is withdrawn, a note explaining why remains at the URL.
Readers are part of the system. Some of our most valuable corrections come from readers who know the subject better than we do. Reporting an error is easy, and every report is reviewed by an editor.
We audit ourselves. Periodically we review published stories against how events actually unfolded — which early claims held, which sources proved reliable, which framings aged badly. That review is uncomfortable and is one of the few mechanisms by which a newsroom genuinely improves.
Accuracy in the Age of Generative AI
The economics of fabrication have collapsed. A fluent, professional-looking article on any subject can be generated in seconds, and synthetic images pass casual inspection. This raises the value of exactly the things that cannot be automated: a reporter who makes phone calls, an editor who knows the beat, a corrections record that stretches back years.
We use AI tools where they are useful — transcription, translation, data processing — under two fixed rules: a human editor remains accountable for everything published, and material AI involvement is disclosed. We do not publish machine-generated articles under human bylines, because a byline is a promise that a person stands behind the work.
How to Test Us
A commitment to accuracy is only credible if it is checkable. Ours is.
Look at our corrections — a publication with none is not one that makes no mistakes, but one that does not admit them. Check whether our stories link to the documents they describe. Compare what we reported during a developing event with what turned out to be true. Look at whether our headlines are supportable by the articles beneath them.
That is the test we apply to ourselves, and we would rather you applied it to us than took our word for anything.
