Readers inherit the finished article and never see the argument that produced it. The argument is the interesting part: inside a working newsroom, publication is not the default but the end of a series of rejections, and understanding those rejections is the most useful media literacy there is. The verification standards that decide what reaches print, drawn from the published ethics codes and practices of the major wire services and broadcasters, are worth knowing from the outside.
Sources are ranked before they are believed
The first thing a desk does with a claim is place its source on a ladder. The rungs look roughly like this from top to bottom: documents created in the ordinary course of business, such as filings, court records and official registries; official statements and on-record interviews; multiple sources with direct knowledge speaking on condition of anonymity; a single anonymous source; and, at the bottom, secondhand and hearsay accounts.
Position on the ladder is not about honesty. A truthful official can be mistaken, a document can be forged, and a single anonymous source can be the most important person in the story. The ladder encodes verifiability: the higher the rung, the more a reader can check without trusting the journalist. That is why statements about what a company owes investors come from filings rather than from what employees say, and why serious reporting on a government's budget starts at the budget document, not the budget press conference.
The two-source rule, and when it bends
The best-known convention in the trade is that a consequential claim needs two independent sources. Two points of craft hide inside the word independent. Sources must not share information with each other, and they must not share a single origin: if three officials all heard the same briefing, the rule's spirit counts that as one.
The rule flexes with the type of claim. A public speech, a vote shown live, a document anyone can download: these are verified by observation, not corroboration, and no outlet sends a second reporter to rewatch a press conference. Allegations of wrongdoing move the opposite way: they harden the requirement, because accusations carry reputational stakes that a mistake cannot undo. This is why documented allegations often take weeks to reach print while the rumour version circulated on day one. The delay is the process.
Documents get interrogated, not just read
A document is evidence only after someone has tried to break it. Standard practice includes confirming who created it and whether that creator could have; checking internal consistency, dates, names and formatting against known authentic examples; tracing its custody, who held it and how it arrived; and testing fragments against independent records, whether the bank statement's dates match the court filing, whether the email's server headers fit its story.
The reason is that document forgeries specifically target newsrooms, and several historic scandals in the profession came from exactly this failure. Outlets that survived them institutionalised the paranoia. The reader's version of the same habit is simpler: when a story turns on one explosive document, check whether the reporting describes how it was authenticated. Outlets that did the work usually say so, because it is expensive and they want credit for it.
Anonymous quotes pass through more hands than you think
The negotiated anonymity visible in print, "a person familiar with the matter", sits on top of an internal process readers never see. Typically: the journalist must know who the source is; the editor must be told, or told enough to judge; the reason for anonymity must meet a published standard, usually that the information is genuinely important and unavailable on the record; and the description of the source must be as narrow as protection allows, because "senior official involved in the negotiation" is more informative to readers than "sources said".
That last phrase, "sources said", with no description at all, is the weakest formulation in journalism and most strong outlets now discourage it, because it asks the reader to extend trust without any basis for doing so. Its presence is a small but real signal about a publication's standards.
What true-but-unpublished looks like
Stories fail verification in characteristic ways, and knowing them explains some of what never reaches print. The sourcing is real but single, and the stakes demand more. The documents are authentic but their meaning is contested, and no independent expert will go on record. The allegation is credible but the accused was never given the chance to respond, and fairness rules out running it anyway. The story is true but it is old, and the outlet that publishes it without the freshest developments will be scooped by its own delay.
None of these are conspiracies. They are the daily triage of a system built on the assumption that being first and being wrong are the same failure, which they are not, a lesson the industry learned expensively and encoded in its standards. For the reader, the practical translation is this: an outlet that sometimes says "we could not independently confirm" is showing you its working, and that visible hesitation is worth more than any amount of confident speed.