You open a story. It says a thing you weren't expecting. You do the responsible thing — you check whether anyone else is reporting it. Three other outlets are. You relax. * You have just been had, possibly. Because "three outlets are reporting it" and "three outlets found it out" are completely different sentences, and the news does an extremely poor job of telling you which one you're looking at. * ### The actual definition * **Circular reporting is when a single unverified claim gets repeated by several outlets, each citing the last, until the volume of coverage looks like independent corroboration.** * One source. Four articles. Zero additional verification. The claim didn't get truer; it got louder, and loudness is what most of us are using as a truth test. * It matters because the ordinary reader's best instinct — *don't trust one source, see if it's widely reported* — is precisely the instinct this exploits. Circular reporting isn't a failure of scepticism. It's a failure that happens **to** sceptics. * ### The Wikipedia version has its own name * The tightest, most demonstrable loop runs through Wikipedia, and it has a name: **citogenesis**, coined by cartoonist **Randall Munroe** in xkcd #978 in **2011**. * The loop has four steps: someone adds an unsourced claim to Wikipedia. A journalist on deadline reads it and prints it. A Wikipedia editor notices the claim has no citation, finds the news article, and adds it as the source. The claim is now **sourced to a professional publication** — which got it from Wikipedia. * The circle is closed. Nobody lied. Everybody followed the rules. The fact was born out of nothing and now has a footnote. * Wikipedia takes this seriously enough to maintain a page called **List of citogenesis incidents**, and it applies a strict test: a Wikipedia error simply being copied by the media doesn't count. It's only citogenesis when that media report is then cited **back** in the article to support the claim. * ### Two cases that actually happened * **The composer who never said it.** In March 2009 the French film composer **Maurice Jarre** died. **Shane Fitzgerald**, then a final-year sociology and economics undergraduate at **University College Dublin**, wrote a quote that Jarre never said, put it on Jarre's Wikipedia page shortly after the death, and waited. * The quote — a graceful little thing about his life being one long soundtrack and a final waltz playing in his head — ran in obituaries in **the Guardian**, **the Independent** in London, on the **BBC Music Magazine** website, and in newspapers in India and Australia. * Fitzgerald said the fabrication took him **under fifteen minutes**. The detail that should stop you is *why* it worked: he wrote it to be exactly what an obituary writer on deadline would want, and did not contradict anything true about Jarre's life. It didn't set off anyone's alarm because it wasn't designed to. It sat undetected for **weeks** — until Fitzgerald emailed the papers himself to tell them. The Guardian corrected it. * Nobody caught it. The correction came from the hoaxer. * **The politician who grew a name.** Also 2009: an editor added the middle name **"Wilhelm"** to the Wikipedia entry of German politician **Karl-Theodor zu Guttenberg** — a man who already had one of the longest legal names in European public life, and none of it was Wilhelm. * German and international outlets printed it. And then the published articles were produced as **reliable sources** in the argument for keeping "Wilhelm" in the Wikipedia entry. The forgery had, in a matter of days, acquired references. * ### Where it happens outside Wikipedia * Wikipedia is just the easiest place to watch the loop, because both ends are timestamped in public. The same mechanism runs through ordinary news all day: * **"According to a report in..."** is the honest version. That phrase is a newsroom telling you plainly that it did not confirm this. It is not a second source. It's a citation. * **Wire copy.** When forty outlets carry the same story, that's usually one wire reporter, syndicated. Not forty confirmations — one, published forty times under different logos. * **One official, several outlets.** A single anonymous official can brief three reporters in an afternoon. Three "independently, sources tell us" stories can be one person's Tuesday. * **The aggregator layer.** A site with no reporters summarizes a story, another site summarizes the summary, and by the fourth hop the qualifier ("may," "is expected to," "one draft proposed") has evaporated and the hedge has become a fact. * **The shell-site layer.** A partisan outlet publishes a claim; a network of sites dressed as local newspapers repeats it; a campaign then cites "local news reports." That's circular reporting used deliberately as a laundering machine — the reason those sites exist at all is to be quoted. * ### How to break the circle in one minute * **Follow the attribution backwards, not sideways.** Don't ask "who else is reporting this." Ask **"where did this one get it."** Open all four tabs and look at what each cites. If they all trace to the same original, you have one source, not four. * **Find the primary document.** Court filing, transcript, filing, press release, dataset. If a story is about a report, the report has a name — read the report. Everything between you and the document is a retelling. * **Check the clock.** If four outlets published within twenty minutes of each other, that is not four investigations converging. That's one press release, or one wire, or one embargo lifting. * **Watch for a hedge that vanished.** If the earliest version says "a draft proposal would," and the newest says "the government will," the change happened in transmission, not in reality. The oldest version is usually the most accurate one. * **On Wikipedia, click the footnote.** Not the number — the actual source. Then check the date it was published against the date the claim was added to the article. If the article predates the source, look harder. * ### The uncomfortable part * Circular reporting is rarely anyone's fault in a way you can point at. Nobody in the Jarre case did anything wrong except trust a reference work on deadline, which is what deadlines are for. That's why it keeps happening: every individual step is reasonable, and the system still produces a fact out of nothing. * Which means the fix cannot be "trust better outlets." It has to be a habit. **Repetition is not evidence.** Count sources, not stories — and a source only counts once, no matter how many mastheads print it.