How to tell if a news outlet is citing a source or just invoking authority

"The FBI's annual report, pages 39-40, confirms that…" Sentences like that have been showing up for weeks in newsrooms that take precision seriously, including — until a few days ago — this one. They sound verified: a weighty institution, a specific document, even a page number. And yet they say nothing checkable, because there is no way to confirm them without redoing the search the journalist already did. That sentence is not a source. It is a name with borrowed authority, and the difference between the two is exactly what this piece teaches you to spot in two minutes, with any outlet — this one included.

A source and an authority are not the same thing

Citing a source means giving the reader a way to check the claim independently: a link to the document, the exact page, something they can open and read. Invoking authority means naming someone or something prestigious — an agency, a study, "experts" — and asking the reader to trust that the journalist represented it accurately, without giving them a way to check. The two forms sound almost identical on the page. Only one is verifiable.

The distinction has a name in classical logic: the argumentum ad verecundiam, or appeal to authority. The Stanford Encyclopedia of Philosophy describes it as the error of "accepting as evidence for a proposition the pronouncement of someone who is taken to be an authority but is either not really an authority or a relevant authority." The nuance matters: relying on a genuine authority is not, by itself, a reasoning error — much of what we know comes precisely from trusting someone who knows more. The problem appears when that authority can't be checked, or when the reader has no way of knowing whether the journalist represented it faithfully. A link solves exactly that problem: it turns the appeal into something verifiable.

This shows up with particular clarity in AI news, because the sector runs on self-issued announcements. "According to OpenAI, the new model improves 40% on reasoning tasks" is, as written, an appeal to authority: it names the company, gives a specific number, and lets you check nothing. "According to the announcement OpenAI published, the new model improves 40%," with that same sentence linked to the actual announcement, is a source: the reader can open the link, find the figure and check what benchmark it was measured on, against what comparison, under what conditions — which is, almost always, where the nuance the headline left out is hiding. The unlinked sentence and the linked one say exactly the same thing in English. Only one can be put to the test.

The uncomfortable case: our own number

Here is where this piece stops being comfortable. This same week, this newsroom's editor-in-chief measured its last 50 published pieces, counting how many carried at least one external link placed at the exact point where a claim is made — not in an end-of-article source list, but where the claim itself is stated. The result: 44 out of 50, zero.

Before writing that figure down, I checked it myself, opening published pieces at random, because repeating a number without checking it would be exactly the error this piece is calling out. What I found explains the number without softening it: almost all of those pieces do carry a "Sources for this piece" section at the end, with real links to press releases, official blogs or studies. But a list of sources at the bottom is not the same as a link at the point where the claim is made. A reader who reads "the company announced a 42% drop" mid-article does not scroll to the bottom to find which of four listed links backs that specific number — and even if they did, they'd have to guess which one supports it. A source that isn't where it's needed functions, in practice, as if it weren't there at all.

This isn't an isolated case of bad faith: it's a habit inherited from how writing has been done for years, inside this newsroom and outside it. And it is exactly the kind of error that only gets fixed by measuring it first.

The two-minute check

This is the part that works with any outlet, not just this one. Next time you read a weighty claim — a figure, a quote, an "according to a report" — ask yourself three questions, in this order:

1. Is there a link right there, or just a name? If the sentence names a source but doesn't link it in that same sentence or paragraph, you can't check it without doing the search the journalist should have done.
2. If there is a link, does it go to the original document, or to another outlet commenting on it? A link to another paper's article that itself cites the report isn't a primary source: it's an echo with a link, which beats nothing but isn't the same as reaching the document.
3. Is the link where the claim is made, or only in an end list nobody scrolls down to check? A correct source in the wrong place has, in practice, the same effect as no source: almost nobody follows it.

If the answer to the first question is no, you already have your diagnosis: that isn't a citation, it's an appeal to authority wearing the grammar of one. You can apply this right here, right now: find the sentence "44 out of 50, zero" above and check that the paragraph around it tells you exactly who measured it and how, even though there's no external public document to link to — because it's an internal figure from this newsroom, not a third party's, and that difference also needs to be stated plainly rather than papered over with a link that doesn't exist.

Why this doesn't expire

This check doesn't depend on which outlet you're reading or which topic it covers. An FBI report, a study cited in an AI story, a statistic from a corporate announcement: the question is always the same, and it's always answered in the same place — the paragraph where the claim was made, not a bibliography that makes the reader do the work of linking the number back to its origin.

The underlying reason isn't cosmetic. An outlet that asks readers to trust a name without giving them the door to check it is asking for faith, not offering rigor. And rigor is only demonstrated by letting anyone, including the most skeptical reader, retrace the exact path the journalist took and land in the same place.

There's a fair objection to all this: checking every link in every article you read is, in practice, exhausting, and no one does it as a habit. You don't need to. The value of the three questions isn't in applying them constantly, it's in knowing how to apply them the moment something feels off — a number that's suspiciously round, a claim that flatters whoever is making it, a headline shared far more than it's actually read. That's the moment it matters to know, on the spot, whether what backs that sentence is a document you can open or just a name that sounds good. That's the skill you leave with: not memorizing which outlets are trustworthy, but knowing how to check, in two minutes and with any text — starting with this one — whether what's in front of you is a source or just the name of one.