Who says so — and how do they know?

This spine deals with only one kind of source: the human being. How to check documents is covered in What does a secret document prove?; what a whole case rests on, in What counts as evidence?. Here it is about four things: proximity, timing, verifiability and interest.

AI illustration

This is not a question of honesty. Even a completely sincere chain loses details with every link — and gains new ones that nobody invented but that emerge in the retelling. That is why the first question to any statement is not “is it true?” but “how many hands lie in between?”.

Anyone retelling smooths things out unconsciously: the unclear is made comprehensible, hesitation disappears, a “maybe three metres” becomes “three metres”. The result is a well-rounded story — and its very roundness is a warning sign, because real observations are rarely smooth.

Rank, clearance and years of service answer a different question: whether someone had access to a field. They do not answer whether they were present at this particular event. This is exactly where debates slip most often — the authority of the person is turned into the authority of the statement.

Anonymity can have good reasons: secrecy obligations, professional consequences, ongoing proceedings. So it is no proof of dishonesty. But it comes at a price, and the statement pays it: without a checkable position it remains a claim with nothing to anchor it. How far this can go is shown by the case of the anonymously surfaced “MJ-12” papers.

The fallacy runs in both directions: “he profits from it, so he is lying” is just as sloppy as “he is risking something, so it is true”. Interest is not a truth value but a reason to look more closely — for critics, incidentally, just as much as for witnesses.

Repetition is not confirmation. Confirmation would be a second, independent source that observed the same thing independently. The test for this is uncomfortable but simple: follow every report back to its origin. If they all end at the same person, you have read one source — twenty times.

Three moves usually suffice: look at the dates of the reports, search the text for the word “according to”, and check what it points at. If every trail leads to the same origin, the number of hits is no longer an argument — it is just an echo.

Arnold compared the flight motion, not the shape; headline writers coined “flying saucer” from it (Wikipedia: Flying saucer). The error was not without consequence: the saucer shape became the image readers carried in their heads — and appeared in reports afterwards. A single imprecise retelling shaped the expectations of an entire field.

The advantage of these four questions is that they work without insinuation. They do not ask whether someone is lying — but what their statement can carry. This lets you take a source seriously and still place it; that is exactly what separates checking from believing or dismissing.

Intelligence officer Jesse Marcel spoke in 1947 of foil and broken wooden beams; in 1978 he called the weather balloon explanation a cover story (Wikipedia: Roswell incident). This need not be dishonesty: memory is rewritten over decades — by what one has since read, seen and told oneself. That is why the timing of a statement counts almost as much as its content.

Asking about origin insinuates nothing. It treats the statement as what it is: the beginning of an inquiry. And you quickly notice that the more interesting question is rarely “is he lying?” but “what exactly can this person actually know?”.