The Second Call

Before you believe a big claim, ask each teller one question: how do you know? If every answer points back to the same person, you have one source — no matter how many people repeat it. A real check comes from a different direction: a second person who saw it themselves, or a paper that proves it.

Verification means confirming a claim before you publish it. The unit of confirmation is the independent source — an account that does not depend on any account you already hold. Independence lives at the origin: trace each account back by asking how do you know, then count distinct origins. Two witnesses who each saw the crash from separate corners: two sources. Two officials briefed by the same aide: one.

The old newsroom standard — the two-source rule — asks for two independent confirmations before a contested claim runs. It has an honest exception. A self-authenticating document — the signed resignation letter, the court filing, the certified budget — is not testimony about the event; often it is the event. When the record itself sits in your hands, one source can be enough.

Triangulation strengthens when the sources differ in kind, not just in name: a witness, a record, a data trail. Accounts of the same kind can share the same failure — everyone at one briefing heard the same wrong figure — while a document fails differently from a memory. And when every outlet carries a claim, check whether each did its own reporting: a hundred stories citing one wire dispatch are still a single origin.

Independence has a second axis: motive. Two genuinely separate sources who share an interest in your believing the claim can still mislead in parallel. The second call is not just to a second person — it is, ideally, to someone with nothing to gain.

Two friends tell you the pool is closed. But both heard it from the same lifeguard. That is one story, told twice. You still know it from only one place.

To be surer, check a different way — read the sign on the gate, or ask someone who walked there and saw. Two separate ways of knowing beat two repeats.

Kovach and Rosenstiel call verification the discipline that separates journalism from everything that merely resembles it. The operative test is conditional dependence: does the probability that source B confirms the claim change once you know what source A said — because B heard A, was briefed alongside A, or coordinates with A? If yes, B's confirmation adds little. The two-source rule is a heuristic approximation of this test, not a substitute for it.

Documents invert the calculus. A certified filing is evidence whose authenticity can be established from its own custody — seal, docket number, provenance — without leaning on any witness's reliability. That is why one authenticated record can outweigh several concurring voices, and why forgery, the attack on custody itself, is the failure mode to check first.

In Bayesian terms, testimonies confirm a hypothesis in proportion to their conditional independence given the evidence; perfectly correlated witnesses collapse to a single likelihood ratio however many there are. The common-cause structure — chief of staff at the root, officials as leaves — is exactly the graph in which naive source-counting most overstates confirmation.

Institutional practice encodes the same logic: wire-service standards require reporters to establish how each confirming source knows, not merely that they agree, and anonymous-sourcing policies demand an editor be told the origin precisely so dependence can be audited. The canonical failures — the aluminum-tube WMD reporting, the Killian memos — were failures of independence and of custody respectively, not of source count.