How many sources do you really have?

How Many Sources Do You Really Have?

Ten reports can represent one source when each repeats the same interview, press release or estimate. Counting documents rewards visibility, not corroboration, and creates false confidence precisely when a claim has travelled far enough for its origin to disappear. Evidence strength depends on independent collection paths, access and incentives�not the number of links in a bibliography.

Construct a source genealogy for every consequential claim. Trace citations back to the earliest observable evidence, mark intermediaries that merely interpret it and group all descendants into one evidential chain. Distinguish primary access, independent confirmation, methodological replication and commentary. Two analysts using the same commercial dataset are not independent observations.

Independence is a spectrum. Sources may share owners, respondents, databases, models or incentives without declaring the connection. Examine who collected the information, when, for what purpose and with what ability to know. The US intelligence community�s analytic standards require the quality and credibility of sources and methods to be described because a precise claim can still rest on weak access.

Triangulation works best across different failure modes. Combine a management statement with customer behaviour, regulatory records, physical capacity, hiring or transaction data. Agreement between sources that could fail for different reasons raises confidence; disagreement is diagnostic. It may expose timing differences, definitions or a deliberate narrative rather than demand an immediate average.

Report the result as evidential depth: number of independent chains, quality of access, recency, known bias and unresolved conflict. Preserve the original source beside every derivative claim so the chain remains auditable. The objective is not to maximise citations. It is to know how many genuinely separate reasons exist for believing the conclusion.