Source reliability assessment is the practice of systematically evaluating the trustworthiness of intelligence sources and the credibility of the information they provide. The most widely used framework is the admiralty system (also called the NATO system), which rates sources on a six-point alphabetic scale (A through F) and information on a six-point numeric scale (1 through 6). An “A-1” rating indicates a completely reliable source providing confirmed information; an “F-6” rating indicates a source whose reliability cannot be judged providing information whose truth cannot be determined. The two dimensions are evaluated independently: a highly reliable source can report information that turns out to be wrong, and an untested source can provide information that proves accurate.
The system forces analysts to separate two questions that intuition tends to fuse: “do I trust this source?” and “do I believe this report?” A HUMINT source with a long track record of accurate reporting (rated B — usually reliable) who reports something inconsistent with all other collection (information rated 4 — doubtfully true) presents a different analytic problem than an untested source (rated E) reporting something confirmed by multiple independent sources (information rated 1). The admiralty system makes these combinations explicit, allowing consumers to understand not just what the intelligence says but how much weight it should bear.
In practice, source reliability assessment is less precise than the rating system implies. A source’s reliability is itself an analytic judgment, based on access (does the source have the positioning to know what they claim to know?), track record (has previous reporting from this source proven accurate?), motivation (does the source have reasons to fabricate or embellish?), and corroboration (does other reporting support or contradict this source?). Each of these criteria involves uncertainty: access can be fabricated, track records can reflect adversary management of a double agent, motivations can be mixed or hidden, and corroboration can be circular if multiple sources draw on the same underlying information.
The Iraq WMD intelligence failure exposed the system’s vulnerability to institutional pressure. The source codenamed CURVEBALL — an Iraqi defector whose fabricated reporting on mobile biological weapons laboratories became central to the case for war — carried reliability warnings from the German intelligence service that handled him, but these warnings were either not transmitted to or not heeded by the analysts who relied on his reporting. The failure was not in the rating system itself but in the institutional willingness to apply it honestly when the resulting assessment would be unwelcome — a form of politicization operating at the source evaluation level.
Related terms
- HUMINT — the collection discipline where source reliability assessment is most critical
- All-source analysis — the analytic practice that integrates source-rated reporting
- Double agent — the counterintelligence condition that corrupts source reliability from within
- Estimative language — the vocabulary for conveying the confidence that source reliability informs