The decision test for intelligence

A briefing can be accurate, carefully sourced and still leave its reader unable to decide. My proposed test for intelligence is whether it makes a consequential uncertainty easier to act on. It may change a decision, strengthen the case for waiting, or show why the current course remains justified.

The starting point

In his essay “When everything is intelligence — nothing is intelligence,” Wilhelm Agrell distinguishes intelligence analysis from information processing and places the handling of uncertainty at its center. This is an argument about analytical practice, not an empirical demonstration that one briefing format outperforms another.

My practical extension is a decision test. Before adding another source to a product, write down the choice that the reader faces and the unresolved question that could change it.

Three things the product should make visible

First, specify the decision. “Understand regional instability” is a research interest. “Decide whether to change a planned visit before Friday” creates a question with a deadline and consequences. This is a hypothetical example, not advice about a particular location.

Second, identify the uncertainty that separates the options. A list of disruptions is useful context. The analytical problem is whether those disruptions are likely to affect the planned activity, which mechanisms could connect them, and what evidence would distinguish a temporary interruption from a persistent constraint.

Third, state the revision condition. “Monitor developments” leaves the next analyst with an open-ended collection task. A more useful instruction names the observation that would strengthen or weaken the judgment and explains why that observation matters.

Separate the judgment from the action

An analyst should not disguise a risk preference as an intelligence finding. Two readers can accept the same assessment and choose different actions because they face different costs, responsibilities or tolerances for disruption.

I would therefore put three short fields at the top of a briefing: what the evidence supports; what remains unresolved; and which choice the uncertainty affects. The decision owner can then see where analysis ends and their own trade-off begins.

A useful fourth field is the strongest alternative explanation. It should be a plausible account of the same observations, with a distinguishing indicator. An alternative included only to be dismissed adds ceremony rather than scrutiny.

How I would evaluate the method

The test should not reward dramatic recommendations. A well-supported judgment that no change is warranted can be valuable. Nor should an eventual good outcome automatically validate the analysis: luck and unrelated developments can intervene.

I would retain the original judgment, its evidence, the alternatives and its revision conditions. After the decision window closes, review whether the reasoning used information available at the time and whether new observations triggered appropriate updates. That makes the proposed process inspectable without pretending it has already been validated.

The goal is a briefing whose reader can identify the analytical contribution without counting its pages: this is the uncertainty, this is why it matters, and this is what would change the assessment.

Source and scope

Wilhelm Agrell, “When everything is intelligence — nothing is intelligence,” October 2002.

Source selected from OV INT’s research collection. The decision test and hypothetical example above are my methodological proposals, not findings attributed to Agrell. Hosting by the CIA does not imply agency endorsement of this article or the source author’s views.