State Uncertainty Precisely

In March 1951 an American National Intelligence Estimate concluded that an attack on Yugoslavia that year should be considered “a serious possibility.” Shortly after, the chairman of the State Department’s Policy Planning Staff asked Sherman Kent, of the Board of National Estimates that had worked on it, what odds the phrase had meant. Kent said his own estimate was about sixty-five to thirty-five in favor of an attack. The chairman was somewhat jolted, because he and his colleagues had read the words to mean odds considerably lower. Kent then asked the members of his own board what they had had in mind when they agreed on the wording. Their answers ran from about twenty percent to about eighty.

Years later Kent wrote the episode up in an essay the CIA has since released, arguing for consistent, unambiguous use of a few key expressions. Left unmodified, he wrote, “reportedly” carries “no evaluative weight whatsoever,” and analysts handle reports whose credibility runs from almost certain truth to almost certain nonsense. The writer who reaches for such a word intends honesty, and the reader gets fog, because it does not say what is unknown, why, or what would settle it.

Precision often means fewer hedges, each with content, beginning with which kind of uncertainty you have. It may be that no one knows, because the decision was never made or no record exists. If the sources conflict, you name them; if you are offering an estimate, you give its basis. “Reportedly” is followed by who reported it and how well placed they were. Readers often act on your uncertainty, and if it is blurred they act on a guess about what you meant, which the Yugoslavia episode shows can scatter widely. Define a word once, early, and keep to it.

Kent himself later judged his first table, eleven phrases each with attached odds, to be unjustifiably sharp given the inexactness of the data, and cut it to five gradations. A figure with no basis misleads in its own way, so if you have none, do not supply one.

Practice. Highlight every hedge in one chapter: may, might, perhaps, reportedly, apparently, seems, some say, it is thought, likely, possibly, approximately, disputed. For each, decide what kind of uncertainty it expresses, and rewrite it as a sentence that says what is unknown, why (no record, sources conflict, not checked, unknowable), and what would settle it. Where the honest answer is that the hedge was a habit, delete it. After an hour you have a before-and-after list of the hedges you changed, with the number deleted written at the top.