Kent's board agreed on "serious possibility" while privately meaning 20 to 80 percent

In 1951 the intelligence community issued an estimate judging a Soviet attack on Yugoslavia a "serious possibility." A State Department reader asked Sherman Kent what that actually meant. Kent said the odds were about 65 to 35 in favor. Bothered by the question, he went back and polled the other board members who had written and approved the same sentence. Their private odds ranged from 20 percent to 80 percent. The board had reached unanimous agreement on a phrase while disagreeing, four-to-one, about the event it described.

This is the failure that made Kent push for numbers. A shared word is not a shared judgment. The phrase sits in the document like an empty shell, and each reader, analyst and policymaker alike, fills it with the odds he already believed. Consensus on language disguises the absence of consensus on the world. Heuer later ran the experiment cleanly: NATO officers reading identical sentences with the probability words swapped assigned wildly different percentages, agreeing only on "better than even."

Kent's colleagues resisted numbers anyway. They thought a percentage faked a precision intelligence does not have, and that estimation was an art a number would cheapen. Kent's reply was that the ambiguity they were protecting was not humility about uncertainty. It was a way to never be measurably wrong.

Who read "serious possibility"	Odds they had in mind
Kent, who wrote it	~65%
The board that approved it	20% to 80%

Source claim: Polling the board that unanimously approved the phrase "serious possibility," Kent found they privately held odds ranging from 20 to 80 percent, proving that agreement on estimative words conceals disagreement about the actual probability.