Vague probability words let everyone silently disagree while appearing to agree
In 1951, a US intelligence estimate called a Soviet move a "serious possibility." Sherman Kent, bothered by the phrase, polled the board members who had all signed it. Their private probability estimates ran from 20% to 80%. They had unanimously agreed on the phrase while privately disagreeing four-to-one about the likelihood of the event.
"Possible," "likely," "significant risk" are empty containers. Every reader fills them with the number they already believed. When a team publishes a readout saying attrition is "a real concern" or a plan "probably works," they haven't communicated a shared judgment. They've licensed every stakeholder to hear what they wanted to hear.
Putting an actual number on a judgment makes you measurably wrong. That's exactly why people avoid it, and exactly why it matters. A stated probability can be scored against what actually happened. It surfaces disagreement at the point where it can still be argued, not after the decision is made. The calibration habit follows: express uncertainty as a number, track whether outcomes match, and adjust until stated confidence matches real accuracy.
In the Kent case, every board member signed the same phrase. Their private estimates spanned 20% to 80%. The document looked like agreement. It wasn't.
Source claim: Shared probability language only looks like agreement: "serious possibility" privately meant 20% to 80% to the same signatories, so replacing vague words with actual numbers is the minimum unit of honest forecasting.