Where the cone comes from
The Cone of Uncertainty traces back to Barry Boehm's research on software cost estimation, popularized further by Steve McConnell in Software Estimation: Demystifying the Black Art. It describes a consistent pattern: estimates made at the idea stage can be off by 4x in either direction, while estimates made right before implementation are far tighter.
Why the range narrows without anyone getting better at guessing
The narrowing isn't about estimation skill improving - it's about more actually being known. A project idea genuinely could take a quarter of the guessed time or four times as long; by the time requirements and design are locked in, the realistic range has collapsed dramatically because most of the unknowns have been resolved.
The honest use of this tool
Presenting a single-number estimate at the idea stage isn't more useful than presenting a range - it's a discarded range, dressed up as false precision. The range this tool shows is the more honest communication, especially to a stakeholder who will hold that number against the project later.