Estimating software when nobody can

Software estimates are guesses wearing a suit. The honest move is not to pretend otherwise, but to make the uncertainty explicit and then shrink it deliberately.

We estimate in ranges, not single numbers, and we tighten those ranges by building the riskiest, least-understood part first rather than saving it for the end.

A plan that admits what it does not yet know is far more useful than a confident one that turns out to be wrong.