Calibration goes bad when the room compares people instead of evidence.
The point is simple: make standards portable across managers and teams. A "strong performer" in one group should not mean something completely different somewhere else. A promotion should not depend on which manager tells the best story. An underperformance call should not depend on who has the least patience.
Politics enters through ambiguity. If standards are vague, the loudest narrative wins. If evidence is thin, reputation fills the gap. If managers arrive unprepared, calibration becomes compensation defense, promotion lobbying, or quiet punishment.
Better calibration starts with packets, not speeches. What standard is being applied? What evidence supports the rating or decision? What examples show the bar? What counter-evidence exists? What decision is actually needed?
Operator artifact: require a calibration packet for one people decision: role expectations, evidence summary, examples of work, manager recommendation, risks, and proposed next step.
Field test: in the next calibration conversation, ban reputation-only claims. If someone says "everyone knows," ask for the work evidence.
This is part 5 of 10 in People Systems That Actually Raise the Bar.
