Partial measurement is worse than no measurement.
Austin’s formal model proves that when you can only observe some dimensions of an employee’s output (which is always the case in software), measurement-based management systematically destroys value in the unmeasured dimensions. Effort shifts entirely toward what’s measured and away from what isn’t. The rational response to being measured on lines of code is to write more lines of code, not better code. Applied to directive governance: the metrics that reach the C-suite don’t just miss the important stuff. They actively redirect effort away from it.