Goodhart's Law highlights how metrics in software engineering can become counterproductive when they turn into targets. The law's weak version suggests that people may manipulate metrics, while the strong version warns that even honest pursuit of these metrics can be detrimental. Examples in engineering include using test coverage as a proxy for thorough testing and lines of code for productivity. The solution is elusive, but combining multiple metrics and using best engineering judgment may help mitigate the effects.
Sort: