Test coverage is almost the perfect illustration of Goodhart’s law. Good programming practices do result in high test coverage, coverage is very easy to measure, but very easy to fake with useless “tests”. So, when the coverage is measured, the coverage goes up, but stops being meaningful.