The point was to change the code structure without changing the tests (possibly to enable a new feature or other change). The challenge being when the tests are at the wrong "level", probably by team policy IME. If you change the tests, how can you be sure the behavior really matches what it was before?
No comments yet.