I think this holds for well written tests, but if one is not talented in writing them then more tests will simply bolster the bugs in the main code. For example

I think this holds for well written tests, but if one is not talented in writing them then more tests will simply bolster the bugs in the main code. For example, if a developer generate test data by running the program and copying it's output (something I see far more often than I'm comfortable with), they can tell themselves that section was tested thoroughly and must be bug free. Thus when someones notices an end result someplace is wrong, the first place checked may not be that "well tested" function. Tests are only as good as the domain knowledge of the person writing them.