IMO most tests are fundamentally flawed. The way most testing is done it would be easier and better to just write everything twice, have a method to compare results, and hope you got it right at least once.