When to remove hacks?

A coworker of mine ran into a bummer of a problem and asked for advice. Over lunch we talked about what to do, I personally felt that option 2 was the safest solution. I didn't really like adding bad code to cover up bad code and I began to think about how you would know when the hack could be removed. Then I threw out the idea of writing a test to verify that the bad condition is present. When the test fails you will know to remove (or at least re-evaluate) the hack. He thought it was a neat idea, can't wait to see what he comes up with.