This could covers issues such as resource allocation where the programmer writes code that does not properly release memory, and the QA did not have a test suitable to trigger the memory problem (possibly because they didn't know such a thing coudl happen - and it is hard to test for things you do not know about)...