This issue definitely isn't unique to Codecademy. I spent way too much time figuring out the exact combination of whitespace that would satisfy CodeEval. I suppose finding the right set of fine-grain test cases remains a problem. Perhaps they should start using QuickCheck or some other random testing + test case distribution combinator system (if they aren't already). Those can systematically build up from small base cases to odd corner cases.