[kenjackson]

It depends on how the test was given. For example, did they ask you to write quicksort == you submit a version and then they grade it. If so I can imagine a lot of little mistakes... from the integer overflow issue (if they did pick a language), to off by one errors, to not handling empty lists, or the case where the item isn't in the list, or other errors that if someone were to say, "it doesn't work with this input" you could easily fix it, but if not given a chance for correction many could easily get wrong.