In an algorithm class, professor introduced us to some simple search algorithms (bubble sort, selection sort, insertion sort, shell sort). He did a quite decent job and most of the students were able to grasp the code and understand the differences in those algorithms. But then he spoiled his whole lecture with one additional slide. There he proposed an optimization: Instead of using a temporary swap variable, we just could use the first array element (or the zeroth element, respectively: the one ad index 0) for doing all the swapping. We just had to document that, so that the caller would "leave the first position of the array empty", resulting in "cleaner code". And he did that in the same class where he used Big-O notation to argue about runtime complexity. But having the caller to resize the array and to shift all the elements by one position did not matter to him at all, because it was "not part of the actual algorithm".