Will Wright, Game Designer extra-ordinaire, is the biggest advocate of prototypes that I know. At a recent conference, he said (and I'm paraphrasing here) that he will make 10 prototypes for every one feature that actually makes it into the game. These prototypes are little more than stick figures in 2d running around the screen, but like others have already said in this thread, you get a feel for whether a feature will be fun or not immediately. He then went on to say that for every 10 feature prototypes that make it into a game, one will actually go in with the code as it is. In other words, and this is important, code reuse is NOT a priority for prototyping according to Will Wright. Get it done quick and dirty and buggy, but get to the gameplay ASAP. Optimize and make it stable later.