The power of simulation is a controversial topic in Philosophy and Computing. Whether there are processes and objects we can't simulate, what conclusions we should draw from our attempts, and whether computer simulations have any epistemic value are difficult questions with as many answers as there are scholars. Much of the disagreement seems to come from disjoint concepts of simulation and different understandings of exactly what simulations should do. I propose a goal oriented understanding of simulatability which splits simulation practices into prediction, analysis, and reproduction. Additionally, I argue that understanding simulations and models more generally in terms of these distinct goals allows us to cut through confusion around fundamentally different simulations and better evaluate arguments like Searle's Chinese Room.