In this paper, I develop the distinction between the problems a computation can be used for and the problem a computation is used for, if it is used at all. I argue that the identity of an algorithm is determined solely by its rule structure, independent of any intended or realized use. Whether an algorithm can be used successfully for a particular problem is an extrinsic fact relating the algorithm and the problem, not an intrinsic feature of the problem. I go on to develop what this usability relation looks like: an algorithm is usable for a problem when there exists a correct and honest coding that translates between the problem and the algorithm's computational states. With the use/usability distinction in hand, I show how it exposes issues in arguments for extrinsic accounts of physical computation and how it clarifies computational explanations on an intrinsic account. In particular, I object to Shagrir's master argument—which exploits the phenomenon of simultaneous implementation to advance extrinsic accounts—by arguing that the extrinsic context Shagrir appeals to does not individuate a computation at all; rather, it individuates the coding used to translate between the system and its context. Finally, I provide a straightforward account of computational explanations made available to intrinsic accounts by usability: a system’s performing a task is explained by the fact that it performs a computation that is usable for that task. Importantly, this account captures computationally relevant differences in how two systems perform the same task.