In this commentary, I discuss why, despite the existence of gradience in phonetics and phonology, there is still a need for abstract representations. Most proponents of exemplar models assume multiple levels of abstraction, allowing for an integration of the gradient and the categorical. Ben Ambridge's dismissal of generative models such as Optimality Theory (OT) is problematic because OT not only allows for the abstract, but can also handle a variety of phenomena, including gradient representations, and similarity among output forms. [For Ben Ambridge's "Against Stored Abstractions: A Radical Exemplar Model of Language Acquisition," see EJ1269951.]