It has been argued in the literature on (language) testing that any act of testing/assessment can impact: (1) educators' curriculum design; (2) teachers' teaching practices; and (3) students' learning behaviors. This quality of any given testing situation or act of assessment has been called washback, or backwash if you will. Washback falls into the two categories of positive or negative--that is, beneficial or harmful. After an overview of the existing scholarly knowledge on washback, this paper argues that washback is not necessarily a test quality. Drawing on the notion of test method facets, the paper lends support to claims that see washback as a main function of teaching, learning, and policy-making situations or conditions rather than a quality of any given test. The paper also argues that the concepts of facet design and analysis including formal research designs, structural hypothesis testing, and measurement are inevitable and inescapable in any comprehensive model of washback. A possible borderline between backwash and washback is also proposed.