The Next Generation Science Standards (NGSS) [Achieve, Inc. [2013]] represent a broad consensus that teaching and learning expectations must change. Rather than memorizing and reciting information, students are now expected to engage in science practices to develop a deep understanding of core science ideas. While we want to share in the optimism about NGSS, the standards are not a silver bullet for transforming science classrooms. They are, instead, another reform document designed to suggest opportunities for students to actively engage in knowledge construction themselves--to be "doers of science," rather than "receivers of facts." A foundational contradiction underlies these efforts--while we want students to do science, we seem to mean that students should mimic practices others have selected as important to learn, and content others have selected as foundational. As a result, students are rarely positioned with epistemic agency: the power to shape the knowledge production and practices of a community [Stroupe [2014] Science Education 98:487-516]. We argue that unless the field tackles significant questions around precisely how students can be active agents in knowledge construction, we will likely continue to implement learning environments that position students as receivers of scientific facts and practices, even as classrooms adopt NGSS. In this conceptual analysis article, we unpack the construct of "epistemic agency" and its relationship to the NGSS, using a vignette to illustrate how students are typically positioned in researcher-developed curricula. The vignette, which describes a seventh-grade class exploring which of two lakes is more at risk for invasion by the spiny water flea, provides an exemplar of what we take to be a loose consensus about learning environments consistent with the NGSS. However, when we look beneath the surface of the consensus, the vignette reveals contradictions and unresolved issues around epistemic agency.