The ongoing attempts to measure educational quality, to hold schools responsible for their performance against these measures, and to use the resulting data to design improvement initiatives have yet to yield the promised results. A universal definition of "good education" seems to elude our grasp. Despite broad public support for "accountability" in education, actual efforts to hold educators accountable for their quality over the past 15 years have tended to inspire rebellion from the very publics that these efforts are meant to serve. In what follows, the author attributes these persistent failures of accountability policy to two closely related but distinct assumptions about educational quality, the objects that policy seek to measure, to evaluate, and to act upon. The first of these pertains to the ontology of educational quality; the second pertains to the politics involved in measuring, judging, and improving educational processes. Taken together, these assumptions would leave the tasks of defining and measuring educational quality, and of prescribing interventions in educational practice, to be undertaken by experts in social and data science, as mandated by federal and state statute.