Writing portfolio assessment and communal (shared, dialogical) assessment are two of our field's most creative, courageous, and influential innovations. Because they are also relatively expensive innovations, however, they remain vulnerable to cost-cutting by university administrators and to attacks from testing corporations. This article lays a theoretical foundation for those two powerful and valuable practices in teaching and assessing writing. Building on the concept of "complementarity" as developed in the fields of quantum physics (Bohr; Kafatos & Nadeau) and rhetoric (Bizzell) and adapted for educational evaluation (Guba & Lincoln), we provide some of the "epistemological basis," called for by Huot, on which portfolio and communal assessment are based and by which those practices can be justified. If we must look to science to validate our assessment practices (and perhaps we must), we should not settle for outdated theories of psychometrics that support techniques like multiple-choice testing. Instead, from more recent scientific theorizing we can garner strong support for many of our best practices, including communal and portfolio assessment. By looking to the new science--including the new psychometrics (Cronbach, Moss)--we can strengthen and protect assessment practices that are vibrantly and unapologetically rhetorical.