Composition assessment scholars have exhibited uneasiness with the language of norming grounded in distaste for the psychometric assumption that achievement of consensus in a communal assessment setting is desirable even at the cost of individual pedagogical values. Responding to the problems of a "reliability" defined by homogenous agreement, compositionists have moved to reframe "validity" as a reflective measure of local context, often relying on hermeneutic approaches that foreground expression of difference as criteria for success, not failure. Such approaches, while serving the goals of assessment as reflective practice, foster a tension between conflict and consensus that is arguably unproductive for assessments that may benefit, in various ways, from the achievement of agreement. The present study, drawing from group development models, reframes conflict and consensus in dynamic integration. Reader training experiences in two large-scale portfolio assessments at one institution provide evidence that assessment practitioners may benefit, theoretically and pragmatically, from an approach to conflict and consensus neither hierarchical nor mutually exclusive.