Computation in the sciences does not sufficiently account for aesthetics, which prevents the information from being corroborated. This argument is made through a comparison between virtual earth modelling communities and epistemic culture and an elaboration on modelling methods as exemplary of the negotiations between scholarly knowledge, aesthetics, and computer resources necessary for visualization. The salience of these negotiations to institutionalized epistemic practice is reinforced through three histories of visualization in the sciences: computation in ecology, empirical modelling and physically based rendering, and visualization in scientific computing. The article flags the significance of this argument for its role in shaping public policy.