I argue that if we make explicit the role of the user of scientific representations not only in the application but also in the construction of a model or representation, then inconsistent modeling assumptions do not pose an insurmountable obstacle to our representational practices.