The suggestion that new courses can be constructed from existing learning objects appears technically self-evident but remains unproven. Despite increasing evidence that learning objects can provide a suitable structure for constructing courses most evidence is based on creating learning objects, either through restructuring existing materials or as a set of newly created learning objects. There is little evidence for building a new course from existing learning objects where any significant number of these have been developed for a different course elsewhere. Authors who adopt a constructivist model of knowledge might view this lack of evidence as confirmation of the inherent difficulty of integrating resources created within different communities. Those following a scientific model of knowledge would view this as a temporary problem that will be resolved when a sufficient volume of materials exists in repositories. This paper argues that reconfiguration of learning objects to create new courses is significantly more complex than is currently recognized, even within a scientific framework of knowledge. By implication, far more research is needed to understand reconfiguration and re-integration than initial "creation."