During research on the management of alcoholic liver disease the authors found that it was difficult to keep the condition in focus through the course of the study. Perhaps this was a sign of methodological failure, but this paper explores an alternative possibility: that social science methods are ill adapted for the study of complex and messy objects. The paper reviews arguments about the character of complex objects as these have been recently elaborated within science, technology and society (STS), and applies these to alcoholic liver disease. Three versions of the object (as region, network and fluid) are found to be relevant. But so, too, is a fourth, fire version, which treats objects as patterns of discontinuity between absence and presence. It is argued that the messiness of alcoholic liver disease in part becomes comprehensible if we imagine it as a fire object.