I’ll be interested to find out whether there’s a distinct difference in data management conventions for LIS professionals as opposed to whoever is creating the data. The blog post about the scarf and the overview on the Penn State library website both seemed to be addressing people who create data and, more generally, assuming an expert understanding of the data’s significance. As a result, they seemed to place a significant fraction of the responsibility for facilitating access to the data on researchers. If researchers are expected to have the skills to create their own metadata and maintain their data, does that mean that the role of librarians in a scientific setting is more advisory than custodial? In the scarf analogy, the person who publishes knitting patterns may not have created the actual pieces or even be able to, and the person who makes the scarf may not know the instruction-manual industry conventions and best practices for representing physical actions in a language any other knitter can understand. At the same time, the author of the manual has to know enough about the process of knitting to be able to foresee what knitters need in a pattern. So how much do scientific librarians need to know about the research that produced a certain dataset? What exactly are they doing with the data that researchers aren’t expected to either do themselves or direct in minute detail?