Discussion of the potential of open data for impact -- including in education -- has grown. Open data from Wikipedia, the world's largest open educational resource, is used in a variety of learning contexts. A sociotechnical concern in this data is that it arises in a platform-based social dynamic, governed by norms and conventions, and negotiated by sets of volunteers for the particular purpose of writing an encyclopedia. This paper uses one feature of data, citation, as a boundary object. Through the exegesis of a reflective design case regarding a citation learning game (Citelearn) we highlight the continuities and discontinuities in citation across sites (Wikipedia and scholarly writing), and the significance of translation of such data across sites. The paper exemplifies a method for close case analyses of learning data, its representation, and re-representation, drawing attention to the need for such analysis in adopting data-infrastructure across context.