Metadata friction - tensions and costs in time, money, energy and attention caused by attempts to produce metadata ("Science friction: Data, metadata, and collaboration", Social Studies of Science, 2011, vol. 41, no. 5, pp. 667-690 ). Rather than viewing metadata as products (i.e., a set of descriptors or tags), the authors suggest to look at it as a process. The process of metadata production is often manual and of ad hoc quality. It is fragmented, i.e., many individuals contribute, they may do different things, such as creating websites or spreadsheets, answering questionnaires, etc. The process is divergent, i.e., several versions of metadata can be created. It is iterative because of the need to reconcile versions, repair mistakes and overcome miscommunications. It is also locally oriented, i.e., the local use of data is often privileged over the desire to contribute to the general project.
The authors offer the analogy with engineering, where friction is reduced by precision – making interacting parts mesh better – and by using lubricants. Typically, metadata discussions address precision - how to join parts (datasets) more perfectly. Metadata as process focuses more on lubrication: "the practices through which people overcome friction without precise solutions or the need to modify components." (p. 684). These practices are imprecise, therefore misunderstandings are inevitable. But they are an important part of metadata creation and exchange, therefore more attention should be paid to these practices as forms of scholarly communication.