Abstract This essay closely examines the ongoing development and use of an open-source software tool commonly used in microbiome research in order to make three interlocking contributions. First, I identify data cleaning as a set of richly epistemic practices which are functionally inextricable from data analysis. This means that ‘good’ data cleaning decisions are not necessarily universal, but must be made suitably for the specific analytic purposes intended by later users of shared data. Second, I examine how the repair and modularity of data processing software can offer data users a variety of cleaning choices in order to enable divergent analytic goals. In doing so, this software facilitates the development of epistemically diverse data communities. Finally, turning to a high-profile paper retraction which hinged on a data processing error, I explore how repair at different points in the research production process can serve to enable or constrain the growth of data communities. Through this analysis, I argue that error serves as a site of productive negotiation over the interpretive flexibility of shared data, and that repair plays a critical role in stabilizing dat
📖 افتح في inklap 🔗 DOI 📮 اطلب بحثاً