Data Integration without Unification

Life scientists generate big data by pooling many smaller datasets and by ensuring that those datasets combine to form a trustworthy body of information with a net increase in use value. Most proceed by constructing a maximally comprehensive dataset based on universal standards for representing the data’s empirical content and fit for different uses. We argue that this approach rests on an regulative ideal to create unified datasets, but following this ideal isn’t necessary: there are alternatives that enable the benefits of data pooling to be realized through infrastructure supporting lateral exchange and customization of data among multiple sources. We illustrate data integration without unification in the context of big data for biodiversity, which aims to address rapid biodiversity losses across the globe.