This session will describe how the California Institute of Technology (Caltech) Library is moving away from stock to flows. Stock is knowledge embedded in media that have a low delta of change, whereas flows, such as cutting-edge research, have a high delta of change. Thanks to our profession's roots in print collections, librarians associate static and stable with sustainability, and that association with solid objects was translated into early digital library architectures. In moving from stock to flows we are proposing that it is, in fact, dynamic and mutable architectures that are the most viable for resilient and sustainable services.
Data decoupled from software flows from curated repository applications, is augmented by external data sources, and is stored in a simple, auto-updating Dataset (a Go package for managing JSON documents), against which simple scripts and APIs make the data available to users, serve it up to library applications (websites, harvesters/indexers, analytical tools), and, through batch update scripts, back into repositories. Dataset is highly mutable: regularly refreshed and re-indexed in cycles appropriate to the data source, and easily restructured or reconstructed. This strategy enables the library to minimize costly technology changes and to optimize a small development team by abstracting skillsets from platforms.
Committing to move from stock to flows means disinvesting from pseudo-stock (rented access to journals) while redirecting resources toward Caltech-contextualized stock (institutional bibliography) and open infrastructure. The flows model promotes both library ownership of data (public purpose) with a simultaneous architecture for sharing (public domain), so more closely aligning dynamic data and research outputs generated within the institution to the trusted role of the library in society.
caltechlibrary.github.io/dataset/
