Building a Digital Platform for Traceable Historical Research

The proliferation of digital tools, methods, and sources available to historians over the last few decades has jeopardized our shared understanding of the provenance and traceability of historical research. The rise of generative AI poses still further risks, not just in the fear of faked sources, but also in the naïve misuse of LLM-based co-pilots in the hands of researchers seeking assistance with datafication and analysis. Today’s historians are more likely than not to engage with their materials in digital and even quantitative ways, practices which inevitably result in calls for thinking more critically and carefully about traceability and transparency. To respond to this challenge, our interdisciplinary team has created kiara, free and open-source data orchestration software aimed at making the methodological processes of historical research more transparent and open to commentary, traceability, and criticism. Built with Python and interoperable with Jupyter, the popular interactive computing platform, kiara allows historians to create and annotate data, analyze it, and visualize their results in an environment which allows them and their audiences to engage critically with digital sources and their transformations from end to end. By examining our project’s dead-ends and setbacks as well as its insights and progress, this paper will argue for the ways in which research software development must patiently consider the benefits and drawbacks of automation rather than blindly responding to the fast-changing pace of the research landscape. It will also reflect on the influence on kiara of a wide range of other research software developed by our own lab’s teams, including Zotero and Tropy.