What stories might historians tease from these detailed but happenstance sources? Before the advent of web-based search, the normal course was to turn select pieces of semi-structured information into structured data and aggregate it for quantitative analysis. The numbers produced would be guided by questions shaped by qualitative readings of other sources and by place-based expertise accumulated along the road to stumbling across these pages in this archive in the first place. In an era of massive digitization, an array of other possibilities opens. We can apply microhistorical methods across a corpus of target sources not limited by place—and indeed, whose outer edges may not be known to us at all. What are the potential rewards; and what are the concerns?
Meanwhile, the availability of off-the-shelf “Artificial Intelligence” tools raises a new set of possibilities, and of cautions. Aggregating information across multiple sources once required human-guided curation, but AI models seem to hold out the promise of algorithmic assistance—or even, fully outsourcing of the tasks of summarizing and finding. As a discipline, we need to explore and explain how archival fragments can generate knowledge in a digital age: including the hard-won caveats and concerns we may have to share.