DSci Happy Hour with Jo Guldi

The Dates that Dominate Wikipedia's Telling of Global History - Making global histories is difficult. Research in global history has grappled for at least twenty years with the limitations that available methods pose to explaining global movements, connections, and phenomena (Clarence-Smith, Pomerance and Vries, 2006; Gänger and Jürgen Osterhammel, 2024). While these limitations can be pragmatic—from archival access and restrictions to individual scholars’ limited linguistic backgrounds—current theories in global history are most useful for national, regional, or small-scale comparative scopes rather than the large-scale global resonance of historical phenomena. Here, we introduce a comparably large-scale and computational approach by looking for convergences and divergences in the dates mentioned across hundreds of thousands of Wikipedia articles written in hundreds of languages. We find that four dates dominate a significant proportion of the dataset, indicating that articles across these languages exhibit similar chronological histories. We identify a version of global history that is consistent across roughly 150 out of 365 languages. That said, the vast proportion of languages on Wikipedia don’t agree with the dominant dates, prompting suggesting further work to identify and interpret alternative, dissenting views on which events matter in the global past, as well as raising questions about the geopolitics of narratives on the internet. Further, we show that it is possible to manufacture a ranking of dissent and consensus to contextualize AI-generated text.