Talk at the conference Building Models of Change: Bridging Sciences and Humanities

On March 13, 2025, I gave a talk on “Computational historical models and historiographical uncertainty” at the conference Building Models of Change: Bridging Sciences and Humanities at Bielefeld University. Here’s the abstract:

The biggest impact of computers on scientific research has probably not been for calculations, but as universal modeling machines: computational models and simulations have considerably changed the natural and engineering sciences.

What about the humanities, and history in particular? As Gordon Leff remarked, “Historians as a profession are not given to constructing or employing models in any formal or explicit sense.” In fact, historians do construct models of their research objects, but these are rarely made explicit, let alone formalized. There are good reasons for this: given the little reliable information we have about the past, there is a high degree of uncertainty, and historians need to fill a lot of gaps to be able to produce coherent and defensible narratives. Consequently, the use of computational models and simulations in historical research (in the widest) sense is relatively limited and generally exclude the most important aspect, the causal relations inherent in a historical narrative.

Here, we are confronted with another level of uncertainty, historiographical uncertainty, which pertains to the interpretations and choices made by historians, i.e., uncertainty that relates to the narratives rather than the facts of the past. Is there any chance that we may deal with this type of uncertainty in a way that would make computational models as useful for historical research as for research in the natural sciences?