Watch the seminar here

AI generated summary of the talk:

The seminar, hosted by Zef Segal and led by Sergi Lozano, focused on network models as analytical and conceptual tools in historical research. Rather than treating networks as a single method, Lozano examined different types of models and the assumptions they embed. He began with Leonhard Euler’s solution to the Königsberg bridges problem, using it to illustrate modeling as a process of abstraction. By reducing a complex physical setting to nodes and edges, Euler demonstrated how simplification enables new forms of reasoning. This set the stage for understanding network models as deliberate constructions that highlight certain relationships while omitting others. Lozano then outlined several major models, including random graphs, preferential attachment, and spatially constrained networks. Each model, he argued, reflects a different hypothesis about how connections form. Random models serve as baselines, preferential attachment captures cumulative advantage and hierarchy, and spatial models emphasize constraints such as geography. Choosing a model is therefore an interpretive act, tied to the researcher’s question. A central distinction framed the talk: models as interpretational tools and as evolutionary tools. In the first sense, models help explain observed structures by suggesting underlying mechanisms. In the second, they allow researchers to simulate how networks might develop over time, offering ways to explore processes even when historical data are incomplete. The Q&A addressed model selection, the role of simulation, and the limits of abstraction. Lozano emphasized that no single model is inherently correct; different models illuminate different aspects of the same phenomenon. Simulations were described as structured thought experiments rather than reconstructions of the past. Overall, the seminar positioned network models as integral to historical reasoning, shaping both explanation and interpretation.