Situated modeling: applying a modeler's manifesto to collaborative data science in rural Zimbabwe

Colloquium | April 25 | 3:30-5 p.m. | 575 McCone Hall

 M.V. Eitzel Solera, University of California Santa Cruz

 Department of Geography

As a practicing modeler by training and experience, I have encountered many issues concerning the way data are handled, how models are made, and, in particular, how truth claims emerge from modeling. "Big data" are increasingly discussed in academic venues and elsewhere as an unquestioned good. However, there is dangerous potential for models to marginalize people based on biased or inappropriate data, affording them no recourse to those same tools to defend themselves. Therefore, modeling needs to be practiced more critically and less automatically. In this talk, I summarize a "modeler's manifesto" applying Haraway's concept of situated knowledge to modeling as a way of knowing. Through the manifesto, I suggest practices grounded in feminist objectivity, requiring more descriptive methods and ways to model responsibly and collaboratively.

I will then apply this manifesto to my collaborative modeling efforts with the Muonde Trust in Mazvihwa Communal Area, Zimbabwe. The community's concerns revolve around land-use decisions in their agro-pastoral system, choosing how to allocate land to arable production versus woodland grazing area while also maintaining enough livestock. We created an agent-based model in NetLogo to represent the feedbacks between these components using an iterative approach with the Muonde Trust research team, grounded in their 35 years of community-based data. We held workshops with community members and local leaders demonstrating and discussing the model at several points, which both led to alterations of the model and of land-use policy in the area. The current model includes a range of management tactics employed by farmers in the community as well as several ways to simulate higher rainfall variability due to climate change. Local leaders are now in conversation with Muonde's researchers about ways to re-cultivate fallow fields rather than converting woodland to new arable production. Our modeling and community engagement process exemplifies many of the suggested practices from the modeler's manifesto, and I will evaluate the success of the manifesto as a framing for this collaborative project.

 jsmandel@berkeley.edu, 510-642-3903