Presentation short abstract

Models make uncertainty computable, but social life holds unruly uncertainties that resist quantification. This wound in the algorithm reveals what models erase relations, refusals, and histories. We call for modeling that stays partial, accountable, and open to the unmodeled.

Presentation long abstract

Amid accelerating climate turbulence, mathematical models have become central instruments for decision-making. Across domains such as stochastic optimization, chaos theory, and machine learning, uncertainty is framed as a computable artifact: a sensitivity parameter, a probability distribution, a scenario waiting to be enumerated. We take uncertainty as an entry point not simply to improve modeling, but to interrogate the ontological assumptions through which models craft the worlds they claim to represent. There is, we argue, a wound in the algorithm: an opening that exposes the limits of approaches premised on a world that is fully knowable, quantifiable, and optimizable. Yet many uncertainties encountered in environmental and political life cannot, and should not, be forced into numerical containers. Unruliness names this dimension of uncertainty that resists translation: political refusal, spiritual causality, collective memory, land-as-kin, trauma, sovereignty, story. These are not epistemic gaps awaiting better data; they are forms of relation that do not belong to the quantitative. When models discipline uncertainty by rendering only computable forms legible, they silence alternative worldviews, erase relational knowledge, and reproduce political hierarchies under the guise of mathematical rigor. Drawing on case studies from climate adaptation, energy governance, and risk assessment, we show how modeling frameworks filter out the very uncertainties rooted in politics, history, and relational life. In response, we propose frictional modeling: an approach that keeps models wounded, partial, and accountable, preserves opacity, and holds space for relational and political sovereignties that live in the unmodeled.