Decision-Making in Complex Infrastructure Work: Data-Driven Intuition?

Decision-making processes for developing and maintaining urban infrastructures are characterized by high degrees of complexity and uncertainty.
Increasingly, digitized technologies and data analytics are adopted to support (or
replace) such decision-making processes. Yet, urban infrastructures are inherently complicated because their development and maintenance are fraught with divergent stakeholder values and multiple object positions, entailing multi-faceted considerations, risks, and power relationships. In such contexts, judgement, morality, intuition and engaging in here-and-now materiality are often needed in order to bring together heterogeneous worlds. The inherent tension between multiplicity and uncertainty in working with infrastructures and the evidence-based reasoning of automated expert systems requires scholarly attention to issues such as the extent to which intelligent technologies should be authorized to make decisions independently and what can be lost when control is relinquished to the algorithm. We present an ethnographic study on pipeline maintenance and repair work in an urban environment to understand (1) the specificity of infrastructure maintenance practices and (2) how practices of deciding are changing with the increased digitalization of pipeline monitoring. The study follows district heating maintenance workers in Denmark on their transition from intuition-based to data-driven planning and decision-making, as their work is increasingly supported by drone-operated thermographic cameras, active monitoring and data visualizations that render leaks from underground pipes ‘visible’. We reflect how digitally mediated decisions differ from previous reliance on vigilant citizens and workers’ firsthand experience. Empirically, the study investigates practices of deciding – a state of relational entanglement that necessarily creates encounters among multiple worlds. Conceptually, it begins to explore the consequences of algorithmic computing and multiple computational ontologies for how decision-making emerges in the context of complex (infrastructural) work. In such an account lies the opportunity to further specify what it means to be deciding or enacting a decision in an era of extensive digitalization