Possibility theory: A foundation for theoretical and empirical explorations of uncertainty

摘要

The field of operations and supply chain management (OSCM) has a long history of identifying and engaging with risk and uncertainty in operational practices.1 We provide a brief review of uncertainty in the OSCM domain, alongside an overview of our special issue (SI) call and accepted manuscripts. This serves as a starting point for the introduction of a new theoretical framework that reframes uncertainty as unresolved states of possibility. In this framework, the term possibility can refer to a broad array of OSCM actions and solutions including the novel application of existing approaches or technology as well as completely novel practices that enhance organizational outcomes. We illustrate the path-dependent evolution in these possibilities, alongside the limitations and opportunities imposed on the set of available possibilities resulting from concurrent evolution in the broader socio-technical system. We present the benefits of deploying a broader array of methodologies in the empirical study of what is, and is not, possible at discrete points in time, as well as the dual process of constraint and expansion in possibilities over time. The resulting empirical efforts to understand possibilities in turn enable novel theory development, elaboration of existing OSCM theory, and opportunities for bridging to other disciplines. When uncertainty is foreseeable or predictable it can be modeled, under various states of nature, as a set of choices and their outcomes. This lends itself to the application of statistical approaches and analytical modeling tools (e.g. maximax, minimax regret, Hurwitz criterion, and expected opportunity loss). Considerable research effort has explored the events and conditions that may impact these probabilistic models and underlying probabilities, as well as developing extensions to optimize the response to uncertainty in operational outcomes (e.g. utilization, yield, throughput) and efficacious ways to respond to external influences like customer demand and preferences (e.g., Boute et al., 2022; Charpin et al., 2021; Gao & Chen, 2015; Leung & Sun, 2021). A parallel stream of research explores the extent to which uncertainty is knowable. One example of such uncertainty arises from managerial ignorance—events that are knowable in principle, but not taken into consideration—with the resulting epistemic uncertainty labeled knowable unknowns (Packard & Clark, 2020; Ramasesh & Browning, 2014). In the face of this form of uncertainty, firms benefit from revisiting their decisions around complexity and complicatedness, and the heuristics they draw upon when responding to, or mitigating, operational consequences of unanticipated events (e.g. Feduzi et al., 2022; Ramasesh & Browning, 2014). Another type of uncertainty is unforeseeable, or aleatory, uncertainty (so-called unknown unknowns). Here, the probabilistic approaches and assumptions that predominate OSCM research fall short as this form of uncertainty is the result of “…some inherent, a priori causal indeterminism that cannot be mitigated…” (Packard & Clark, 2020, p.770). In practice, tools like insurance and forward contracts can provide straightforward and fruitful pathways for managerial action in the presence of unknown unknowns (Holmes & Westgren, 2020). At the same time, mitigation of known or knowable forms of uncertainty may not be viable due to the cost/benefit ratio or bounded rationality. Nevertheless, distinguishing epistemic uncertainty from aleatory uncertainty guides the use of predictive and non-predictive strategies. This distinction may also be helpful in rationalizing perceptions of uncertainty that can emerge at the intersection of high complexity processes and the cognitive limitations of managers and workers; where one can effectively consider the unfortunate emergence of unknown (or incomprehensible) knowns' (Bendoly et al., 2006; Sterman et al., 2015). In developing the proposal for the SI, we envisioned that the operational disru