Causal modelling of knowing how to prepare: A concept for a ‘strategic Bayesian operations room’ exercise
Abstract as pdf
Corresponding Author:
Tapio Reinekoski
Tampere University Finland
tapio.reinekoski@tuni.fi
Author(s):
Marko Ahvenainen1 , Nina Janasik2 and Annukka Lehikoinen3
1University Of Turku, Marko.Ahvenainen@Utu.Fi, FIN
2University Of Helsinki, Nina.Janasik@Helsinki.Fi, FIN
3University Of Helsinki And Kotka Maritime Research Centre, Annukka.Lehikoinen@Helsinki.Fi, FIN
Constructing situational awareness from uncertain knowledge in crises is crucial for taking decisive action. Gaining such insight is a very ‘Bayesian’¹ practice in the sense that observations of new information update prior beliefs and result in a new posterior understanding. Priors and posteriors are not only distributed across people, organisations, and devices, but they also extend in time, from acute operative management to long-term strategic planning. To systematically take on these dimensions in crisis governance, we present a ‘strategic Bayesian operations room’: a table-top exercise for diagnosing long-term risks with causal modelling.
We carried out the exercise with expert officials of a city in Finland. The initial scenario involved an industrial logistics accident and a chemical spill in an urban area, with the scope of the short-term outcomes represented as a Bayesian network model. The participants were to follow the model’s causal logic and discuss how the causal mechanisms and mediating factors related to the accident could unfold and endanger the city’s long-term goals. The participants were also asked to describe and quantify the strengths and uncertainties of the causal links that they identified. After all this input was coded into a calculable Bayesian network, we presented the model and the preliminary analyses that we ran on it to the participants. Their task with this new knowledge was to diagnose what they considered critical risks for the city.
We derive three strands of approaches and results. First, the participants found the exercise to be an attractive method for communicating across expertise boundaries and for taming uncertainties into risks for which to prepare strategically. Second, the probability-based model amplified two kindred socio-cognitive notions: articulations of what is both plausible and possible. We conclude all three categories to be at play when assessing causal mechanisms and viable courses of action. Further, and third, the exercise enacted two ‘separate but mutually constitutive’ modes of qualculation², that is, of both quantifiable and qualitative judgment. They were prominent when translating the deliberations into model features as well as in the participants’ evaluations of the worth of the city’s ecosystems, financial stability, and health and safety.
¹ Bayes, T & Price, R (1736). An essay towards solving a problem in the doctrine of changes. Philosophical transactions (1683–1775) (53), 370–418.
² Callon, M & Law, J (2005). On qualculation, agency, and otherness. Environment and Planning D: Society and Space (23), 717–733.
On the Epistemology of Crises and Disasters