Planning under uncertainty
Even when working under conditions of great uncertainty, there are strategies which can be used to ensure a more effective response.
We have previously looked at factors that can generate uncertainty about systems operations and future predictions, including perceptual unreliability, gaps in knowledge, conflicting perspectives, powerlessness and irreducible randomness.
However, the total impact of uncertainty on planning ranges on a spectrum falling somewhere between total determinism and total uncertainty. The polar extremes do not exist in practice, since this would require either a universe of zero free will or the existence of purely arbitrary magic, respectively.
To develop a suitable approach in response, there are four basic modalities of uncertainty that have distinct strategies for goal-setting and prediction, ranging from the most stable to the most unpredictable environments:
Level 1 uncertainty
When in a state of level 1 uncertainty, outcomes are assumed to proceed predictably in almost all circumstances. Planning assessments focus on the impacts of low-probability deviations from expected results (planned loss events), and the vulnerability to loss if any underlying assumptions are wrong (unplanned loss events).
Key tools used in level 1 include:
Single adverse event exposure management
Sensitivity analysis of assumptions
Level 2 uncertainty
In level 2 uncertainty, the system under assessment yields useful results when modelled, including the assignment of probabilities of relating to input variance, resultant actions, and value outcomes from expected future scenarios.
Key tools used in level 2 include:
Assessment of expected value / utility
Conventional risk management strategies based on modelled futures
Level 3 uncertainty
When operating under level 3 uncertainty, the probabilities of inputs or outcomes for the system cannot be determined with confidence, but there are a limited range of plausible futures resulting from the situation to consider.
Key tools used in level 3 include:
Scenario analysis to consider planned policy responses that will produce an acceptable planned or interim outcome for each scenario
Fitness-informed policy design
Planned outcomes are those in which the actioned policy will deliver a benefit or acceptable level of loss.
Interim outcomes are ones in which the actioned policy has passed a threshold for continuation, requiring the activation of a contingent policy response to develop a new approach.
For scenarios with interim outcomes, no attempt is made to forecast an acceptable outcome beyond the trigger for the contingency response. For example, disaster recovery plans do not plan in detail how system restoration will be achieved, but only for the successful activation of the disaster recovery team who will then design and implement a plan to stabilise and respond.
Policy design is always selected based on a system fitness trade off, for example:
Productivity policy – A policy with low sensitivity to loss and low contingent planning will deliver successful outcomes at lower cost if the forecast scenarios come to pass. Vulnerable to catastrophic collapse.
Resilience policy – A policy with high sensitivity to loss and high contingent planning. Low exposure to loss but unpredictable in delivery due to rapid pivots, which can make the organisation an unreliable partner.
Robustness policy – A policy that focuses on mitigating loss exposure to the organisation within default policy actions. This can be expensive to implement, but makes the organisation predictable across a greater range of scenarios.
Level 4 uncertainty
Finally, in level 4 and the deepest level of recognized uncertainty, either indicative but non-comprehensive sets of plausible futures can be described (level 4a), or the only thing known is that the set of possible futures is not known (level 4b). Level 4 uncertainty is rare, requiring that either:
No amount of information gathering would reduce analytical uncertainty, due to a comprehensive lack of knowledge about how to explain or predict the system mechanisms at play in the current situation, or
Future scenarios can be described and would be recognised as plausible after the fact, but there is no known way to leverage existing knowledge and past data to predict the circumstances under which it could arise in the future
Key tools to manage level 4 uncertainty include:
Action constraint analysis
Counterfactual exploration of unknowns to build assumption sets
Broaden environment scanning and weak signal association
Deep resilience development using contingent policy and assumed adaptation
A disciplined approach to uncertainty assessements can transform uncertainty from a source of action paralysis into a strategic driver of preparations across your organisation’s activities, supporting better outcomes delivery across the full spectrum from strong predictability to the fundamentally unknown.