Uncertainty Typology
When analyzing the nature of risks facing an organization to determine appropriate redundancy strategy.
A classification system for different types of uncertainty that require different redundancy strategies: known unknowns (risk), unknown unknowns (uncertainty), epistemic uncertainty, aleatory uncertainty, and ambiguity.
When to Use Uncertainty Typology
When analyzing the nature of risks facing an organization to determine appropriate redundancy strategy. Different uncertainty types require fundamentally different approaches.
How to Apply
Known Unknowns (Risk)
You know possible failure modes and can estimate probabilities, even if you can't predict specific occurrences. Example: Aircraft engine failure modes with documented historical rates.
Questions to Ask
- Can we enumerate the failure modes?
- Do we have historical data on failure rates?
- Can we use probabilistic models?
Outputs
- Design redundancy using probabilistic methods
- Calculate expected failure rates
- Optimize cost vs. reliability through quantitative analysis
Unknown Unknowns (Uncertainty)
You don't know what might fail or how. Black swan events that haven't occurred before. Example: COVID-19's impact on aviation.
Questions to Ask
- Are we facing unprecedented situations?
- Could something happen that's not in our risk models?
- Are we over-optimized for known risks?
Outputs
- Build general resilience rather than specific backups
- Maintain financial reserves and diverse capabilities
- Keep flexible capacity that can adapt to unanticipated disruptions
Epistemic Uncertainty
Uncertainty exists because information is incomplete, but more data could reduce it. Example: New supplier reliability before trial orders.
Questions to Ask
- Would more information reduce this uncertainty?
- Can we run pilots or trials to learn?
- Is this a learning curve situation?
Outputs
- Maintain backup options while gathering information
- Run new systems parallel to proven ones
- Redundancy provides insurance during learning curves
Aleatory Uncertainty
Inherent randomness that can't be eliminated through more information. Example: Weather variability, genetic mutations.
Questions to Ask
- Is this fundamentally random?
- Can better forecasting eliminate this uncertainty?
- Does prediction have inherent limits here?
Outputs
- Accept that prediction has limits
- Build buffers to absorb irreducible randomness
- Geographic diversification hedges location-specific randomness
Ambiguity
Multiple plausible interpretations; experts disagree about risk levels or appropriate responses. Example: Climate change impacts on specific regions.
Questions to Ask
- Do experts offer conflicting assessments?
- Are there multiple valid models of this risk?
- Is there genuine scientific disagreement?
Outputs
- Implement diverse redundancy covering multiple scenarios
- Don't optimize for single most-likely outcome
- Hedge multiple scenarios rather than betting on one