Don't know, can't know: embracing deeper uncertainties when analysing risks.

Area of Science:

• Risk analysis and modeling
• Decision science
• Climate change science

Background:

• Formal models are essential for risk analysis but are subject to various uncertainties.
• Understanding and communicating these uncertainties is critical for effective decision-making.

Purpose of the Study:

• To propose a structured approach for assessing and communicating uncertainties in risk models.
• To examine the challenges in expressing uncertainty, particularly in climate change assessments.

Main Methods:

• Development of a five-level framework for uncertainty assessment: event, parameter, model uncertainty, and extra-model uncertainties (acknowledged and unknown inadequacies).
• Analysis of how uncertainties are expressed and communicated across these levels.
• Drawing parallels with evidence-based medicine for assessing the quality of evidence.

Main Results:

• Uncertainties in risk models are contingent, arising from judgments based on potentially inadequate assumptions.
• A five-level structure helps differentiate and communicate various sources of uncertainty.
• Criticisms of existing methods, like those by the Intergovernmental Panel on Climate Change, highlight challenges in separating likelihood from confidence.

Conclusions:

• The contingent nature of risk modeling must be explicitly communicated to policymakers.
• Unconditional expressions of uncertainty in risk assessments remain an ideal rather than a current reality.