In a new paper published in Earth’s Future, Klaus Keller (Dartmouth, formerly PSU) and I try to link the exploratory-modeling, robustness-centric #DMDU philosophy back to probabilistic models.

Using formal probability models has many advantages. For example, we can deliberately over-sample low-probability regions of the parameter space without distorting our estimates. We can also use simulation-visualization techniques to explore the implications of our (inevitably flawed) assumptions.

Many classical approaches to climate risk management have used excessively narrow estimates of future risk (sup 👋 bulletin 17C) as a basis for “objective” cost-benefit analyses and the like.

A growing literature on decision making under deep uncertainty (#DMDU) rightly rejects this, instead emphasizing methods that identify plans that perform well over many possible futures.