Morgan said that he and David Keith have critiqued scenarios on various grounds: detailed story lines fixate people on those particular stories, whereas there are other ways to get to the same end point. Overconfidence is ubiquitous; putting probabilities on scenarios is problematic; and path dependencies—the order in which different changes happen—can make a huge difference. Evidence indicates that consensus processes tend to understate the level of uncertainty, as indicated by individual experts’ judgments. The literature produces a range that is narrower than the judgments of the individual experts.

  1. Bounding analysis. An alternative to story lines is to ask people to suggest all the conditions that could lead to very high and very low values of a parameter of interest and have the list reviewed by experts to cull out infeasible conditions and suggest how the extreme values might come to pass.

  2. Working the problem backward. If one propagates probability distributions down the causal chain, the uncertainties get huge. An alternative is to go backward on the chain, from outcomes of interest to the paths that could produce them. What possible outcomes do people most care about, and how could they get there? Which things would be problems that need to be addressed and avoided? People have trouble with this method. Research has found stakeholders refusing to do it, because they saw it as unwarranted speculation about bad outcomes.

Morgan said that methodological uncertainties in scenarios are often not appreciated. For example, time preference is modeled in ways that do not correspond to what people do, and important feedbacks in physical or socioeconomic systems are sometimes not taken into account. The uncertainties in the methodology can be larger than those that apply to the technical issues. Also, assumptions about who makes the decisions are sources of uncertainty.


Brian O’Neill spoke about the relationships of demographic drivers and emissions on climate outcomes.2 First, he noted that the relationships between story lines and demographic drivers are often thought to be much more solid than they are. Second, emission scenarios can be consistent with a wide range of demographic drivers. The A2 SRES

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement