Climate change and many other environmentally related policy problems are members of a class of “wicked problems” (Rittel and Webber, 1973)—problems with no definite formulation and no clear point at which the problem is solved. They have been described as having five key characteristics (see Dietz and Stern, 1998):

1. Multidimensionality: A single environmental process or policy can have many different types of effects, distributed unevenly so that those affected face unequal shares of the costs, risks, and benefits.

2. Scientific uncertainty: Current understanding is primitive in comparison with what decision makers want to know—and sometimes the degree of uncertainty is itself uncertain. In addition, the consequences unfold at an unfamiliar tempo, with some effects delayed and others disconcertingly prompt.

3. Value conflict and uncertainty: People differ in the importance they attach to the different effects of any action, and these judgments change as people experience how their own and others’ actions affect the things they value.

4. Mistrust: Decision makers are often mistrusted by those their decisions affect; their analyses are also often mistrusted.

5. Urgency: It is not feasible to postpone action until scientific uncertainties are resolved.

In addition to these characteristics, climate change presents a dynamic decision context and unfolds over a time scale that extends beyond the planning horizons of most organizations and over a geographic scale that exceeds their control.

Learning by doing under such conditions creates challenges for leadership. Although it makes sense to treat all decisions as provisional, such an approach is not easily reconciled with conventional notions of accountability. Decision makers will have to discard well-accepted standard procedures that offer them a kind of protection in favor of new ones that may be more effective, but that will open them to criticism when, inevitably, errors occur. Another challenge is that most climate change decisions will be undertaken in a decentralized fashion, as local and state governments, firms, and other institutions respond to a changing climate.

Thus, the federal role in decision support will have to be aimed at creating and informing a distributed capacity to make sensible choices. This is both functionally necessary and advantageous to the nation as a whole, since decentralized decision making will generally be better able to cope with surprises and specific local conditions. Nevertheless, federal