"And then what?"
Because ecosystems are complex, the environmental consequences of human activities are uncertain (Brooks, 1986; Costanza, 1993; Holling, 1993; Ludwig et al., 1993). Thus, uncertainties are an important consideration in engineering designs and management plans. It can be useful to distinguish three types of uncertainty. "Risks" refer to situations where probabilities may be ascribed to various potential consequences. "Unknowns" refer to situations where the range of possible consequences is thought to be reasonably well understood, but the probabilities of the various consequences are unknown. "Unknown unknowns'' are phenomena that one is not even aware one fails to expect or understand.
Even though new technologies are often environmentally preferable to those they replace, their large-scale adoption often results in unanticipated undesirable environmental impacts (Gray, 1989).3 Hindsight suggests there is room for improvement in our ability to anticipate these impacts. Myers (1995) notes that neither global warming nor acid rain were major concerns at the 1972 United Nations conference on the environment in Stockholm, even though Arrhenius had warned of global warming 100 years earlier and biologists were aware that massive quantities of sulfur dioxide and nitrous oxides were being emitted into the atmosphere. Perhaps some unknowns would be better described as unappreciated knowns or ignored knowns. Whole cadres of environmental scientists, policy analysts, and others work on risk assessment, but few are studying approaches to anticipating that Which would otherwise come as a surprise.
Differences in approaches to uncertainty have major implications for environmental policy (Costanza, 1993). Approaches that focus on risks use quantitative models in attempts to identify most likely scenarios, and then use the results as the basis of policy recommendations (Committee on Science, Engineering, and Public Policy, 1992; Nordhaus, 1992). An alternative approach, focused on unknowns, attempts to identify the policy option that minimizes the likelihood of a catastrophic outcome. This approach, which is embodied in the precautionary principle, does not attempt to ascribe probabilities to alternative possible outcomes (Cameron and Abouchar, 1991; Costanza, 1989, in this volume; Daily et al., 1991).
At the extremes neither approach is perfect. Policies based on a most likely scenario will be unfortunate, if not catastrophic, if the eventual outcome is not the one that was deemed most likely. Conversely, reliance on the precautionary principle begs the question, "How much precaution?" Environmental impacts are but one of many important consequences of new technologies. An excess of precaution could reduce innovation and its associated benefits. In addition, even though a new technology may have unanticipated effects, its aggregate environmental impacts may be preferable to those of the older technology that it replaces.