As nations engage in the long-term process of negotiating a protocol on controlling greenhouse gas emissions, the danger of costly mistakes looms large, both from doing too little or too much. Researchers try to provide insight and guidance on this difficult problem, many relying on the tools of mathematical simulation models, but two important shortcomings remain in much of the current analysis. One is the treatment of the uncertainty inherent in projections of economic activity over the next century or more; and the other is the way economic activity is modeled to impact the poorly understood physical and biological systems of the earth.
    This paper presents several simple illustrations of the performance of current policy proposals in the face of a long-term climate-based goal when uncertainties in economic growth and technology development are made explicit. Several conclusions emerge. Analyses that rely on deterministic emissions paths through time obscure the underlying uncertainties, and explicitly incorporating these uncertainties can produce qualitatively different conclusions. Moreover, apparent differences in the climate impacts of proposals now being debated within the Framework Convention on Climate Change may be negligible when viewed in the light of likely uncertainties. Seeking a less stringent protocol with a higher probability of compliance may thus produce preferable outcomes. Most importantly, the analyses presented are meant to highlight the need for flexibility in any response to climate change, because uncertainty is an unavoidable aspect of the problem.