The aggregated term structure of social discount rates that results from Weitzman's (2001) survey of expert opinion is shown to be highly sensitive to the nature of the responses. If variation reflects irreducible differences in ethical judgments, the term structure can decline rapidly. If variation occurred because respondents were forecasting future rates under uncertainty, the term structure is much flatter because additional experts provide new information. The former approach triples the social cost of carbon when compared to the latter. The distinction between heterogeneity and uncertainty illustrates the need for a nuanced treatment of survey data in intergenerational policy making.