The literature on energy technology costs, diffusion, and learning has been characterized by data limitations, partial or arbitrary data sets, apples to oranges comparisons, and imprecision in the use of key concepts and terminology. Two responses to our paper, Lovering et al. (2016), by Koomey et al. and Gilbert et al. reflect many of these problems, conflating learning curves with experience curves, trends in actual costs with the relationship between cost estimates and final construction costs, and component costs with total installed costs. The respondents use inconsistent definitions of demonstration, first-of-a-kind, and commercial deployment across different energy technologies. They also propose to compare final installed costs for nuclear power plants, encompassing construction and finance costs, across different national economies and time periods encompassing a wide range of macro-economic circumstances and finance arrangements that overwhelm any signal from trends associated with the actual construction costs of the plants in question. In this response, we address the specific issues raised in these papers and suggest better practices for comparing energy technology costs, trends, and technological learning.