In the ongoing debate concerning the origin of hot spots, and more generally, the nature of mantle convection and its relation to plate tectonics, it is important to recognize that viable models need to satisfy all of the data. Much of the recent literature has emphasized one or two subsets of geoscientific data (typically geochemical/petrological, tomographic, or local or whole mantle convection), almost to the exclusion of other data types. In their Comment, and in other contributions, Liu, Kolenkiewicz, and their colleagues provide important reminders of this need, as they explore the implications of global gravity for convection models.