I've said it before, and I'll say it again: An empirical investigation of the upper bound of the selection approach to dialogue