I think on the whole we're going to care less and less about the distinction between language bindings in data science as time goes on. If something like Apache

I think on the whole we're going to care less and less about the distinction between language bindings in data science as time goes on. If something like Apache Arrow takes off and we end up with a decent standard representation for dataframes (either in memory or distributed), and most of the heavyweight processing (e.g. XGBoost, TensorFlow etc) is written in C/C++ anyway, then I don't massively care what languages people are using to express themselves (and I personally think tidyverse on the ingestion side and ggplot2 on the output side win here).