I dislike these search based metrics for popularity. When I use python, I end up doing a lot more searches, because its a much more fragmented environment . . what version of python, what OS, . . . matplotlib, sns . . . I typically have to try a few tries to get something to work. In contrast with R, you can get a base R solution from a newsgroup that is 10 years old and it works. I think blindly looking at total searches or posts on stackoverflow is not the best metric for measuring the popularity of a language.