These two challenges are probably common for any organization, no matter how tightly regulated:
-
Convincing an organization's lawyers to approve the different open source licenses. Sadly, this could also include explaining that "open source" is not the "freeware" of yesteryear which could come with malicious aspects.
-
Legacy code. Converting working programs can be huge effort with no immediate benefit. But if the org has some programs in SAS and others in R, then it has to maintain environments for both languages. Big cost no matter what.
Statisticians aren't seen as developers in my org, so we're allowed to use any tool as long as legal and some tech guys say it's not dangerous. This is great for experimentation, but not so great for collaboration.
I've been convincing others to use R by showing results SAS can't do: reports made with bookdown, nice GUIs for internal tools made with Shiny. I first tried proselytizing the developer benefits: more flexible than SAS, functions are better than macro abuse, sharing code is a breeze. Nobody was buying it, and for good reason:
They hadn't experienced it
\Rightarrow They didn't believe me
\Rightarrow They didn't try learning it
\Rightarrow They never would experience it