The problem is that people doing data science are not developers, so instead of just using whatever is there they are reinventing a terrible version of package

The problem is that people doing data science are not developers, so instead of just using whatever is there they are reinventing a terrible version of package management.