I was listening to a podcast interview with @andrewnez and, TIL:
The package management for the programming language R, used mainly in science, has no version pinning and will just remove packages in the event of compatibility problems. As was noted, this would be a problem if science ever concerned itself with reproducibility.




