Earlier quoted context omitted.
At least for code, git being distributed means that there's always multiple copies with typical workflows. For example, in addition to the copy on GitHub whoever pushed last should have a completely current copy. And everyone else working with the repo should have something pretty current.
We have 643 git repos. I'm not sure we could reconstruct all of those by going round 300 engineers' machines...
Our company has thousands of repos, some of them years old.