It's a shared/copy-on-write backend. An "entire fork" consists of 20-byte commit ID and whatever metadata identifies the owner and organization, and you create the "entire fork" in the UX by clicking on a single button. I don't understand the criticism here, do you just not like the use of the word "fork" because it implies a copy that doesn't exist?
Not doubting you, but do you have a source where I could read about this? I feel that would create some wild problems if the original repository was deleted.
Edit: I guess they would just not delete the data if there were more references to it.
