The reason why is that each file gets a specific hash ID, this lets the machine find the exact sample file regardless of where it is in the file system. If you were able to load duplicates you could run into a situation where you do not know which file with the exact same content is loaded/used by your project.
It is possible to have aliases in the file system, but I don’t know if it’s a possibility to solve that via Transfer, or if it even is such a good idea to allow for that in the first place. (Could get confusing quickly!)