I don't understand what "doesn't make sense".
You are trying to establish an equipotential grounding grid within a building. If you parallel neutral current in there it will see the voltage drop (actually "rise") on the EGC.
An example. Lets say you compute a VD (rise) on the neutral conductor in a sub-panel 100' from the service panel at 3 volts. Now if I used the neutral at that distant sub as the EGC the case of my equipment would be 3v above the potential of building steel. If I pulled an equal sized EGC and bonded it the rise would only be 1.5v but that is still not a good thing.
If this sub panel was bonded to building steel the steel becomes part of the neutral path, still not a good idea.
Once you get to another building, with no parallel paths you can start over with a new "ground". It will probably be a different level between buildings but if you don't have any other paths, who cares?
In the data world we had a parallel path and it was going to be the thing that resolved "ground shift" differences. Buy a few routers, LAN cards or system boards and a fat copper wire starts looking pretty cheap.
IBM was very careful to make the wire black and call it a "drain" so we didn't kick the article 250 tar baby.
