Whoops, guess my idea of scaling is still of.
I guess it still matters what use-case you have in mind … I presume you can model a bridge with 300 BUs if you want, then you would see on the “center stage” (the blender grid) that each square is 10x10 meters, so you would see people walking, and have relatively detailed cars visible. I think the main point of this lecture was: if you need this much detail, you are probably not going to model the entire 3km’s of the bridge and you’re probably more focused on what is on the bridge.
I think you can also think of it as: if you would draw the bridge on a piece of paper, if you draw the entire bridge, then you wouldn’t see the individual people and cars either?