Ive personally always understood edgeloops as making sure your geometry doesnt try and deform diagonally on a grid
Well yes and no its a side effect of the reason you use them. Its more because the screen and your eyes and your brain sees a 2d world. They operate well with lines and if you have 2 crossing lines that form a well defined surface with lines next to them then its easy to understand. It doenst form undesireable diagonal chenges because you nderstand the relationship of those moves better, so you also see that the bind works better.
It doenet mena that the surface in question can not behave right with other shapes. There are in fact a unlimitted of those shapes. They are just not well understood thats all.
I gues I'll go midways and just add extra edgeloops for abs or byceps/chest.
Understanding part of the surface amy be better than nothing. yes. But in my experience they coount more in faces, shoulder area, ankles and hands
(which would be quite jaggy).
jaggines again is because you have no visual clue as to when the surface aligns up to a even shape, as you have no curve going that way. A surface is even if its even in 2 directions, hence edge loops that try to cross as perpendiculary as possible you make those edges even and that ensures that the others get even.
If your main curve intersect a area that uneven youd see it instantly and not make thet mistake. So you choose lines that count MOST to you.