Balanced signals are widely used in professional gear, and twisted pairs indeed reduce loop area, but it’s not perfect. I once helped out in a sound booth where the cables picked up AM radio, and some nonlinearity in the system made it audible.
This right here is part of the reason the "coathanger versus cable test" annoys me so much. All these wonderful graphs "proving" that cables don't matter never take environmental noise into account, which is frustrating because in reality that's exactly what you're testing.
The fact that you could pick up AM radio inside a sound studio means that, for experimental purposes, that environment is simply not good enough.
Are expensive cables mostly bullshit? sure. Do better cables still make a difference? My personal experience tells me absolutely yes, just usually not enough to matter to 95%+ of people.