buzz,
Good post, man!

A concept that has helped me in the amplification thinking is the concept that the input to the amplifier is not what's fed in on the -+ lead. Instead, it's the difference of voltages between the i+ and the i-. That, all by itself, explains noise being injected via the ground, without any fancy inducing stuff.

The technique used by some audio cables is slightly different -- there's a hot wire and a neutral wire and a shield. Using your designations, i+ of the cable goes to its other end, o+, i- of the cable goes to its other end, o-. The shield, however, is connected to i-.

The purpose of this is to route electrostatic noise to ground only at one end. Since the source end of any well-designed pair of component will have a lower impedance, the overall system will have less noise if the shield is grounded at its source end.

This is how those puzzling first Monster Cable cables with the arrow were wired. The arrow was to make you plug the source end into the source. I worked briefly for Ryder Sound (Academy Award winning sound company) in the late 60s; their double-phone plug to phone plug rack interconnection cables were wired this way, indicating that this practice goes back at least to the 1940s.

To say it simply, the reason that balanced cables are quieter than other cables is that the wires are arranged with opposing voltages on the two wires, and are twisted such that neither wire is closer, on average, to any surrounding fields that might induce a voltage. As a result, when voltages are induced in the wires, the same voltage is induced into each wire; at the other end, the input "sees" the different voltages on the two wires, which are our signal, and rejects the same voltages on each wire, which are noise. Identical voltages induced into both wires are called common mode voltages, since the voltage is the same on each. The ability of an input to ignore the same voltage on both wires is called CMRR, meaning common mode rejection ratio.

And yes, that's saying it simply.