What's better, one conductor or two conductors for an RCA interconnect?

You are looking at this more theoretically than practically. Excuse me as my expertise is physics (and managing EEs) not day to day EE and a lot of time around semiconductor processing equipment, some of it quite noisy while trying to do precision measurements. We are not passing a reference point. We are connecting the reference point on two pieces of equipment and making them the same. The impedance on the ground connection is not the same as the signal connection. An equal induced current on the ground creates a small voltage only related to the ground circuit. Induced current on the signal creates a voltage a factor of the impedance . You are making the assumption there is not a relevant secondary path for ground, for instance totally disconnected or a large value resistor. How large is a large value?  How large does that resistor need to be to ground so that it is not there?  Poked around enough in tube amps to know that resistor may not be very large at all. What if there is a capacitor?  10 nanofarads? That is only 16K resistance to ground at 1KHZ.

Is it better to connect the ground to each side of the RCA cable?

Single ended RCA cable, the ground is connected on both sides, there is no choice. If there is an additional shield, typically is it better to only connect it on one side.

@deludedaudiophile So, you are not EE.  Well, I am - designing low level electronics for 40 years.  Sorry I cannot explain it to you better.  We agree to disagree.

I have a PhD in materials science and close to 2 decades in semiconductors, semiconductor processing, and associate equipment, including forays into process measurement and control ... well exposed to signals in noisy environment. Doing much of the same in batteries now.  I understand the overall issues quite fine and could probably derive a lot from first principles. I understand your argument, I don't agree with all of it and I think you have poorly argued some points specifically as it may apply to real world applications including equipment ground connections and how they relate to loop resistance or impedance. Given the indeterminate ground connection and how that related to the loop resistance all due to the signal connector, I believe you are making erroneous conclusions about what would be common mode noise injection.

A speaker connection is balanced from the standpoint of the speaker, but if you induce noise on the speaker cables, it is not balanced at the amplifier input hence why the amplifier circuits for sensors and bridges, both often floating, are still fully differential circuits.