What is the experimental or theoretical basis of your disagreement? And what sort of sonic deficiencies are you envisioning, beyond the possibility I mentioned of a minor increase in background noise levels?
Elaborating on my previous response, leaving pins 2 and 3 of a line-level xlr input not connected together will only cause audible symptoms if ALL of the following conditions are satisfied:
1)Noise/hum/emi/rfi etc. is picked up or coupled onto pin 2 and/or pin 3, or the associated internal wiring.
2)That pickup occurs to a DIFFERENT degree on pin 2 than on pin 3, OR that pickup is great enough to overcome the common mode noise rejection capability of the input stage to which those pins are connected (and if the noise environment were that great the un-jumpered input is likely to be the least of the problems that are present).
3)There is a path for the differential noise to be coupled into the internal signal path through which the unbalanced rca input is processed. If a switch is present to select between the balanced and unbalanced inputs, no such path exists. If there is no switch, any such coupling is still likely to be minimal.
4)All of the above factors are present to an audibly significant degree.
In terms of my experience, one of the two balanced inputs on my Classe CP-60 preamp is unused, and I have never bothered to put a shorting jumper on it. More significantly, however, in the same living room I have extensive experience using high quality pro audio equipment for voice recording purposes, the equipment having numerous unterminated xlr microphone level inputs. Since they are microphone level, and subject to amplification that is vastly greater than the amplification that is applied to the line-level signals the op is apparently referring to, that sort of situation figures to be far more sensitive to the effects I have listed above than the op's situation.
My results have been exemplary in terms of noise performance and sound quality (to the extent that I can judge it on spoken voice, and by measuring the noise levels via professional software). Furthermore, that sort of quality pro audio equipment is normally not supplied with xlr jumpers for either mic or line-level inputs, and I have never heard of that kind of equipment being used with jumpers. I see no reason that an xlr input on a piece of consumer gear would be any different in that respect, especially at line-level.