If you're going to do this for real, I'd suggest measuring the output with and without the attenuator to determine if it's exactly 12 dB or whatever the real value is. When doing the comparison, with attenuator out, compensate with -12 dB (actually, the exact value you measure) in software. Doing the normalization by rotating the volume knob is difficult and would require redoing measurements every time; also, doing -12 dB in software simulates the alternative solution.
To be honest, -12 dB in software is not such a huge deal at all, particularly for a 24-bit DAC (though Modi's effective number of bits is unknown). Technically using the attenuator could result in better SNR, but it depends on the DAC and probably wouldn't be noticed anyway. Sure, 5x gain may be overkill for some people, but that's just 14 dB over no gain and is probably no issue to reduce in software unless maybe you're using very sensitive IEMs.
If anyone wants to buy (or make) attenuators, they could easily also buy (or make) new RCA cables as well that aren't so unwieldy. Don't need to use the thickest things in the world for something like line-level audio interconnects at minimal distance.
Edited by mikeaj - 1/23/13 at 3:07pm