There are a couple of things going on.
Indeed in Redcloud there was a bug (fixed in Snowmass) that running near 100 with loud or dynamic music (music that stayed for a little above -3dBFS) a crackle could happen. The bug was very dependent on the material being played, but a volume below about 84 would keep it from happening. In a technical sense the bug has nothing to do with the volume control, but that’s a little beside the point.
The DS has a fixed analog noise floor. The volume control is completely lossless but if you are running with the volume set at around, say, 20 then you are running with the noise floor about 40dB louder than it needs to be. The noise may or may not be bothersome, but you will be missing some details there.
If you are happy with the DS volume at, say 100, and you change nothing in the system except the volume, the DS volume is more transparent than turning down a preamp, any preamp.
But if you change the DS’s volume and you change your preamp you can either get better sound or worse sound depending on the gain structure of the rest of your system.
The 20dB attenuator is actually very good, BUT there’s an inherent interaction between the output transformers and the interconnects. This is true for DACs systems without output transformers as well, but with them higher capacitance cables usually roll the top off. With the DS if the 20dB attenuator isn’t being used (“High Level”) the transformer cable interaction is reversed from what most people are used to: higher capacitance (or longer) cables tip the top end of the frequency response up. The peak level of that tip up depends on the cables, but it’s about 6 - 10 dB. With cables that have more capacitance than anyone would sell that whole 6 to 10dB would be in the audio band. That would be very noticeable. With the attenuator engaged the opposite happens - longer or higher capacitance cables will roll the top off more like a typical DAC.
Then there’s the simple matter that louder is almost always perceived as better. Doing controlled experiments with a great preamp to check that lowering the volume on the DS and exactly compensating with a preamp are very hard to do. You’d need to measure a test track with a sound level meter each time you changed volumes. It’s also hard to do with the 20dB attenuator simply because dialing in exactly 20dB with a preamp to compensate is just as hard as an arbitrary volume setting.
I’m not claiming that the 20dB attenuator doesn’t affect the sound, but the method by which the sound is affected is not what most people think.
And for those that haven’t seen some of this before - there is absolutely nothing special about 100. If one claims that the volume control is broken then it’s still broken at 100. 100 doesn’t cause a multiply one somewhere or cause the volume control to be skipped, it’s just a multiply by another really weird number.