Do you want to argue with math?
In fact, let me put it into better perspective. So, 3900 inaudible samples per second are added to the signal and 3900 samples per second that are part of the original recording are removed. We can combine both of these numbers into one, as they both result in worse quality. So in total there are 7800 samples per second that are not what was originally in the recording.
So thats 7800 samples out of 44,100 that are in every second of the recorded audio.
Now if we do some cross multiplication, you will find that it is equal to about 18 out of 100.
Lets say you are watching a movie that is recorded at 100 frames per second. So in every second you see 100 frames. What would happen to the movie if out of every 100 frames, we made 9 frames blank, and then simply deleted another 9 frames. What would that look like? The effect would be very visable.
And that is pretty much what happens when you play 44.1KHz audio on the Audigy2 ZS.