The latter. Basically you need to identify which A or B is the 8th generation copy, or inversely, which A or B is the same as the reference. There are real measured differences between A and B, and while not apples to apples they are arguably more significant than the wider skirt at the base of the fundamental with the Gustard's PLL. I could pick one correctly, others just sounded the same to my ears and I gave up.
Okay, My wife and I both did this, at different times. She is always busy...
Result: I picked 4 out of 4 correctly.
My wife picked 3 out of 4 correctly.
The thing that we both noticed is that in at least 2 of the tracks, the 8X version sounded more focused. This made it difficult, because the altered track actually sounded better than the original, but didn't match the reference. Usually, we listen for the better sounding track, not trying to match two tracks, so we had to resist this tendency.
I have a theory about this improvement in focus in the 8X tracks:
As I believe I mentioned in the other thread, some of the early re-clockers used several PLL receivers in series to reduce jitter because one was not good enough. They went from receiver to transmitter to receiver etc.. S/PDIF to I2S to S/PDIF to I2S and on and on. In this case, we are going from S/PDIF or I2S to analog to S/PDIF or I2S to analog and on and on. I think the effect of the series PLL's might be the same, reducing the jitter in the end result and thereby improving focus. Because the altered tracks actually sounded better than the originals indicates perhaps that the compression and harmonic distortion added by the multiple A/D and D/A is less obvious than eliminating some jitter. This reinforces my desire to see the relevance of these different distortions studied with DBT tests. It's not unlike amplitude sensitivity. If you cannot hear it, it's not important.
Steve N.