If George is saying that the datastream itself is inverted, does that turn lands into pits and pits into lands? I don't believe that acoustic polarity can necessarily be changed simply by inverting the digital signal - but it may be possible to make dramatic changes by doing that very thing.
For instance: could a digital word with a value of, say, 0110110100001110 theoretically be changed to a value of 1001001011110001 by a polarity inversion of the data read from the laser or the machester code of the S/PDIF stream? And are these values simply acoustically polar-opposites of the same amplitude value on the same waveform?
I don't think it works that way.
"In theory there is no difference between theory and practice. In practice there is." - Yogi Berra