Interesting about the spiral on the label side being clearly visible from the dye side. Certainly if the reflective layer is too thin, that which is on the other side of it could affect the ability to read the data. The correlation between color of disc and burn quality is intriguing in this respect. I'd like to speculate that with an excessively thin reflective layer some label colors, when viewed from the recording side of the disc, enhance the contrast/reflectivity of the recorded dye/reflective layer while others degrade it. I can't offer any proof of this, though. And the spiral itself offers some argument against the hypothesis:
All of the discs, regardless of color, share a common characteristic: the silver portion of the spiral. That would rule out the color of the label improving the contrast/reflectivity, because such an argument wouldn't apply to the silver portion of the spiral that every discs shares. Errors on the silver portion would show up on every color disc. But it doesn't rule out the color of the label, as seen through an excessively thin reflective layer, degrading the contrast/reflectivity. The discs might be readable on the silver portions and green/orange portions, but not the red/blue portions.
It would be interesting if we could correlate the color of the label to the electromagnetic spectrum and the laser wavelength, but it doesn't work out so neatly. The laser and the poorly performing red discs are together at the opposite end of the spectrum from the poorly performing blue discs, with the passable green and orange lying in between. On the other hand, the visible spectrum characteristics of the label pigments might have nothing to do with their characteristics in the infrared portion of the spectrum.
It could also be that the correlation by color has to do with the manufacturing timeline. It is possible, although not likely in my opinion, that the different colored discs were made at different times, and that explains the observed discrepancies. It is more likely that they were at least made on different lines within the plant, but again, who knows?
This is all just empty speculation on my part. Why don't you write to Fuji and see what they have to say. They do provide a lifetime warranty, and they might be interested in knowing they had a quality control problem.
P.S. Did you try running CD Doctor at 8x CLV and see how the discs read at slower speeds?