I'm jumping into this late but here goes.
First, don't assume that people who can't hear the difference in digital cables have a bad system or bad ears. Maybe they are just fortunate that their system components match well enough that the cable isn't much of an issue. Having a system that is very sensitive to cable changes may not always be such a good thing.
Second, I've tested a few ADC's (similar to DAC's) and clock jitter is extremely important in reducing harmonic distortion at audio accuracy levels. As mentioned above, errors in the time position of the sine wave samples will distort the sine wave, just as voltage errors will.
What we're talking about is two issues: data recovery from a serial bit stream and also clock recovery from the same serial bit stream.
Although data bits errors should be rare in a well-designed and well-matched system, they will cause serious problems when they occur. As noted above, there is no error detection or recovery mechanism for data errors. All serial data transmission schemes will miss data occasionally and the measure of that is the "bit-error rate".
Clock recovery circuit design is almost an art. I'm not familiar with the CD standards. I assume that the DAC's clock recovery circuit uses a PLL. Is this correct? The PLL will filter out alot of the high frequency jitter. Better PLL designs will filter out more jitter.
We have equipment to measure jitter where I work. The accuracy is in tens of picosends. The inputs are all 50 ohm. This equpment should be usable for testing the transport/cable jitter. Does anyone have ideas for some easy experiments using this equipment? It might be interesting to measure several types of transports.
