Hi Lynne,
I agree with Steve's comments about length, and I don't doubt that his cable is an excellent one. As he indicated above and in his Positive Feedback paper, the optimal length is dependent on signal risetimes and falltimes, which will vary from transport to transport and are usually unspecified. But as indicated in his paper, what probably stands the best chance of being optimal in any given setup is a length of around 1.5 meters (about 5 feet), assuming that a very short length is not practicable.
In addition to signal risetimes and falltimes (the amount of time it takes for the signal to change from its low voltage state to its high voltage state, and vice versa), what length is optimal in any given setup can also be affected by the propagation velocity of the particular cable, by the accuracy of the impedances of the cable and the components it is connecting, and by the amount of electrical noise that may be riding on the signal that is provided by the source, or that may occur as a result of ground loop effects or RFI pickup. So it's all somewhat unpredictable.
Regarding your second question, most likely the length of the ethernet cable is unrelated to the poor sound. The timing of D/A conversion in the AVR (which is where jitter matters) and the timing of data transmission via ethernet are inherently unrelated. It's conceivable that the cable type and/or its length could have a slight effect on sonics by affecting the amount of digital noise that could radiate or couple from the cable to circuit points within the AVR, but I'd expect that kind of effect to be subtle at most.
Best,
-- Al