Judit - I agree wholeheartedly with your concluding sentiment, but I'm not sure such a thing is indeed achievable in practice. I'm also not so sure about some of the assumptions put forth on the physics side, but undoubtedly the EE's will come along to expound. :-)
Benthar, you and I will probably never totally comprehend this in depth, and I'm not sure that it is possible for even the cable designers themselves to completely understand everything that goes on. As to the "Why?", and omitting the obvious "What?" and "How?", I think the simplest fundamentally true explanation is that different cables are just that - physically different from one another. No condutor or circuit is perfect, and the same input passed through different ones will result in different outputs. What I am saying, in other words, is this: It wouldn't make sense if different cables sounded the *same*. They *sound* different because they *are* different. While we may struggle with the issues of prediction or explanation, there is really no more mystery as to why dissimilar wires should sound unlike each other than there is about the fact that dissimilar components don't sound the same. For some reason, people just tend to be less disturbed by the latter proposition, but the underlying reality is the same. They sound different because they pass the signal differently (and different signals differently as well), and also because they electrically interact with the other components they're connected to differently, which themselves vary just as much or more. The variables are unlimited in theory (including the whole system with its environment, the source material, and the listener), so the sonic differences are too. Which is why, as we all know, there can never be one 'right' or 'best' choice based on anything other than what you hear and prefer.
P.S. - A while ago when I was comparing some cables, the ultimate ramification of the situation fully hit me: If we were really to attempt to do this cable thing the whole way, then we would have to audition and reselect the best cables to use with *every recording* we listened to. And that's why I say I sympathize with Judit's closing statement - in the final analysis, all you can really do is try to choose the wires that best accomplish her stated goals of neutrality and accuracy (as best you can judge them) in one's system with the majority of well-recorded material, and then let the chips fall where they may after that.