Wire, Dire Wire
I have not read all the posts in this thread yet, but I will still inflict my useless take on it anyway.

All I have to back it up is fifty plus years of listening experience, and being raised by a DIY father who was an engineer at Shure Bros.
I will preface by stating that I have never been a fan of meter men who pay attention to readouts instead of music, and that I think Julian Hirsch was an ass and deaf to boot. But snake oil is snake oil and the wire debate seems to be awash with it. Don’t get me started on power conditioners.
I will also say that I am an atypical listener in that I listen at night and without distractions, so this stuff matters to me. For some reason unknown to me I am incapable of casual listening. If I don't like it enough to pay attention to it, I don't want to hear it at all. (After giving it several fair tries, of course, usually over several years.)
Crossovers are indeed an evil source of distortion and especially phase problems. One of the big advantages of headphones is their lack of crossovers. When not using headphones, go Stax, I use different kinds of speakers, but for attentive listening I use a single driver augmented, crossover less design. Hammer Dynamics Super 12s, if you care to know. But the fact that crossovers can do many times more signal damage than speaker wire does not change the fact that wire could matter. But does it matter?
I have currently settled on magnet wire for speaker use. I do not use large conductors. For many years it has been supposed by users of high efficiency, crossover less speakers that wire more closely related to voice coil diameter than the wire industry’s recommendations is appropriate. They have observed that too much wire adversely affects the mids and highs, and my personal experience agrees with this. Things simply sound less clear, given a system with enough resolving power to render the difference. I admit that most systems do not resolve the differences, but that does not change the experience of owners of systems that do.
If you are not familiar with magnet wire, it is high quality drawn solid wire coated once or twice with a thin dielectric designed to resist extreme thermal and mechanical conditions (those found in transformer windings and electric motors for instance). On good magnet wire the two coatings are different. The coatings are also mechanically very tough, as scraping the stuff off to make a connection will demonstrate. Those not up for that challenge can spend a lot of money and buy a solder pot. Magnet wire is very inexpensive. Being an industrial item it is not oxygen free, so if that instills O2 paranoia too bad for you! It is just high quality, reasonably priced wire that works in boiling temperatures and conditions of extreme magnetic and electrical flux without any leakage from insulation shortcomings. I order one or two pound spools of the sizes I need for a few dollars each. I have never paid more than twenty dollars for such a spool of double insulated magnet wire.
If you wish to have a mental illustration of how senseless fire hose cables are, picture one connected directly to a voice coil wire. How much of the conductor do you suppose is overkill?
I think that interconnect wire for low level signals suffers from a different set of misinformation. While balanced is superior in theory there is seldom if ever any audible advantage in the short runs typically found in home music reproduction. Sound reinforcement is another but unrelated matter, where small signal wires run a great distance over terrain littered with AC and speaker lines. If you don’t live under a high tension power line or next to a radar station, forget it. Some use long lines to amps because they are concerned with losses in speaker cable runs. It is your rig, do as you like. It is still obvious that small signal interconnects are more prone to distortion and loss than speaker lines, and by several orders of magnitude. So keep interconnects short and run speaker wire where you need to go with it. Even when using balanced for the long runs to the amps, it is a wrongheaded approach that puts the cart before the horse. In a world of expensive speaker cables keeping the runs short at the expense of interconnect length becomes more an economic than audio quality decision. It is like cutting down the dosage of a life saving drug you need because it is expensive.
What does seem to make sense is braided wire at low signal levels. Not at all necessary, but a cheap and engineering backed precaution. I sometimes use Kimber’s least expensive line of interconnects, and I really should be making my own from ordinary wire but I need to support my local audio shop with a high profit item now and then.
What about skin effect? A simple engineering table from a reliable source (that would be the IEEE) will show when skin effect is a factor. It ends up interconnect wires can be pretty small, and speaker wire need not be large to negate the possibility of audible effects. Unless of course you use low efficiency speakers (never a good idea if you want the best sound) and need an audio bandwidth down to 5HZ.
How about all those differences heard in A/B listening tests that seem unlikely or make no sense but are heard anyway? I had those put to rest for me when an amp was changed over from pentode to triode in an audio shop. Most present commented, at length no less, on the differences, quantifying and qualifying them with the usual audiophile terms. Then it was discovered that the switches had not been toggled after all. The amp was still in pentode. I rest my case.
Clark