Why do you think Bi-Wiring improves the sound ?

As for kool-aid  It worked for the Grateful Dead !

Well, that kind of kool-aid may improve the sound of any system. I know for a fact i prefer the sound of my system after a glass of good wine or Scotch.  really.

I wont even comment on the benefits of bi-wiring, + or -.  I have my biases and i have not done it it too long to feel confident.

But i will say that Millercarbon's comment is the most important one that we should all keep in mind, and almost no one ever asks.Many things work. The question is always "where can i get the most benefit for s given expenditure"
Right now i am placing most of my effort and money on sources (DAC, low jitter/noise sources, cartridge) and on room/setup. The sound, as it sits, is frankly amazing so I'm not out to change any fundamentals. 

I believe most people can get the biggest changes from setup, room and source material (e.g.: the best recordings).  Oh, and i make (or more accurately have and will) the costly electronics stuff i'm down-playing :-)
The great thing about bi-wiring is that with many speakers you can do a low-cost test -- even if the test is with lesser cables than the ones you own.  This way you can separate two effects:  a) wire sound (if any) and b) bi-wirign affect (if any).
G

Violating my own promise, I will make a few very simple observations:
For reference my system is quite extensive and has been tuned (e.g.: room, TT setup) pretty darned well, although i am not a maniac about it.
1. I have heard small, but meaningful differences between interconnect cables, and much smaller ones for speaker cables. real, but generally small.
2. I mostly heard difference vs cables that we would all agree are crap.  That's a technical term, but think generic, 40 year old, from some box somewhere int eh attic
3. I have seen myriad cases, in my system and elsewhere of cables that are basically broken.  Corrosion. loose terminations, poor contact etc.

I speculate that sometimes changing wires fixes these issues and is heard, but the reason is sim-attributed.
4. Want to improve wire gauge?  Make the wires shorter.  Now, to be honest, with an 8-ohm speaker impedance, or 30,000 ohm amp input resistance, does the difference between 0.05 and 0.03 ohms matter?  I'll leave that up to you.
5. Construction - e.g. dialectic material, is much more important than many other esoteric factors. I do go out of my way to use interconnects with expanded polyethelene or foam teflon or something similar. Its nto all that expensive anyway.
6. Many cables are so think and stuff that they place significant strain on connections.  I have seen many, many examples of bad connections because someone is using $500 cables that that thicker than my thumb and stuff as a garden hose. Oh wait, they were garden hoses, scratch that. In fact i have had to service equipment that was actually damaged by the strains (like expensive [brand omitted] terminals breaking... think about that)
And as MC said, in most cases the money and or time could likely have made a much bigger difference elsewhere.
having had too many Vandys to count, staring in the very early 80s, i never benefited from bi wiring BTW. Of course, since my system is also a test bed, i really hat to have complicated connections, rinnign from lab ot system and spending 20 minuets fighting with wires is not my idea of productive use of time. SO like i said, i'm biased.
G