Although not guitar related this is synth related but I am sure there are similarities.  Lets say you have a synth creating a nice pad sound but you want to make it more lush and you start to overdub it etc to make it bigger etc.  The fact is once you go above three tracks the extra you get is very little.  3 is sort of magic number.
 
I used to have a Yamaha TX816 module with all 8 DX7 modules in it.  There was a mode where you could send the same sound to all 8 modules at once.  I even had the outputs connected to an 8 channel mixer so I could mix and pan all the modules.  I used to detune them all slightly too for a bigger effect.
 
Anway what I found was one module on its own sounded OK.  Two together  (producing the same sound) definitely sounded better and fatter and three was even better.  But after that there was little to be gained. Adding the 4th/5th etc.. modules in did not do that much and so on.  There was not a lot of difference between 8 modules doing the same sound as compared to three in fact.  I also got a very wide stereo effect too by panning them Left, Centre, and Right.
 
The older Kurzweil synths only allowed you to layer three sounds on one note and I can sort of see why.  Once you get three layers going things are pretty complex and fat. (New ones allow you to layer 32 layers! but that is extreme I think)

I haven't had the chance to listen to the links posted but there seems to be some contradiction, well, maybe not contradiction but different approaches to this. Many of the most powerful sounding metal bands use lots of tracks and get BIG results...how they do it exactly ...I don't know, but they do it well. I would agree that , as it seems to me, the more tracks you get the more chances of phase cancellations if tracks are not exact and the more chance for tracks fighting one another. Are they cutting off tails? or playing with timing after that fact? I have no idea, but there is a technique to it that I don't think those who use are chomping at the bit to tell us about. I guess the rest of us use two or three tracks with great success.
 
Jeff mentioned the similarities to tracking synths multiple times...another example that comes to mind is Enya...she( or her engineer) used dozens of tracks, and sold millions of albums, but I think they were after the distant dreamy effect and it worked well for that.
 
I think the more tracks of the same or a similar thing you record, the more important alignment becomes.

davdud101
So, I've always wondered.... A lot of artists, especially in metal, like to track 4, 8, 16, ridiculous numbers the same guitar part. But I've always wondered- in a situation where it were possible, what would it sound like if someone were to track 100 guitars and pan each one at a single interval on the panning spectrum?

If you used the same sound 100 times, you'd gain volume and a a nice stereo field based on the human timing inconsistencies. The key in today's times is no one is using the same sound. When we record multiple guitars today, we call it layering. But you don't want to use the same sound. You'd want a different timbre. A different guitar, different eq curve, a different amp. This, as well as the human timing inconsistencies AND your panning, creates incredible stereo sound. Much cooler than the same sound 100 times, I promise. :)
 
I have backing vocal parts of myself where I literally sang the part 30 times. It basically sounded like 30 me's...which was cool for the times. But then we got software that gives us a feature called "formant" and THIS allows you to change the sound of the voices. So now when I do a 30 part vocal of myself, using the formant feature alters the sound of my voice just like me changing guitars, amps and eq settings for each new guitar sound. I'm doing the same thing with my vocals. Each vocal I record using the formant will need to be eq'd for that voice sound. When you add all THESE vocals up, it obliterates the 30 vocals I did with the same voice when compared to 30 voices using the formant option, understand?
 
Now some guys like to record the same sound multiple times. If you do it tight enough, the sound gets really phat but it doesn't actually gain sound size like it would if you altered the sound. When you don't alter a sound multiple times, you only pick up stereo from human timing inconsistencies of the multiple takes and any panning. When you alter the sound and record each time you alter the sound playing the same part, you still get the human timing inconsistencies but now you get the stereo sounds of multiple eq's/sounds as well. So it makes a much bigger difference.
 
There are times when layering is the better bet...or times when triple tracking the same sound may be in your best interest. Randy Rhoads, who used to play with Ozzy was a master at multiple takes. He tracked them so perfectly being able to play consistently, he was a machine at it. "Diary of a Madman" is one of the tracks where he went nuts tracking multiple guitars. Though a dark tune, it's a beautiful piece of music when you just hone in on the acoustic guitar parts. :)
 
-Danny

I think 100 is too far, I could see 99? But 100? it's madness, MADNESS I SAY!

Actually, TB, your comment made me think- 0C-thru-100L is 100 steps... 1R-thru-100R is 99 steps, so there are actually 200 places to put stuff!! That is correct, right?