Okay, so as the discussion continues, let me present this.

4 rhythm tracks

drums
bass
guitar
piano

3 vocal tracks

lead
bgv1
bgv2

2 solo tracks

guitar
sax

All of the above are recorded at this mystical magic number of -6db.

I start playing back. (I am not even looking at the index marks on the channel faders. Just doing things by ear.) Bass is not loud enough. I push up the fader a couple of lines. (Here I did not raise the recorded db level of the bass track, it is still -6db, but I DID raise the db level as far as the VU meter sees it.) So now I have the bass sounding nice and full, but man, now that scratching guitar track seems kind of week, so I slide that fader up a line. Nice. Bass thumps in harmony with the kick, the scratch guitar is playing off the hat real nice, but man, this piano is really lost now. Let me move that up a notch.

So here I have had to slide 3 faders up to give the thing some balls. Why did I not just record those three tracks hotter, at maybe -3db? In the logic that has been prevalent in this thread, recording low and boosting is better than recording hot and cutting. However, let's use that quick example above to say "Maybe that piano is up as high as it can go and there is no more fader left unless I take a saw and cut a groove in the mixer." How did recording soft satisfy me there? And if the answer is "do a gain change", how is recording at -6db and then doing a gain change of +3db not the same as recording at -3?

Why don't I want the wave form to fill as much as the track limits as possible as long as I don't hit the top or bottom barriers?