First off, let me start by saying that I tried to search for this sort of thing but quickly found out that I'm not exactly sure how I can get the results I'm looking for.
But consider this:
You have a collection of tracks to be put together on a CD. You have the order and spacing figured out. You're content with the consistency of volume/panning/timbre between the tracks. You don't hear any other common offenders (noise, etc.) in the mixes themselves. The mixes themselves sound good individually. In short, you're happy.
BUT, for various reasons, let's say that your volume levels are quantitatively low. And by that, I mean your PEAK gain on the songs maxes out well below full-scale. For convenience, let's call it -6 dB. You don't need to compress it or anything, you're happy with the intrinsic loudness/crest factor you've got. Is there any real harm in bumping up the fader an extra 5-6 dB?
On the one hand, my intuition tells me that it's a transparent change, but it will improve SNR on playback systems.
On the other hand, my guilty conscience (abused by the horror stories of the loudness wars) tells me that this is basically "using up all the bits" and shouldn't be done, because it doesn't "improve" the way the mixes sound. That this change is just trying to keep up with Jones' a little bit in terms of overall loudness.
On a third, mutant hand, I get the distinct impression that maybe I've already over-thought this.
So I'm curious as to what everybody's reaction to the scenario is, and why.