If you are producing your own CDs, as many here are, then there is no reason to even conceive of a mastering process as anything not integrated with the whole production process.
While it is possible that an all-in-one-guy production *can* take care of some of "mastering's" nominal tasks during mixing or on the mixdown itself, I don't see it as making a whole lot of sense to finalize the mix volume until you have an actual mix to finalize and a target finiaization "sound" to actually shoot for.
Is it possible to just throw some (or even all) of your 2-mix processing right on the master bus *as* you create the mix (as opposed to creating the mix and saving it first before moving on to the next logical post production phase)? Sure, if all you're doing is creating a single with absolutely no future chance of having to integrate it into a collection of some type such as a compliation, anthology, EP or full album. But even in the case of making a single only, it still makes complete sense to me to use the mixdown as a stage at which you save what you have so far before moving on to the next steps. And if you're mixing a song intentionally as part of a collection, it makes no sense to me to do any but the most minimal processing to the 2mix until you have all the 2mixes together to examine as a group.
It's no different or better IMHO, FDIC, to master a song solo when it's going to be used in a collection of songs than it is to "mix" an instrument track to get it sounding great solo before you check to see how it fits or not in the rest of the mix.
And as far as the -10dB thang, I can only speak for myself, but it's not that I meant or even advocated that -10dBFS - or any other specific number, FTM - as a *target*. I only extremely rarely advocate mixing or mastering to specific numbers, as you probably well know. I also was treating that -10dB number as a ballpark figure meant in context to indicate that having the mix numbers come in well below full-scale is not a bad thing.
For me, the actual numbers that the mix comes out with - with a master buss set at unity and with no master buss processing, as I almost always mix - will vary depending upon track content and can easily *naturally* come in at something along the lines of -10dBFS peaks with RMS levels in the -17 to -22dBFS range. Sometimes those numbers may wind up a few or more dB higher. Sometimes (though not as often, probably) even a couple of dB lower.
Now, granted, there are more than one legit styles of mixing, and mine is not the only way, though it is not unique by any long shot. I *DO* almost always start with a faders up check, but more often than not I'll then move to a default automation setting on all tracks of -3dB (to give me some automation "headroom"), and when adjusting individual relative track levels, I'll be pulling most of them down off of unity gain, leaving the loudest ones at unity. This more often than not winds up the kind of numbers I described in the above paragraph.
Numbers which, BTW, I personally find attractive (even if only on an old-school aesthetic level) because they tend to keep a continuity on the digital side with analog line level. At the same time they almost never compromise usable dynamic range, noise level, or headroom, which to me coincides with good gain structure strategy.
Again, I'm not saying that's the ONLY good or legit mix strategy, but it is *A* good one, and it easily and naturally results in the kind of -10 (+/-) peak numbers about which we're currently examining, without purposely setting any specific target levels.
G.
