So from what you're all saying, it's probably best to record my songs to a master output signal of -6dBfs (to give the Mastering Engineer headroom) and then after mixdown (whilst mastering) you bring this level up to 0db then?
So why does the Mastering Engineer need this headroom, for what purpose? seeing as he's just going to maximize it to 0db anyway?
No, in fact it's not. The only thing you need to ensure is that it does not clip, not even the inter sample peaks may clip. Because if you overdrive it in mixing, you are really doing mastering work at that point and that's the boundary condition between mixing and mastering. That's your only focus, beyond that you should use every tiny bit of headroom you have available (everywhere in the music creation process), so that your mix balance will be as little off as possible. If you leave 6 dBFS of headroom in the mix, that will likely be a really big challenge for the mastering engineer because your converter is likely not top notch, so it means that in his mastering room it will sound very weak and unbalanced to begin with, unless of course the mastering engineer has equally poor gear as you, then both of you are working towards poor final quality. But if he has better gear than you, and he should have, then if I were to receive that material I would send it back to you and ask you to make it use the full dynamic headroom and make it sound very finished at mixing, but not dynamically pushed to its death. So I would prefer it closer to -10 LUFS than -7 LUFS in the chorus, so that I can during mastering increase it to whatever level that particular production can be at. (the sweet spot of that particular production)
In my opinion, masters can with the right gear and skills, be greater sounding all the way up to -6.2 LUFS, even when a few of the sound sources are sampled, after that most mixes start to break down. Most pro mixing engineers out there use outboard brickwall peak limiters during the post mixing process, to ensure the mix can be handed over to the mastering engineer as close to the final sound as possible. This serves the purpose of letting the mix dictate the dynamic characteristics, what it can and should be at least when it is done. It also serves the purpose of allowing the mixing engineer to remove the gap that otherwise might exist between mixing and mastering, that potentially might allow the mix to become negatively altered during the mastering process. Simply put, gaining impacts the mix balance. The mastering engineer then processes those dynamic ideas and fine tunes those at mastering so that they are perfect on all playback volume levels and systems. His work is also to do the final pushing of it to its power and loudness sweet spot, now overdrive is allowed.
His work is not to take the mix apart in order to get the right dynamic ranges on the various sound sources prior to gaining. Many do so, but it should be done in mixing.
Think of the mastering engineer as the provider of a mix insurance service. It's the mastering engineer's main responsibility to determine whether the mix is ready to be mastered or not. Because with his/her gear and skills he/she has within the music creation process the right conditions to do so. He/she can instantly tell how the mix will come across, similarly to a mixing engineer that can tell how a mix will end up based on a particular production and recording. So, when you hand over the content for mastering, he/she might ask you what type of gear you used, why you left so much unused dynamic range in the content and stuff like that. And that's good, that's what mastering engineers are for...