It’s impossible to mix by numbers. Reading your message and taking it to a ‘logical’ (ish) conclusion, one could hypothetically mix by turning off the speakers, setting the numbers so they read the right values and calling it finished. Obviously this is daft – it’s way way way way way way (you get the idea) more complex than that. Which, because the auditory system takes care of all this without you having to think about it, actually makes it way more simple; ‘make it sound right’. In today’s internet-led, analysis-centric world however, everyone wants to know what the right numbers are, asking if I can tell them what numbers they should be working with; it just doesn’t work like that.Tied into the above is that the numbers on your meters (assuming you are talking about the peak readings and not the fader values) are only peak readings. Peak readings have only a small actual relation to loudness perception. While it’s true to say that snap/dynamics/impact (whatever people are calling it this week) does play a role in loudness perception, it’s only a small part of the equation.
Mixing by relying mainly on peak readings is a bit like flying a plane but only looking at the altimeter. You might know what height you’re at, but it tells you nothing at all about where you are in the world, what direction you’re going, etc. You might well have zero above sea level on your altimeter, but if you’ve landed in the middle of the Pacific then you’re bollocksed.
Not a perfect analogy, I’ll have to think of a better one… but you get the idea. It’s disregarding a whole load of equally- or more-important things, that all interact and have a bearing on the final result.