Hi Ken
When I was young I just experimented with little foundation in knowledge, and emulated the approach and techniques of other mix engineers. It wasn't until I was older that I reapproached mixing based upon an understanding of what I was doing to sound. In fact, when I reapproached mixing I had a far, far better understanding of what sound was, and what is looked like. It made a world of difference.
I found that training my ears was, of course, very important. But the big difference for me was in visualising sound, and visualising what I was doing to sound, with every tweak and every effect.
Effects and processors can do quite complex things to sound across 3 aspects: amplitude, in the time domain and amplitude in the frequency domain. Use DAW tool spectrum analysis to help with understanding frequency domain, and use a wave editor to understand the time domain.
I found using 4 test wave forms helped with time domain effects and processing, including EQ. I used sine, square, triangle and saw. To help with getting what was going on in the frequency domain I tried both known wave forms at mixed frequencies mixed together and white/pink noise. Each had different benefits. On top of that I learned and understood the maths involved... but that was because I was learning about designing digital and analogue effects! So I don't recommend most people fo this!. I also learned what theoretically Each effect and treatment should do in the digital and analogue domain and the limitations of electronic circuitry.
All that did improve my understanding. It helped me visualise what is going on, whether that is cutting an EQ hole in a pad to allow other instruments to cut through, or applying a chorus effect etc. One of the reasons I love Isotope tools is their visualisation.
My point here (yes there is one) is that anything that improves understanding is good. Experimentation using your ears is necessary, an absolute minimum... trial and error. But you can greatly improve the speed and accuracy if your understanding of what is going on is developed in parallel.... and as part of the experimentation. Simple waves like sine, square, triangle and saw make visual change pretty obvious. Different waves also let you see the effect that quicker transitions can have.
I realise few would go to the lengths I did, but it doesn't mean doing some of what I did wouldn't be very useful. This at least allows you to experiment with more focus, and with the ability to improve your learning.
On mixing itself, times have changed (and with recent development come full circle). When I started mixing there was no automation. Ok perhaps on very high end Neve desks. Mixes had to be rehearsed. Group faders were essential, as were trainee engineers to manage sections of faders. You learned your mix much like playing a musical instrument. It introduced another performance element and level of variation mix to mix. Latest mixing control systems seem to be reintroducing this as a feature. I always enjoyed that, it has to be said.
These days you can control and automate your mix to a fine level if detail. All the more reason to understand your console knobs, faders and switches, and the effects and processors you use from VSTs to console EQ.
Testing is best done using test signals (many consoles can generate them) and by using reference recordings. Reference recordings are essential for getting to know your system, especially when getting to know the effects of amplifier and monitors on a recording. You know how your reference recordings will sound on different systems, so you can work out how your new mix sould sound (ball park) in order to achieve a similar balance on other systems... Moore of an issue when you do your own mastering.
I hope this rambling is of some use! Lol
Cheers
John