I think what Gary is talking about are standard techniques. A track should translate well all its elements in mono. The stereo information is generally regarded that carries "effects". (ambience, polishing tonal color etc). This happens for many reasons but mainly because someone likes the song to have a great amount of its energy be carried by both speakers. This is somewhat of a rule for the whole track and strict for the low frequencies. Also mixing is all about contrasts and if everything in a mix is stereo then nothing really is. A sound can create the sense of space by positioning in contrast to another sound that is positioned differently. And the greatest extend of this arc is the contrast between a mono sound and a stereo one and this is what is mainly used because at the same time it fulfils all other neccecities of mono playback like radio mono compatibility, more constant loudness curve, better articulation of audio images etc...
From another point of view:
pan is just a tool on a mixer channel. It allows to route and crossfade a sound to 2 destinations
In my way of working i use custom mixers that are simple adders. They do not have pan control.
If I want to pan something I will feed part of the signal to another empty channel. But I won't stop there. I can feed some filtered part of a sound to the right lets say and then I can take a filtered version of the sum and feed it reversed the left to the right.
If I want to make a drum sound I can mix different samples to the left and the right and then send part of the sounds to the opposite channels... In general the possibilities are many... but in general someone should have enough experience with audio so to have a mental picture of what is mono and what is stereo in a mix. I have programmed a phase analyzer that displays lissagous curves and I have spend a lot of time with it. The process involves if I remeber correctly by heart taking the sine of the L-R and the sine of the L+R and displaying it as x-y coordinates... Now I don't have to use it anymore but I use it often because I like to watch the damn graphics. Also m-s techniques are powerful but not nessecary, also just a tool. Anything can be done with filters and basic effects/dynamics. Strict digital ITB is harder but possible for some genres with a lot of experience. The main problem with only digital is that it is complete abstract and you can paint just nonsense. Poor DACS replicate with accuracy this nonsense... sounds bad. Hybrid systems are better than what people used at the old days because beside analog processing there is also digital which is powerful. Also analog offers a repair service for nonsense created by digital means, they take them back to reality by forcing electrons to take the strange path. Its a magical eraser of ugliness.
Strict analog is great for personal fun and "wow even fck noise sounds interesting", or "i can't make it sound bad" moments.
But you can't sell this experiense it is kind of personal and gets a bit aphasic in moments...
But I still remember the day that I soldered my first sine oscillator, a pathetic diode lowpass self resonating, with uncontrollable distortion because of feedback instability and I hit it with a cute 555 D-envelope triggered by a simple RC binary counter and connected straigth to one of my old 5 inch monitors... this sounded good... and yes it was real. The whole circuit cost around 3euro... anyway we are in the digital era and kind of digital zealots. But how much difficult...
https://en.m.wikipedia.org/wiki/Lissajous_curve