As for the talk about shadow functions in the OP (it came up in a PM so I'll post it here as well):
I'm not so sure I really believe in shadow functions. We prefer some functions and sure, under stress when our preferred method doesn't work, we are from time to time forced to try the less used ones which we use ineffectively and in a crude manner, but I don't think they're shadowing anything. They're just functions we're less good at using. The "shadow" functions we exhibit vary with the type of stress we're in and can also often just be the the functions we use more frequently applied in a negative/new way. Instead of exhibiting a "shadow type" depending on our MBTI type like the theory seems to suggest, I think we instead make a crude patchwork "type" out of desperation and frustration, glued together from different pieces of other functions. This "type" could in some situations very well happen to be the shadow type, but in other situations it could as easily be something completely different or not really a type at all.
One of the reasons I started doubting the neat setup of the shadow functions is the realization that the letters aren't really a binary thing, like dissonance mentioned above. If the strength of the letters isn't a binary thing, it means the function order can't be a neat system either. How on earth can the shadow function order be as neat and tidy as the theory seems to suggest if the general function order isn't?
It makes a lot more sense to me (and fits much better with my empirical data) that the function order is a jumble of random strengths and weaknesses governed only by a moderate level of probability.