I can't speak for Jeffster, but my personal thought is that a lot of people take the theory, and cognitive function order, as a given, without questioning it. Now granted, this is a forum devoted to the theory, so I doubt it's the place to actively raise eyebrows at the theory itself. But based on people I've known in real life, and observation, I've never truly been convinced myself on the validity of all of it - specifically the iron-clad function orders. Then I try to remind myself that mbti is a pretty system, and as such it needs a structured cognitive function order to complete the framework.
It's possible though I misunderstand the theory or am mis-applying it in some way.
Ironclad functional order is dumb.
The functions themselves don't have much use beyond self-examination. It's nigh impossible to tell what's actually going on in someone else's subconscious mental processes.
The real use of MBTI is in its labels and system of categorization of externalized behaviors. When you consider it as a system of four independent variables of behavioral preferences,
without attempting to compare their strengths relative to each other, it works well enough to provide an added degree of understanding/predictability.
This is the way the system was introduced to me in a book called
Type Talk, and it's the way I still use it today. The authors mention the functions, but only in passing and kind of as a supplement, a bit of a "here this might be interesting to think about" afterthought.
So basically, screw functions. Read about what they do so that you can identify the processes in yourself, but when considering the types of others, ask four questions:
1) E vs. I?
2) N vs. S?
3) F vs. T?
4) P vs. J?
There are ample resources on the internet for what each of those means...but remember that these are
independent variables, meaning that if you're an ISTJ, you prefer I over E, S over N, T over F and J over P, but that which preference is stronger than the others is entirely dependent upon the person. Insisting that all ISTJs are best in Si, then 2nd best in Te, etc. is just silly and makes no intuitive sense.
I totally get where Jeffster is coming from. If we're to expect any real use out of MBTI beyond a cute parlor game, we have to consider it purely as an arbitrary categorization system of externalized behaviors.