We have all seen typology systems like the Myers-Briggs fall to shambles because of the way it's employed on the internet. Systems, when constructed, begin as wholes, and if the veracity of the systems is to be kept, definitions should exist to allow the systems to exist as wholes. Only with that set of anchoring definitions can systems be employed consistently and make the same predictions (hopefully to similar degrees of accuracy) in different contexts. The main problem that you see with "MBTI" variants online is that people are really disguising their own idiosyncratic variants of these systems as The System. Even when showcasing typings within their own systems to others, the pretence is that these typings are done under the "real MBTI" or the real system in general.
Online Socionics communities that I have attended tend to have a similar problem these days, you see a similar splintering occur. People are employing overcomplicated, impractical and completely flawed (VI best falls under this category) metrics to type without sets of definitions against which typing can occur. In order to stop Socionics from regressing to systems like these "MBTI" variants, it is best to establish a baseline whereby which typings occur. Since the theory is fundamentally describing information metabolism (through IMEs) and intertype relations (through interactions b/t these IMEs), I suggest we stick strictly to Model A. Though anecdotes and personal methods are valuable for typing, ultimately, if they don't work within the system, the typings are not within the system. And this causes numerous problems, including erroneous and idiosyncratic function->behavior mappings (a direct product of this kind of typing).
I myself am an LIE-ENTj. Thought I'd introduce myself because I am new here.