This is my new model:
Model D builds on Socionics Model A and Model B, and to some extent on Harold Grant's model. It consists of two sets of functions; main functions and secondary functions. These are separate systems (of functions), but for the most part it is only the main system that works independently. For example, information frequently goes from ILI's Ni+ directly to Se-, then to Ti+ and finally back to the main system (Fe-). This is why we often don't see Ne- in ILI. But if an ILI extroverts a lot, then he or she occasionally uses the complete secondary system.
The basic structure of Model A, Vladimir Yermak's definitions of +/- functions and Viktor Gulenko's approximate descriptions of +/- functions apply to Model D:
EDIT: The Super-Id block contains 3D and 4D functions.
Model D is different from Model B in three ways:
1) The model does not use information aspects/IM elements. It uses Jung's/Berens' cognitive processes with some improvements: Te is about logical deductions, EDIT: Te is indirectly related to facts, Si is related to comfort, EDIT: Si notices details of objects, EDIT: Ni is about scenario thinking etc. However, it is possible to replace the Jungian cognitive processes with information aspects/IM elements. It would still be a very accurate model.
2) The secondary functions are noticeable almost every day, unlike Bukalov's "Shadow functions" which are noticeable in some special situations. The conscious/mental functions in the secondary system are much more obvious than the unconscious/vital functions in the main system. This corresponds with Irina Eglit's and Vladimir Yermak's research on plus and minus functions.
3) Aleksandr Bukalov claims that ILI's Shadow Ego is NiFe. I claim that ILI's second Ego is NeFi.
Is there any evidence that supports Model D? Yes. For example, ILI can spend months or years on a new theoretical model (Ti-) but has no patience whatsoever for concrete technical problems, like repairing the car (Ti+). Also, ILI oversimplifies any solutions (Ti+) to physical threats (Se-). ILI's Ti+ is obviously very weak. Another example is IEI who is more creative than ILI. This makes no sense if both IEI and ILI have Ne as Ignoring function in Model A. A third example is LII who sometimes has surprisingly high emotional intelligence (Tina Fey).
A key advantage of Model D is that it explains why people of the same type can be so different from each other. This is my new subtype system:
1) Strengthening one function produces a domino effect across the entire functional layout, so all strong (or weak) functions are strengthened and all weak (or strong) functions are weakened.
2) Some subtypes use their secondary functions more than others due to variations in extraversion/introversion.
3) 100 subtypes are needed in order to accurately describe all kinds of personalities. Hence, a very introverted or extroverted person with very strong Ego and Id functions is subtype 1,1. An ambivert with very strong Ego and Id functions is subtype 1,10. Here are some examples: David Keirsey Jr is LII-8,2. Tina Fey is LII-7,8. Bill Gates is LII-2,7. Seth Lloyd is LII-1,4. And Terence Tao is LII-1,1.