To explain why do I think so, we have to review two basic assumptions of socionics: that there is a certain amount of types and that everybody fits inside one of them.
Let's see the first assumption:
there is a certain amount of types.
The main difference between a computer and a brain (not only human) is that a computer has a finite set of states, while a brain is in essence infinite.
This is because a computer is a static element which simulates dynamism. Although a computer can perform different operations, the truth is that the data buses, the logical and arithmetical units and other parts of a microprocessor are wired the same way always. So if you let a computer run forever, it will eventually reach all possibilities and start running in circles.
A brain, not only the human brain but all of them, has a characteristic that computers doesn't have: it's a truly dynamic element. A brain can do something a computer will never be able to: to process some information, rewire itself and process the same information yet getting different results.
To be explicit, a brain is a perpetual recycling machine. Each time it changes, a new set of possibilities is established and it never runs long enough to deplete it, because it changes again, over and over.
So, can you imagine a dynamic element structured like a static one? I can't. And I can't because it makes no sense. I can only explain such way of thinking by the abuse of statistics, like in many other areas of knowledge. Some people tends to believe that anything which is statistically significant has a direct correlation with reality, while in fact it doesn't.
The trick goes as follows: one can pick up any dichotomy, divide a large group into sections, then apply another dichotomy and further divide it and so. In the end we end up with smaller groups that share certain characteristics and that are ordered in an orthogonal way. So if you ask a member of a group to make a choice, you're likely to get the same response from other members. The degree of agreement will be in direct relationship of the "quality" of the dichotomies you used to build the group.
This explains the interrelationship aspect of socionics, but it proves nothing about the way we process information and it causes an endless (and useless) discussion on why members of a type are similar and different at the same time. Summarized, it goes like:
* The groups are only consistent to the degree of dichotomies applied to them. Outside the dichotomies the characteristics of an individual are in essence unpredictable.
* Given enough dichotomies one ends up with groups of a single individual (which is cycling thinking).