Theyre all flawed but theyre fun to learn from.
I think one area where Socionics is limited, or that is, where there's a potential for the understanding to be extended, is in the area of the flexibility of the human mind. That's something that relates to this thread actually: The fact that INTp and ENTp aren't as far from each other as might be predicted in theory.
In theory, an INTp's mental apparatus is in a certain relatively fixed state, where N is pointing in one direction, and T is pointing the other way, so that to act like an ENTp, one would have to rotate these so to speak, which, in theory would be a difficult task; or at best, one's emphasizing functions 7&8, but not living out of one's main functions.
Hence, if an INTp were to become more extraverted, s/he would appear like an ENTj, but never like an ENTp.
This is true in certain circumstances, I think, like when an INTp becomes a lawyer, politician, etc; in these cases, INTp's

comes out very strongly; but...
...Overall, I think there's more of a continuum. Here's one way I look at it; you may not agree, but I think there's something to it:
When you imagine something, it automatically has structure. There is a degree of internal structure and external structure. Roughly, I think we can say that these correspond to

and

(not the underlying motivations, but the ways of thinking, in this case).
And when you imagine something, I think the mix of these kinds of structures can be different...anywhere from almost all internal (

) to almost all external (

)...and everything in between.
[EDIT - Of course, I realize that in Socionics, T isn't just structure; it's evaluative; but it seems to me that the internal/external quality of the N is still inverse to the internal/external quality of the structure; I know this may not make sense to everyone, but try to see it.

]
So, unless someone can show me why there should be such a sharp cut-off, I think there's a lot of gray area between INTp and ENTp.
In fact, with the billions of people around, you probably have people representing every which combination between the types that you could think of. The theory suggests some sort of "cut-off" between the types, but when you look at data on this sort of thing, you see normal distributions in the various scales, and it's very hard to find evidence that people are clustered within types. There's a lot of "in between."