The main problem with socionics is that it attempts to reduce complex, fuzzy phenomena to a simple mathematical model. Psychological phenomena are not like physical phenomena; they don't give way to a nice geometric pattern. Take mental illness as an example. For each mental illness, different people get different symptoms, and further, distinct mental illnesses have been found to be related. This makes it difficult to classify them. The problem is that it is difficult to reduce something like this to discrete categories. A similar problem occurs in socionics. While I don't believe that Augusta's ideas are a work of fiction, I do believe, for example, that not every INTp seeks Se more than Fi. I believe that there is a general type that can be called an INTp, but, because we are dealing with complex, vague phenomena, not every aspect of the type will fit together; these are only trends. Therefore, it seems to me that there is sometimes a problem in typing people. If, for example, you say that someone can't be an ENTp, because they probably have Se PoLR, as opposed to Fi PoLR, you are making a mistake. In order to determine if they are an ENTp, you have to consider their type as a whole. What are their dichotomies? What is their leading function? What appear to be their superego functions? What appear to be their super id functions? Their type is not determined by the answer to any one of these questions, but by the pattern that emerges as a whole; the type that is most consistent with the answers. I've seen this problem come up several times on this site, and I think that it's something worth considering.

Jason