Problems with "predictions" (and explanations) in Socionics
In principle, Socionics can predict how anyone or any groups of people are going to act collectively as a group. But the problem with this approach is that a prediction is not the same as an explanation.
Let me explain what I mean by that. Say that you have two people in a room, and say that you can predict that they're going to conflict, because they're a Conflictor relation and their Fi and Ti are going to conflict or whatever. And that prediction turns out to be correct, they did end up with a conflict and it's all good and well.
But the problem is that it's not an EXPLANATION of why they have conflicted. Let's suppose that the real reason that they have conflicted, is because one person accidentally stepped on another person's feet while they were in the waiting room, and the person never apologized for it, which made the another person angry. Or whatever, it could be any kind of an arbitrary reason. The whole point is that we are supposing that we "know" the truth behind the reason for their conflict, and this conflict could have never been explained by Socionics, as stepping on another's feet or whatever have nothing to do with Fi or Ti or anything that can be explained within the frameworks Socionics.
So Socionics (in principle) can PREDICT that two people are going to conflict. But you can not explain why they have ended up in a conflict, even though the RESULT is the same: They have conflicted. But there are no "details" or the "process" of WHY it happened. I mean if you can predict something but you can not explain it, then it's not going to be much use since it could be just a coincidence, or there could be no correlation. And if you're going to say something like "Fi conflicts with Ti", then again that's just an assumption without an explanation.
You could say, "Well of course you can explain it, since this and that is related to Fi and such and such", but then that would be your OWN explanation that came after the fact. And anything can be "explained" after the fact. It was outside of the framework of Socionics, as it were, and so you arrived at a certain truth OUTSIDE of the system of Socionics (and if you do this, it will no longer be Socionics because it's not relying on functions or whatever).
In the end, in principle, you could predict how anyone is going to act using Socionics. But there is not going to be an explanation of why, other than from the framework within Socionics, so it's going to be pretty pointless. You can not say, "This is why so and so acted in this way", without "adding" extra explanations on top of Socionics, which is obviously not a Socionics explanation, but it was your own, ad-hoc, personal explanation.
And of course, this is even assuming that Socionics CAN successfully predict relationships, which is a pretty dubious claim. And also it is the EXPLANATIONS that allow something to be successfully predicted in the first place (and Socionics is not in the business of offering explanations, it offers a priori assumptions such as "Fi conflicts with Ti" or "LIEs do not get along with SEIs"). And understanding things and coming up with explanations is difficult, because it is the creative process of a rational being, which itself can not be predicted (because then we can make a prediction of future explanations).
We cannot truly predict something without an explanation, and Socionics does not offer proper explanations. In order to predict people's behavior, we need explanations for why people behave the way they do. The explanations are roughly either cognitive or behavioral. So psychology at large is in the realm of offering explanations for people's behavior (if they're "good" psychology). Of course, there could be an entirely new theory of consciousness that could explain people's behavior and cognition in an objective way.