I've thought about it, and the most preposterous premise in socionics is the whole notion of Fi choosing based on 'likes and dislikes.' That notion is just absolutely preposterous. I simply don't think there is any way to choose something unless you 'like it.' Quite simply, there is no difference between 'valuing' something and 'liking it.' This therefore undermines the whole idea of quadra values - as this assumes that certain types 'like Fe' or 'like Si' or 'don't like Se', etc.; all types therefore value what they value because of what they 'like' and what they 'dislike.' It therefore is stupid to assume that only certain types choose based on liking or disliking something as this is present throughout the whole theory.
Wouldn't you agree?
EDIT: Also assume that people who are Fi 'know what other people like.' If the person who created the theory is Fi-PoLR, then she didn't know this at all. But it is a theory entirely based around what people like and what they dislike. Then, if this was her weakest function, how could she surmise what people like (or don't) like? She doesn't know this at all, but the whole theory is based around it. That's another reason why the whole concept should be thrown into the garbage...