I have come to the conclusion that socionics is about 70-75% accurate. People often have clear, well-defined types. There are sometimes exceptions, but I think it holds up quite often. On the Ti side, I think a hard proof might someday exist for the theory. What do you think?