Socionics has, in our context, been reduced from a personality theory to something on the level of a philosophical idealization of different ways that the brain can function. "My friend is a scientist, and loves to do empirical research and crunch numbers," says one person. A socionist responds "That has nothing to do with type. This person may be an SEI, for all we know."
My problem with socionics is that it isn't tied to anything. It doesn't predict ANYTHING if not behavior. I think it's been reduced, in our case, to VI and some very, very specific and, IMO, minor details about a person's personality that, in reality, have a minimal effect on how they interact with people. What exactly IS it that we are predicting if it's not sets of behavioral characteristics? It seems to be that the actual behaviors that are, indeed, connected to type are so narrow in their spectrum that there is little, if any, predictive power in Socionics any more. So what ARE we measuring?