I would agree that the three-letter code approach is an improvement. Where Myers-Briggs were troubled over J/P, the three-letter code never even encounters such an issue. Judging and perceiving properties are implied from which type of function is dominant(e.g. logic vs intuition).
However, beyond that it deviates just as well from Jung's work, so overall I'd definitely say it is the more complex theory, but it may not be an improvement over MBTI. There are all sorts of concepts and theories within Socionics, that also are in need of strong backing. In my personal opinion, though, since Socionics's fundamental basis is closer to Jung, I'd prefer to work from that over MBTI, if not branch off into a new personal typology altogether.