To avoid reinventing the wheel, below is some existing quality material on socionics. The quality of socionics information online is uneven, so here are some links that are of high quality and which have contributed to my own understanding. I don't necessarily agree with all of the information provided but it is overall very good.
Socionics as a Potential Scientific Theory The main issue here is how to make type diagnosis scientific, i.e. replicable. This can be done either 1) using biological methods (such as brain scans or blood tests), or 2) using clinical methods, i.e. standardized questionnaires. We can then correlate either method with measures of stress as mentioned in the article, or with other clinical measures. However, the latter method seems to me a lot of work for very little reward (other than recognition by the scientific community which may appeal to some). But there is another way to make socionics "real" or agreed upon: using mathematics—self-evident axioms that the rest of the theory can be derived from. In this light we can still keep the focus on information metabolism which is the foundation. (The interaction between elements does not even necessarily occur between two different types; it is also a process that happens within the mind, or abstractly, as a consequence of making decisions.)
The Socionics Community I'm happy to say that the English-speaking community (however small it may be) has definitely made real progress, with people proposing original hypotheses and definitions. Real synergy has been a somewhat more elusive goal.
The Socion, part 2 Here Augusta explains some of the fundamental concepts of socionics. Many socionists would do well to emulate her approach: "There isn't a drop here of 'pure theory' that doesn't come out of our observations." Her descriptions of IM elements contain minor inaccuracies but they remain among the best to this day.
The Dual Nature of Man, commentary Augusta's seminal work. The full translation is below, but the initial part is the most important. (Augusta devotes considerable space to elaborating type dichotomy traits and behavioral duality descriptions, neither of which are conceptually central to socionics.)
The Dual Nature of Man, machine translated a rough translation of the whole work
Bukalov and Karpenko interview about the history and development of socionics
Two Important Steps to Make Time has shown the functional approach to be the correct one. IM elements are the foundation of the model, with type dichotomies being a derivative phenomenon.
Filatova portraits (I do NOT endorse using visual identification to type people, except as described by Rick DeLong at the above link. However, these portraits are interesting nonetheless.)
Model A: Blocks and Functions of the Socionic Model of the Psyche A more detailed treatment of functions and their categories. In my opinion, some of these dichotomy descriptions are inaccurate and need to be reformulated, if they are meaningful at all.
My Personal Typology some other important factors to consider when analyzing personality
Socionics Model and Associations While every experienced socionist draws on comparisons to type people, argument-by-comparison is not a good argument while discussing typings with other people! You must consider the possibility than any of your past typings may be wrong, or at least you cannot expect other people to accept them at face value. This is why model-based arguments are needed (and formulating them will refine your own understanding, to the point where the model and your own experience converge).
The Reinin Dichotomies are Dead I actually do think there is potential in certain Reinin dichotomies (in particular the ones that correspond to groups of relationships, such as Process/Result, Positivist/Negativist, and Static/Dynamic) — but they 1) require new interpretations and 2) should be considered as a byproduct of the functional ordering. The standalone traits as classically described are not very useful for typing and in fact can often be positively misleading. For example, I would not rule out someone being a Declaring type just because they ask a lot of questions.