I want people to stop using the Myers-Briggs test, because it is not a useful decision-making tool ("bogus" for short). It generates spuriously precise results that garner false confidence, and probably crowds out other, more valid instruments.
Your left- / right-handed analogy is not appropriate. The population is strongly clustered in terms of hand-dominance: most people clearly favour one hand over another. For the kinds of personality traits the MTBI measures, the distribution is roughly bell-shaped: most people cluster towards the middle.
This bell-shape means that a simple binary classification is just about the worst way to do things. Take the I/E axis. Most psychologists will tell you that there is an intro- / extraversion continuum, and only a few people are found at the extremes. A statistician will add that, because of the wide error bands (which we can measure by checking the correlation of the results from, say, the even-numbered with the odd-numbered questions) the MTBI can only reliably identify the extreme types. For the majority of people who are close to the I/E borderline, the results of the test are basically random.
So if my MTBI tells me that I'm an "I", all I really know is that I'm not an extreme "E". That isn't very helpful.
You mentioned having the larger dataset at our disposal. With the larger dataset, we don't need the MTBI. Without it, the MTBI tells us very little. So what good is it?