On the one hand, Myers-Briggs is very useful for correlative anecdotes (ie. ENTJs have these problems, INFJs respond these ways, etc.). On the other hand, the test/re-test validity is low because people tend to mistype a lot. Maybe the model just isn't good enough, but the concept is useful in areas.
