What are the most important and least important criteria for a good personality theory?
The foremost ongoing challenge for anyone trying to develop a theory of personality is measurement. In other words, one needs to develop some account that can measure personality traits accurate in the individual and be replicated with identical results by other raters for high inter-reader reliability. Next, the theory must in some way by useful rather than purely descriptive, and should include criteria of verifiability and falsifiability.
Much of the discipline of pyschology, both that grounded in Freudian and associated humanistic theories (those of Jung and Reich, e.g.) as well as much of the psycho-pharmaceutical industry has come under attack in the past decade for being either ineffective or making unproven claims.