I think the paragraphs on the Research Domain Criteria are my response to that. The question of the future is going to be how much of the hundreds of thousands of research articles will we throw out because the categories used didn't really make sense, or what can we extract from those that were useful guideposts for directing future research.

So much current research is based on "pure" clinical samples - individuals who have just one diagnosis. We will learn much more, even if the statistical analysis will be more complex, if we include a greater range of individuals, and document all of the conditions they seem to have.