In reading through the kerfluffle about Berit Brogaard's post yesterday I realized that some of the discussion was a bit too harsh on the DSM.
First off, it should be clear that the psychological disorders delineated by the DSM do not pick out natural kinds in the way the periodic table does. Psychology is not chemistry. For one, the phenomena are maddeningly vague. And for two, one can only delineate the phenomena to the extent that one has a pretty robust normative Aristotelian account of human flourishing (note that this does not mean Chemistry is non-normative, it's just that the norms don't so clearly require Aristotelianism about chemicals, or at the very least no theory of chemical flourishing).
Often in this work, discrete causes are found and then the ontology of disorders can change in new and surprising ways. For example a huge issue right now in dyslexia research concerns whether there are really two distinct disorders that had been lumped together: one caused by impaired spatial abilities combined with something like ADSD, and another caused by impaired ability to track phonemes.
So the DSM isn't complete ideology. It also consists in a lot of bets made about the value of future research projects. These bets are always based on current research, as a non-negligible percentage of abnormal psychology publications actually conclude with the author writing about how her findings might or might not influence the next DSM. And to the extent that the bets are wrong (as they have been with dyslexia, I think), the very fact that the bet prompted research into causes is something that helps get us closer to the truth.