The problems pointed out in this article -- which are in some degree real even if heavily editorialized -- must be understood in light of a wider context which includes:
0) The very poor standards of validity and replicability in most fields of experimental psych-whatever. The Popperian method ain't what it used to be.
1) The Viennese Waltz of bad incentives throughout the psychiatry research pipeline. Most notably, drug companies are incentivized for fraud and deceit by being rewarded for closely-guarded research, hiding negative studies and obtaining patent extensions for new applications of existing drugs.
2) Somewhat related to (1): the unfortunate drug safety model prevalent in all relevant nations in which drugs must be proven to work for some pathology. Besides incentivizing recklessness in industry-led research, this means viable off-label usages go "underground" (topiramate for dissociative disorders is a typical case where FDA-worthy testing might be impossible). It might be better if the FDA and its counterparts instead regulated for drug _safety_ in humans and let the academic and clinical communities evolve their professional consensuses.
3) Much related to (0): the effective failure of "talking cures" in outcompeting drugs (specially "dirty drugs" like anticonvulsants, neuroleptics and non-SSRI antidepressants) for "simple" depression, bereavement and related mental health issues of which one could whip up a "social critique" of drug use and so forth. "Mad in America" is obviously not willing to make a fuss of things like bipolar where people will take their lithium for decades and bear the side-effects because the alternative is to go actually mad-crazy.
Did I miss something?