Even to know (in the double-blind trials "gold standard" sense) whether a medicine works, you need to diagnose reliably what it is supposed to treat.

But the DSM is full of illnesses that are really a half dozen completely different conditions with the same (or just similar) apparent symptoms, each with its own cause, and no reliable way to distinguish it from the others except that maybe drug X helps one, and drugs Y, Z, W, and V are useless for it. Likewise, the one that Y helps, and the one that responds to Z, and the rest where nothing seems to work. In double-blind trials, V, W, X, Y, and Z are all "shown" to be equally useless.

That does NOT mean V, W, X, Y, and Z don't work. The "gold standard" is as subject to GIGO as everything else. You simply cannot tell anything about what it means without looking at underlying facts.