I don't want to downplay depression (which I have no experience of), or come across as being flippant, but could the placebo effect in many of these kinds of trials be as simple as "a caring person (professional and/or stranger) interacting with the depressed subject improves their condition?"
This is neither downplaying depression nor flippant - the therapeutic relationship is clearly very important and I don't think anybody denies that. But I suspect a lot of this is about your inherent world-view. Some people do well after purely online CBT, some don't. Some do well in talking therapy, some don't. It seems likely that how your brain perceives the world (and itself) affects which interventions help.
That's the reason why drugs are tested in double-blind, randomized controlled trials.
The placebo effect is a major part of the effect of SSRI, which only improve 1 in 6-8 patients if you subtract the placebo effect [1].
The "number needed to treat" (NNT) mentioned below is the average number of patients that need to recieve the treatment for one of them to be in a better condition than if all of them had recieved a placebo.
