[tierack]

From your first post (and my real contention):

> Placebos aren't working better because people have more faith in medicine, they're working better because of the Hawthorne effect; people feel like they have someone who cares about them.

The problem isn't that placebos are working better, the problem is that they're working better relative to medications. If it was merely an issue of "someone who cares about [the patients]" then we'd expect that the actual drugs would benefit from the same effect too. As it stands, the problem is that placebos have been working better, while the same drugs have been basically steady. If Prozac couldn't beat placebos now, it tells us something, but doesn't tell us anything about how much better things were back in the old days when when we were participating more in civic organizations and church.

[gruseom]

If Prozac couldn't beat placebos now

If I understand recent findings correctly, it's not that Prozac can't beat placebos now; it never beat placebos and the studies showing this were simply suppressed. (I just posted about this: http://news.ycombinator.com/item?id=784271).

[Alex3917]

>If it was merely an issue of "someone who cares about [the patients]" then we'd expect that the actual drugs would benefit from the same effect too.

Great point, this could completely sink my theory. Two arguments though:

1) If the placebo was less effective than the drug and they both gained the same amount of efficacy, then mathematically the placebo is now more effective relative to the medication.

2) It doesn't seem possible to really debunk my theory without taking a look at how patients are recruited for experimental drug studies, as I can imagine the standard procedure being such that my theory would still hold. but I don't know enough about this to say much offhand.

Anyway I'm not going for a Nobel prize in science, but if I were then this is still what I'd bet on.

edit: I wonder if watching funny movies has become more effective over time. This would be an interesting preliminary test.