[aufreak3]

> Instead, if you could eliminate a variable from the system by preselecting for people for whom the variable doesn't apply, it would reduce the time to experiment

... if the goal is to reduce the time to prove that your drugs work. That goal is different from making drugs that work.

For an analogy - in order to prove that I'm smart, I must compete with those others who're known to be smart, not with those who've (by some agreed on measure) demonstrated a lack of the required aptitude.

Even this article keeps emphasizing throughout that Kaptchuk is at Harvard, he got praise from scholars, etc.

So the real proof of a drug's worthiness ought probably to be that it works over and above the placebo effect shown for those who respond very well to placebos. That would provide more data to help evaluate the risk of a drug to a patient, given placebo predisposition can indeed be measured. That would be important particularly if the drug has significant unwanted side effects.

edit: And fwiw, preselecting for those who would respond well to placebos would also reduce the time to experiment .. but, I think, it would do so in the right way.