So we hook up ChatGPT to procedurally generate medicine sounding names to sugar pills, and then figure out which name gives the best placebo effect in patients? Then we just rotate some of the better ones in patients to maximize the placebo effect, when they say the effect wears off they get a new name on the pill they take.

Yep. And given a significant problem and a placebo, would you rather:

1) Not use the placebo, because even though it seems to work for you, it's been scientifically shown in trials to do nothing

2) Use the placebo, because it seems to work; somehow get good results

A huge number of people will choose #1. They are extremely close to cargo cult without even realizing it.

I'd say that if there's a pos/neg polarization to cargo cult, they are on the negative side, looking for things to reject in culty ways. But whether you're consuming product or rejecting it, the same shaky mindsets apply.

> 1) Not use the placebo, because even though it seems to work for you, it's been scientifically shown in trials to do nothing

This is the opposite of a placebo. A placebo is a treatment that has been scientifically shown in trials to do something even though it shouldn't. For example, taking a pill or drinking a cup of "medicine" might make your headache feel better even though the pill is just sugar and the medicine is just colored water, and just eating a spoonful of sugar or drinking the same amount of water doesn't have the same effect if you don't believe it's medicine.

We use placebos as the comparison in clinical trials precisely because placebos do something, and we want to show that taking the pill with the actual medicine does more than just taking a sugar pill does. Because of it doesn't--might as well just take the sugar pill.

Placebos do something, and if that something is beneficial then by all means take them. If I can take a sugar pill that makes me feel better I don't really care whether it's psychosomatic or pharmacological. The FDA cares, because they want to encourage real treatments and discourage quackery.

That's interesting, I see the term has multiple definitions, to include contradictory definitions around the tested value of the placebo itself.

It reminds me of some other terms in my own areas of specialty, and the meaning they "properly" carry, vs. the meaning they generally carry when used by the public.