does that stand up to experiments pitting a placebo group against a control group that gets no treatment at all?

No, it does not. While Kahneman is a wonderful scientist (psychologist, btw), he has this annoying habit of attempting to fit everything into his models. He's dead wrong on placebo, even though regression to the mean is a real thing.

There is research supporting Kahneman: https://www.ncbi.nlm.nih.gov/pubmed/1280801

So while I don't dispute your link (because it's correct), the placebo effect is normally defined as the effect over and above that seen in the no-treatment group (who don't get a placebo or treatment). Regression to the mean would impact both placebo and no treatment (as well as treatment) groups equally, so cannot be the explanation for the entirity of the placebo effect.

Placebo patients know that they have received a treatment, so that introduces a bias in the reporting of outcomes from the patient, even if the medics are blinded. If you believe you have been treated it is well known that patients report improvements even for illnesses that can be measured not to have improved.

The breaking of blinding and contrary measurements is enough to undermine most claims for placebo.

Yes, I agree. That's what the no-treatment group controls for. And generally, both medics and patients are blinded.

I sometimes wonder about the validity of comparison with the placebo. What if the drug being studied makes the patients feel something, but doesn’t directly help the condition being studied. But that something feeling alerts the patients that they’re getting the real drug and not the placebo, and that awareness causes all the psychological biase we’re worried about?

Kahneman is supported by research. https://www.ncbi.nlm.nih.gov/pubmed/1280801

Good luck finding an ethics board who will let you do that experiment on actually sick people.

"Ah, people with pain. One group will get nothing and the other a placebo."

Pretty sure they do that all the time?

Reminds me of the old joke, where a respected scientist is announcing he has some great cure for the disease of the day and is presenting the impressive results to an audience, when someone pipes up and asks:

'How did this compare to the control group?'

The presenter is indignant and says, "Excuse me? You're asking if I randomly selected half of these poor souls to be deprived of the medicine, just to see what would happen to them?"

'...yes.'

"Of course not! That would have condemned half of them to an avoidable death!"

'...but which half?'

I love this. It's not even a joke really.

so do it with the common cold or something equally untreatable/fairly harmless.

not that I was proposing such a study - I was asking if we have already done them and what the data says if we have.