I've noticed the skeptics here like to raise this argument on every article related to psychedelic research. As if they've found a gaping hole that somehow casts "doubts" on the methodology of decades worth of clinical trials.

Unfortunately this line of thinking doesn't hold up to scrutiny, and completely misses the forest for the trees. Mainly because the only implied alternative is to not blind subjects at all! How exactly would that reduce bias or otherwise improve the clinical trial design...?

> I imagine people guess correctly >90% of the time.

Yes in some cases you're right. Here's a paper from last year that does exactly what you asked [0]. They note that "Placebo could be distinguished well from active substance and correctly identified in 96% of the sessions." The placebos were ethanol and mannitol, for LSD and Psilocybin respectively.

The important takeaway is that subjects still had to guess. They didn't know their group assignment with certainty. And that is key to reducing bias.

I'd really like to see someone articulate a coherent alternative to subject blinding in psychedelic research.

[0] https://www.nature.com/articles/s41386-022-01297-2