The biggest takeaway for me is how unreliable our subjective sense of "this works" can be without proper controls. Also, the way placebo + regression to the mean can make almost anything seem effective in the moment