Blinding gets described as if it deletes the placebo effect. It does not: people who expect to improve still improve. What a good procedure does is give both groups the same expectation, and stop the person scoring the results from knowing which score the hypothesis wants to win. Switch who knows which group is which, and watch how far the measured gap between the two bars sits above the true one.
Switch what is known. The hypothesis is known the whole time.
Who knows which group is which?
Name the procedure