msandford is saying that you can only run an experiment like this and get good results if the treatment is administered entirely by a machine. PhasmaFelis is saying that as long as both groups get similar amounts of professional attention then you can still get good results. I'm saying that to make sure you get equal levels of professional attention you need to blind the professionals to whether the machine is enabled.

(I agree that this is what the "double" in "double blind" means, but that's not what we were talking about.)