In the realm of quackery, it is sometimes very important to use vague words which don’t actually mean anything to improve your sense of self importance and to convey a sense of competence. Double blind is one.
It postulates that not knowing you’re taking a medication will reduce the impact of the placebo effect, which means that if someone believes something will work, it is far more likely to work. Sometimes, someone can get an improvement merely from taking a sugar pill. In the realm of psychology, someone is likely to view the same phenomena through an optimistic or pessimistic lens and ignore or accept signs which don’t correlate with their viewpoint.
The only thing is if you’re taking medication with very obvious side-effects, and some medication have extremely obvious side-effects like dependence, they will know they’re taking the placebo. It is therefore semantically no different than testing someone who knows they’re not receiving treatment.
This doesn’t go into the factor of testing superficial metrics (or phenomena) and not practical (and far harder to measure ones) ones outside the context of a lab.
“Double blind” doesn’t mean what you say it does: it means that both the evaluator and the patient are unaware of which group the patient was assigned to by a third party. This keeps any placebo effects generated by physician’s responses from influencing the result. As for whether or not the treatment and placebo are directly distinguishable by the patient, if a medicine turned the patient turquoise, no one would set up a placebo-controlled trial until they found a placebo that could generate the same shade. Some treatments have perceptible effects – AZT for AIDS was a notable example – and patients could figure out which med they’d got, and start switching out with their friends, which then ruined the trial. However, in cases of more subtle effects, you have to understand that many people will generate various placebo effects for themselves: for example, a patient may imagine she got the treatment and start to feel nauseous out of anxiety because she’s read it has side effects. But in reality she may have got the control and started generating nausea as a displicebo effect. So the picture here is not so simple, in that there are various placebo spin gyres that can be produced.
Many people think that a placebo is a name that only describes a negative, physical medical treatment, like a sugar pill, but in fact, the concept is much more wide ranging. A placebo or displicebo (negative placebo, also called nocebo) can be any item that: 1) seems to have existence, either physically or as a historical event, and 2) has a spin (epistemic self-fulfilling prophecy) effect on the attitude of the person who perceives it.
A double blind trial is the closest you can come to a test of treatment vs. control that neutralizes all the potential placebo effects involved. Like democracy, it’s not perfect, but no one has come up with a better solution.