1) You need something to compute the outcome of the symbols in order to test the accuracy of those symbols. I am not sure how to experimentally test your point without doing so. Perhaps that is the crux of the issue, we can't test symbols without tying them to a "physical medium".
2) We can introduce more communication layers as you desire and still keep in the spirit of the experiment. I opted for the strip down version due to simplicity of describing the experiment and the similarity it has to the turing test.
Comments
1) You need something to compute the outcome of the symbols in order to test the accuracy of those symbols. I am not sure how to experimentally test your point without doing so. Perhaps that is the crux of the issue, we can't test symbols without tying them to a "physical medium".
2) We can introduce more communication layers as you desire and still keep in the spirit of the experiment. I opted for the strip down version due to simplicity of describing the experiment and the similarity it has to the turing test.