Nozick's experience machine is usually described as able to bring about any desired experience. If it can't do that, then it's not a suitable object for the thought experiments Nozick and others build upon it.
Surely our friends Russell and Gödel would point out a weakness in the definition, as outlined by their friend Richards. Suppose that I wish to have the experience of becoming convinced, via formal proof, that the experience machine is physically impossible. Such a proof would be a syntactic object. If the proof exists, then the machine can't exist; if the proof doesn't exist, then the machine can't feed it to me.
I doubt that this is an original line of thought; who has studied this before, and what did they conclude?