Nonsense. You can show no example of random noise ever creating any useful program, much less any software of even "pedestrian" complexity.
So you refute a theorem of mathematics (one that always seemed rather obvious and self-evident to me)? Not only can a random noise source produce a "useful" program, but I can calculate approximately how long it will take to generate any program of a specific length given a generation rate. If you are going to discard mathematics and hijack its terms when it suits you, then we aren't having a real discussion.