Reply to post:

Hmm. Is "near perfect" actually = "good enough" for "impenetrable" encryption/communication?

I'm also a bit wary of this "probabilistic method" - can you really be sure about the size of the error margin you'll be working with?