Someone should just publish that as public domain, and see what happens if Oracle challenges it. How can they prove that a bunch of bytes sent over the network are meant as prior art and so on.
That said, the fact that one can ONLY find copies of the work by looking to 3rd parties (rather than Oracle) should undercut any arguments they try to make against discussing it for the purpose of criticism to be fair use.