On Tue, 7 Jul 1998 mgraffam@mhv.net wrote:
Indeed.. what we need is for someone to testify to the court about natural and computer language, and maybe some relevent material from information theory.
What we need, IMO, is a well-placed stunt or two. On sci.crypt a while ago I mentioned that if perl-RSA got any smaller, one could make it the *subtitle* of one's next book: "Blah blah blah: what ``/usr/bin/perl blah blah blah ...'' means." The result would be a book, with book-like 1st amendment protections, but which would turn ONLINE library card catalogs into online repositories for strong crypto. Need RSA? Look up the book, cut and paste into a shell. Ta Daaaa! You're an evil terrorist. The point of this is to drive home the fact that while source code is something that *does* (i.e., a device), it's also something that *says*. It's information, even if executable information, and cannot be dispatched as easily as a gun or a bomb. Should the government attempt to declare a book title a munition, it'll fer sure end up in CNN fringe.
Pseudo-code from any computer programming textbook would be helpful in making this point too.
You'd need to prove it easily runnable. What's the tech status of OCR via, say, a minicam? If you could just hold a page of _Applied Crypto_ up to your computer and have it gleen C source (much easier than gleening English), It'll demonstrate the fuzziness of these things (again, IMO) in much starker terms than would just using a scanner. Another project would be to collect all our number theory and abstract algebra books, and type in verbatim the sections on RSA. These usually contain instructions somewhere between pseudocode and conversational English. If someone could for a Masters thesis develop a program capable of reading that stuff... -Caj