🚀 go-pugleaf

Mark Riordan:
-----
   RSA requires the user to select an arbitrary encryption key, e.
   From e, p, and q is computed the corresponding decryption key, d.
   Encryption and decryption are almost identical:
      ciphertext = plaintext^e mod pq
      plaintext = ciphertext^d mod pq
-----

Lee Hollaar: 
   You infringe a patent if you make, use, or sell the patented
   invention in the United States (or outer space).

Mike McNally:
   >If I implement RSA without using it, do I immediately have to
   >start paying royalties?

Lee Hollaar: 
   Depends on what you thing a court might think "making" the
   invention means.  It's possible.


Hmm... since we're talking heavy duty action here, why not consider
the implementation of "exponentiation", "modulo" and "multiplication"
for bignums.  If I write a general purpose language with a decent
bignum package, then the above description of RSA would constitute an
encryption/decryption program, with cosmetic changes:
   before   after
    =        =:
    mod      |   (and reverse arguemnt order)
    pq       p*q

I don't think the courts will be able to effectively clamp down on
people publishing things like (if this is the actual algorithm):
cyphered =: (p*q)|text^encrypt_key

It might even be possible that a compiler could provide a decent
optimization for this sort of code as a fall-out of general
techniques.  And I think I'll take my chances on being sued for
implementing such a language :-)

This still isn't a perfect solution, because you couldn't advertise
that you are actually using the thing, but that isn't a big deal for a
lot of people.

For that matter, people into civil disobedience could implement,
publish and/or use trivial versions of this scheme (as I've done in
this article -- the language I mentioned above already exists, but
current implementations do not support bignums).

Software patents have a fundamental conceptual flaw:  A description is
an implementation.  Why not demonstrate that?

Raul Rockwell