From: smb@research.att.com
Date: Mon, 30 May 94 10:09:10 PDT
To: rishab@dxm.ernet.in
Subject: Re: Does Estonian RSA chip violate patents?
Message-ID: <9405301709.AA13972@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


	 As this is in Estonia, I don't suppose you're bothered about
	 the patents?  As far as I know, RSA/PKP patents are for
	 _algorithms_, not respected outside the US, though patents for
	 RSA/PKP _hardware_ would be respected world wide.

	 I'm not sure how algorithm patents can be applied to hardware
	 -- you may or may not be able to sell this chip in the US
	 _without_ violating patents.  (Though you can freely sell RSA
	 hardware or software outside the US.)

	 You may even be able to apply for a European patent for the
	 hardware, which would then be respected everywhere, except in
	 the US where it may be superceded by the algorithm patents.

There's a lot of confusion about what the RSA patent covers, and what's
prohibited by it.

Rivest, Shamir, and Adleman did not patent an algorithm.  Rather, they
patented a cryptographic communication system based on certain equations.
That means that it doesn't matter if you build it using software or
hardware -- if it's still a system using the mechanisms described by
their patent, it would infringe the patent in the U.S.

Both within and outside the U.S., RSA hardware is patentable to the
extent that the circuitry used is new, useful, and non-obvious.  But
what you're patenting is a circuit to do RSA, not RSA itself, of course.
(The circuit might actually be more general, i.e., a modular exponentiation
chip or some such -- in which case you'd be well advised to patent the
more general concept rather just its utility for RSA.)  That doesn't
mean you could use the circuit to do RSA in the U.S. without a license
from PKP.  A patent is *not* the right to do something; rather, it is
the right to prevent others from doing it.  But if the circuit did not
do only RSA, there's no reason why you couldn't sell it within the U.S.
without worrying about the RSA patent.

Finally, many folks have claimed that non-U.S. law does not permit
``algorithm'' patents.  That may very well be.  However -- there have
been a lot of patents like that that have been issued within Europe;
IDEA and Schnorr's signature algorithm come to mind.  Whether these
have been issued due to some quirks of the Patent Co-operation Treaty,
or whether they could be issued without that, or whether they'll stand
up in a European court, I couldn't say -- but the patents *are* being
issued.

		--Steve Bellovin