🚀 go-pugleaf

Let me clarify the structure of patents.  WARNING:  I am not a lawyer,
and I don't want to be one, ever.  The following is based on my
understanding of some conversations I've recently had with a patent
attorney.  I could be wrong about some or all of this.  In fact, before
I posted the article about the #include patent, I went back to him to
check on one key point and discovered I was completely wrong.  So --
take what I say with as much NaCl as you wish...

First, it's necessary to understand the philosophy behind a patent.  A
patent is a grant of a limited-term monopoly in exchange for teaching
the world what you know about your invention.  The *intention* of the
patent system is to encourage you to publish things, rather than
keeping them secret.  (We can debate endlessly whether or not it
actually accomplishes that; thank you, I'd rather not, especially in
sci.crypt.)  Patents can be granted for (more or less) anything that is
(a) new, (b) non-obvious ``to one skilled in the art'', and (c)
useful.

That said, a patent contains two major parts.  The first is, more or
less, a technical paper that describes everything you know about the
invention.  If you omit critical details here, your patent can be
thrown out -- you haven't taught everything important about the
subject, and hence have committed a kind of fraud: you haven't kept
your end of the bargain with the Patent Office.  (In fact, if you come
up with something that's not quite worth patenting -- patenting
something is an expensive process -- you can publish it through the
Patent Office to prevent anyone else from ever trying to patent it.
The advantage of publishing there, as opposed to in some journal, is
that it's much more likely to be found by someone else trying to file a
patent.)

The second major part is a series of claims.  These delimit exactly
what you are saying is new, i.e., what you have invented.  Note that
that is not necessarily obvious from the first part, which of necessity
will include a lot of background material.  The claims are written in
an obscure dialect of legalese, and are very difficult to interpret.
(To be sure, there are arcane requirements in the first part.  For
example, diagrams must be submitted on 8"x11.5" paper.  No, that's not
a typo, at least not on my part -- I speculate that it's actually a
typographical error that's been enshrined into law.  The lawyer was not
amused when I likened that to the hobbits' custom of requiring wills to
be signed by seven witnesses, using in red ink.)

It is quite possible that the RSA patent itself contains a claim
sufficiently broad as to cover Rabin's algorithm.  I haven't checked,
and I won't -- thank you, I've got better things to do than to decode
that much lawyerspeak; that patent has 40 claims!  If Rabin's algorithm
is a non-obvious extension to RSA, it would be eligible for a patent,
too -- but you'd need licenses from both patent holders to use it.
Claims are generally drafted as broadly as possible -- you want to make
sure anyone who does something close to what you've got is stepping on
your patent's toes.  Thus, if you come up with something that involves
squaring a number, you'll talk about squaring in the description, and
maybe even explain why cubing is too expensive -- but the claims will
probably say ``any exponent'', and then you'll have subsequent claims
that say ``the system of claim 1 using an integer exponent'', and ``the
system of claim 2 using the integer exponent 2''.  What you'll build
and sell will rely on using ``2'' -- but if someone finds a nifty way
to use ``3'' while otherwise using your idea, you want to be sure you
can collect royalties.  Btw, if I recall correctly, one of those patents
cited claims the basic concept of public-key cryptography.  If so, and
if that claim is upheld, *any* public-key system would require a license
from RSADSI.

Two more legalistic notes.  First, patents are presumed to be valid.
If you're challenging a patent in court, the burden of proof is on
you.  Second, *knowing infringement* of a patent is much worse; damages
can be doubled or trebled, I believe.  You may or may not like the
public-key patents -- but given that letter, if you go ahead and release
your system, and if you lose in court, you're liable for serious bucks,
because you have been warned.

		--Steve Bellovin