Re: libdts patent issue?
Summary: it looks to me like current US and European law on the
patentability of math, software, and business methods are already
very, very closely aligned. Gripe, if you like, about the USPTO's
ignorance of the prior art in software-intensive fields, and about the
unholy alliance between patent agents with big thesauruses and
corporate assignees who are more interested in obtaining ammunition
for FUD than an accurate statement of the extent of the novelty in
their "invention". But don't blame Congress or the Federal Circuit,
and don't believe everything you read in the funny papers. IANAL,
On 7/17/05, Arnoud Engelfriet <firstname.lastname@example.org> wrote:
> As such. Art. 52(3) EPC. It's actually easier to make sure you
> don't claim math as such than software as such. Just state
> a practical application in the claim.
> Here's a claim that would _not_ be maths as such under European law:
> "A method of encrypting a bitstream A using a key B that is the
> same length as A, comprising computing A XOR B".
> Of course this method is long known, but that's question number two.
That is precisely the distinction applied under current law in the US
as I understand it (IANAL). The latest word on the topic appears to
be AT&T Corp. V. Excel Comm. Inc. [50 USPQ2d 1447, 1452 (Fed. Cir.
1999)] at http://caselaw.lp.findlaw.com/data2/circs/Fed/981338v2.html
. AT&T had patented a method of identifying whether both ends of a
long distance call were in the same network that used simple Boolean
algebra. The district court invalidated the patent on the grounds
that it was not statutory subject matter, for lack of any "physical
limitations" (as required under the older Freeman-Walter-Abele test).
Judge Plager's appellate opinion points out that, subsequent to "In re
Alappat" (a 1994 decision of the Federal Circuit en banc), it applies
a "useful, concrete, tangible result" test instead: "our inquiry here
focuses on whether the mathematical algorithm is applied in a
practical manner to produce a useful result." The circuit court
reversed and remanded, with a reminder that "the ultimate validity of
these claims depends upon their satisfying the other requirements for
patentability such as those set forth in 35 U.S.C. 102, 103, and 112"
-- and sure enough, on remand the patent was invalidated for the right
reasons, anticipation and obviousness.
> I occasionally see courts dealing with the computer program
> exception, but I've never seen caselaw about maths as such.
> Probably because me and my colleagues know that you must
> claim a practical application of the mathematics, and fortunately
> that's usually easy.
And it limits the scope of the monopoly granted to the cited
application(s). US Patent law appears to apply a criterion not
dissimilar from the copyright law "doctrine of merger" -- the Supreme
Court, in Diamond v. Diehr [450 U.S. at 175, 209 USPQ 1 (1981)],
confirmed that "abstract ideas" are excluded from patentability and
hence any "invention" which constitutes an obvious application of an
abstract idea is denied patent protection.
I think it's fair to say that an abstract idea in the form of an
algorithm, "applied" to the making of a "machine" that executes it
(those absurd "register means", "processor means", and "storage means"
of the 80's and 90's), isn't really "applied" yet. It becomes
patentable subject matter when it is applied to a concrete problem
domain in which that algorithm is not an obvious solution and produces
a useful result; and it doesn't really matter whether it can be built
with commodity hardware or has to be realized as an ASIC to be
commercially competitive. Hence it is essentially also true in the US
that software is not patentable "as such", only its application to a
concrete problem domain.
The fact that this distinction is widely misunderstood or
misrepresented (IMHO, IANAL) by commentators, especially those with a
nest to feather or an axe to grind, is not the Federal Circuit's
fault. The USPTO isn't helping much, as they haven't fully understood
this opinion either as far as I am concerned. See discussion of the
continuing relevance of the Koo patent case
http://www.uspto.gov/web/menu/pbmethod/trangmaterials.ppt (Google will
convert it to HTML for you). The "storage in computer-readable
memory" shibboleth from the mid-90's had to do with uncertainty about
whether "concrete", where signals were concerned, meant "must be a
machine instead of a process" (the claim formulas are different in
USPTO guidance), and AFAICT this has also been superseded by AT&T v.
> Try it. I'm able to kill business methods, but I would not
> advise my client to oppose a European patent on the ground that
> it's a computer program as such. Not even if I had Knuth as
> expert witness.
>  See the lowermost example on my webpage
Note that the relevant decision in the US, State Street v. Signature (
http://laws.findlaw.com/fed/961327.html ), would not authorize
"business methods as such" with no technical effect either. Judge
Rich wrote (on behalf of a panel also including Judge Plager):
"Office personnel have had difficulty in properly treating claims
directed to methods of doing business. Claims should not be
categorized as methods of doing business. Instead such claims should
be treated like any other process claims." And like other process
claims, they need to demonstrate a useful, concrete, tangible result
that reliably follows from performing that process.
Once again, business methods also have to meet all the other criteria
for patentability like any other invention: "Whether the patent's
claims are too broad to be patentable is not to be judged under § 101,
but rather under §§ 102, 103 and 112. Assuming the above statement
[Signature's assertion that the patent would foreclose the use of
computers in the entire application area] to be correct, it has
nothing to do with whether what is claimed is statutory subject
matter." I have not yet tracked down the decision on remand, but I
would not be the least bit surprised to find that State Street's
patent was thrown out again on obviousness and/or anticipation
> European law is civil law, and it's the statute that determines
> what is patentable. The European Patent Convention does not say
> that "computer programs" are excluded, but makes a point of
> saying that only "computer programs as such" are excluded.
> Since that's what the law says, that must mean something.
US law also contains a principle of "construction of statutes so as to
give every word meaning", and the Federal Circuit would doubtless
derive, from statutory language denying patentability to "computer
programs as such", similar conclusions about software that it has
(finally) extracted from 35 USC 101's "process, machine, manufacture,
or composition of matter" and the Supreme Court's Gottschalk v. Benson
and Diamond v. Diehr decisions.
It is certainly true that civil law countries do not have the volume
of judge-made law that common law countries do. More importantly, the
practice in Congress and state legislatures of incremental revisions
to statute that either ratify or override major distinctions first
articulated by a judge seems to be almost unique to the US these days.
(AIUI England has gone more and more the "spell it all out afresh
each time you rework the statute" route in recent decades.)
I humbly submit, however, that non-lawyers in civil law countries
would also do well to treat appellate decisions as the primary
literature for the purposes of legal research. Stare decisis or no
stare decisis, appeals courts have a way of commenting incisively on
the lower court's application of statute and other guidance.