ROGER SCHLAFLY

REAL SOFTWARE

MR. SCHLAFLY:  My name is Roger Schlafly, I am self_employed and a software
developer and nobody is paying me to be here.

I have a Ph.D. in mathematics and I have worked for both small and large
companies.  I have one issued patent, and I have been sued for patent
infringement.  I have a couple of patents pending, and I have some pending
patent litigation.

Most of the discussion so far has come from software companies who've argued
about whether software patents are good for the industry, and from patent
lawyers who favor patenting everything.  There have also been a number of
good suggestions for improving the system, but most of these apply equally
to nonsoftware patents.  Instead, I want to focus on some legal and
technical issues related to the scope of software patents.

First of all, I want to say that software patents are not as new as everyone
seems to act that they are.  The first Patent Act explicitly made processes
eligible for patents, and the government's been getting process patents for
200 years. Processes are indistinguishable from algorithms, many process
patents are enforced.

For example, the Polymerase Chain Reaction, PCR invention which won the
Nobel prize in chemistry last year was a process patent.  It was upheld in
court and was sold for $300 million.  It is a recipe for cooking DNA, but
legally it is indistinguishable from an algorithm path.

For an older example, Samuel Morse's patent on the telegraph was upheld by
the U.S. Supreme Court in 1853.  Morse specifically claimed the system of
dots and dashes we now know as Morse code.  This was a software patent.

My next point is that many software patents are not algorithm patents, and
much of the discussion of the legality of software patents focuses on the
patentability of algorithms, but actually many if not most, claim a system
or an apparatus or a machine.  Even if we could reinstate the anti_algorithm
bias of the Benson Decision, it would not eliminate software patents.

Many of these software patents, especially the nonalgorithm patents are
legally indistinguishable from traditional nonsoftware patents.  Many
special purpose electronic circuits and chips are designed using software
techniques, and they often have microcoded programs etched onto chips. 
Nobody is seriously suggesting that patent protection should not be
available for electronic gadgets.

I don't see how you could justify protection for a special_purpose circuit
and deny protection for software on general_purpose computer that performs
the same function.

My next point is that the law does not have to change with new technology. 
Many people here have argued that software is different from other
technology and therefore requires a sui generis protection scheme.  I think
this is a big mistake.

For 200 years copyrights and patents have served to protect intellectual
property without any fundamental change in the law.  I only know of two
cases where sui generis protection scheme was created as a result of
industry pleading that some new technology required it.

One, is plant patents.  As you know, it was eventually decided that ordinary
patents suffice for animal inventions, so the notion that special patents
were needed for plants turned out to be unnecessary.

The other case is The Semiconductor Chip Protection Act, eventually though,
it turned out that ordinary copyright law was sufficient to protect
complicated chip designs, and this special law turned out to be unnecessary
also.

Thus, I think you should reject the notion that the law must change to keep
up with changing technology unless there is clear and compelling evidence to
the contrary.

One advocate here of the sui generis system was Oracle, and to tell you
where they're coming from, I want to quote from their policy on software
patents.  They say, "New developments influential to the software industry
frequently emanate from individuals and small companies that lack
substantial resources".

So from this I suggest that here they are one billion dollar company, what
they want to do is get their innovation for free and not pay for it.

Okay.  My next point is that mathematical algorithms are not distinguishable
from other algorithms.  The Benson decision tried to make this distinction,
most everyone including the Court of Appeals for the Federal Circuit and the
recent Arrhythmia decision, agreed that this was nonsense.  All computer
algorithms are essentially mathematical in character.

I see we have up in the podium the Knuth bible on computer algorithms, and
it's impossible to say that some of these are mathematical and some of them
aren't.  There's just no distinction there.  Fortunately, all these are in
the public domain.

My next point is that many software patents should have failed the novelty
or nonobviousness test.  For example, the Benson patent, which was rejected
by the Supreme Court in 1972 for being unstatutory subject matter, was for
an algorithm for multiplying by 10.  It consists of noticing that the
decimal number 10 is equal to one_zero one_zero in binary notations.  So
that on a binary computer multiplication by 10 can be accomplished by two
shifts and an add.  This trick is obvious to any skilled programer.  If
anyone had tried to publish the trick as a research result, he would be
laughed at.

A lot of patents seem to merely consist of taking some well known method and
putting it on a computer.  These are not novel.  Any method can be put on a
computer, that is inherent in the nature of computers, and people shouldn't
get patents just for the idea of computerizing something.

Now, I want to talk about some of the specific claims that were in the
Notice of Announcement for this hearing.  The last one, Invention G seems to
be in this category of just computerizing something.  It claims an
accounting system implemented on a computer.  It is obvious that any
accounting system can be implemented on a computer.  Accounting systems are
normally not patentable and it shouldn't make any difference that it is
implemented on a computer.  So I don't think that claim should be
patentable.

Inventions A and B are for mathematical algorithms. Patents for these were
thrown out by the Benson decision and reinstated by later decisions. 
Current practice is to look unfavorably towards these, but to allow them as
long as the claim mentioned some hardware.  The hardware mentioned might be
merely some memory chips.  Many older patents have drawn elaborate diagrams
of flip_flops and other electronics hardware, but the Patent Office seems to
be getting more lenient about this all the time.

It seems to me that the hardware requirement is not serving any useful
purpose anymore, it is not required by the law, and does not serve to limit
the claims in any meaningful way.  It might as well be abolished.

Invention claim C_1 and D_1, directly claim a computer listing.  I think
this is a mistake and should not be allowed because it confuses patent and
copyright protection.  Copyright law protects program listings. If someone
else sells a similar program, such a claim will not provide any useful
guidance as to what constitutes an infringement.

On the other hand, copyright law provides a framework for deciding what
plagiarism is.  Software patent claims should distinctly claim the invention
as with other patent claims.

The alternate version of the claims for that invention, Claims C_2 and D_2
more reasonably describe the software invention.  I don't see that it makes
any difference whether such a claim directly refers to a computer program,
or something more tangible such as a disk.  The legal scope of the claim
will be the same.  The traditional Patent Office tangibility requirement is
not accomplishing anything useful here, it should be dropped.

Invention E is a computer data structure, I think the arguments for and
against this are the same as for mathematical algorithms.  Much as people
may not like it, I don't see any basis under current law for rejecting these
claims.

Invention F is a computational diagnostic method performed on a computer. 
If it is a genuinely novel invention, it shouldn't be rejected just because
it uses a computer.  An MRI device might be in this category.  Computer
calculations perform a necessary part of MRI scans, but most of the novelty
lies elsewhere.

A great many electronic devices use microprocessors with software.  But the
claim in Invention F is very similar to the claims in the Meyer and Weissman
patent which was rejected by the Court of Customs and Patent Appeals in 1982
for unstatutory subject matter.  That invention was not really a diagnostic
device, but a computer program for inputting medical test results comparing
them to a list of known (ontcomps) and applying a simple criterion for
determining whether the patient has a problem.

It might be acceptable to grant a patent for some very clever novel way of
implementing such a criterion, but the Patent Office should not give a broad
patent for something like this. It is completely obvious that there are many
criteria to chose from, and that any criterion can be computerized.

In sum, whether anyone likes it or not, it's my opinion that
software_related patents will continue to be issued because that's the law. 
The best we can do, is improve the system by trying to give patents only for
truly novel and nonobvious inventions.

COMMISSIONER LEHMAN:  Thank you very much.

Next I would like to ask Wallace Judd of Mentrix Corporation to come
forward.


Back to The San Jose Index

Forward to Wallace Judd