RUSSELL BRAND

REASONINGS SYSTEMS, INC.

MR. BRAND:  Mr. Commissioner, members of the panel, my name is Russell
Brand, I'm a Senior Computer Scientist and Product Manager at Reasonings
Systems in Palo Alto.  I've been a programmer for more than 15 years.  I
speak only for myself and not for my company.

I'm here today to speak against software patents, as most programmers seem
to, and am here primarily to talk a little bit about part of the history
that seem to have been lost. Before I start with my prepared remarks, there
are enough issues that have been raised in the few hours that I've been in
the room that I think are worthy of attention, that I had not considered
carefully before coming here, did not realize they were issues.

The system that my company sells is a hundred times larger than the stack of
paper you see there on the side.  The system that I write reads source code
so the people who own the source code can figure out what it does.  Often
the source code that we read is a factor of twenty larger than that, written
by a team of a hundred people over ten years.  If we could in any manner
figure out what it did easily, if someone could do that, we'd be out of
business.  It's hard, even with all the help of the people there, to know
everything it does.  It would be impractical to find out what patents it
infringes.  If you could give me a good machine description of every
software patent, I don't think, even with my tools, and my tools are the
best in the world by perhaps twenty years, I could go through the software
in an automated fashion, and find out what patents are being violated.

There have been questions as to whether there are distinctions between
numerical and non_numerical algorithms.  I think the late Admiral Grace
Hopper would turn over in her grave to think that there is no difference and
that we couldn't understand the difference.  She brought the industry
forward perhaps thirty years by the realization that computers could work
with characters and could do things that were not fundamentally thought of
as mathematical.  I learned while sitting in the office that a tool that I
wrote in an afternoon about two years ago to help me deal with dyslexia, to
fix some of the spelling as I type it, probably violates two patents.  There
is nothing nonobvious in it.  There was a problem; I spent an hour to solve
it.  Should I stop using it now?  Should I rely on the patent having been
invalid?

There's been talk about changing the rules to narrow the edges a little bit. 
You're dealing with programmers as one of the groups best suited to find
ways around rules.  Working around social rules, working around machine
restrictions, that's part of what makes us programmers.  You're also dealing
with lawyers, who are probably the second best group at working their way
around rules, and I imagine that microtuning and managing the procedures is
not going to help much.  It will buy you six months or nine months, and
someone will find the new bugs, and there are more of them who will be
looking for ways around it than there can possibly be of you trying to fix
up the rules.

In addition to being a full_time programmer, six to ten hours a day, five to
seven days a week, I'm also a law student three nights a week.  I bring a
laptop with me to class; when class gets dull, I work on programming.  I
have an open lawbook next to my terminal while waiting for compiles.  I have
determined that it's going to take me only probably five years part_time to
learn enough about law to speak intelligently on this, and I imagine to
understand what's going on with algorithms would take someone without
scientific training ten or fifteen years.  All of my free time now goes into
understanding legal issues, primarily issues of information privacy,
constitutional issues, but also in the patent issues.

There's been questions about how do we find the prior art.  If in my field I
could get via Internet all the new patents nominally in my field, and could
send back by Internet mail, here are the things that we have done ten years
ago, here are the articles, check it against my databases online that I use
so I know who to cite when I write articles.  It would take me a little
effort; I could do that, I wouldn't need to charge anyone to do it because
it's a small increment over what I'm normally doing, and it's keeping me
aware of the current research, and I imagine specialists in a hundred other
fields could do the same thing.

Part of what the users of my system do when they are studying software to
find out what's good and what's bad, is they introduce defects and see how
many of them are found.  We send our system for testing, we put in ten
defects; if the testers only find eight of those ten defects and they find a
hundred other defects, we can bet there are twenty other defects that
weren't found.

At this point there is at least a wide belief that many of the software
patents should not have been granted.  My statistical study, grabbing
patents at random and reading them, is more pessimistic that anyone else's
prediction in the room. Nineteen out of every twenty I've read are voidable
on at least three grounds.

Perhaps it's time that we start introducing ridiculous patents, like the
(Letvin Kirchoff) current law patent into the system and see how many get
through.  And if more than one percent of them get through, then we should
address it as a quality control problem, as we would address a quality
control problem in any other industry.

To move on to the history, which is the basis of prepared remarks that I'd
like to make, I'm on a number of committees that run conferences, annual
conferences large and small for professional organizations.  I had a very
hard time getting speakers for one of my conferences this year, a state of
the art technical conference.  More than half of the speakers that I
approached said they couldn't speak this year because of patent_related
restrictions placed upon them by their company's corporate counsel.  In
previous years, ten years ago when this series started, no one had any
problems.  People talked about what they did.  This year I lost half of my
best speakers.  It's going to be another two or three years to find out what
they are doing, and so everyone working in that same field isn't going to be
able to build on that research as quickly.

One of the speakers, in order to give a talk, managed to hack his internal
legal system and get a publication out, such that they started their
one_year clock from that date running, and he was able to talk publicly.  He
had to hack his internal legal system in order to make the information
publicly available and allow us to build on his knowledge in the field.

Five groups of my colleagues doing work in cryptographic technique have
moved or are in the process of moving their work outside the United States. 
They say the patent restrictions and the export restrictions in combination
here prevent them from doing development, prevent them from doing marketing,
prevent them from starting a company.  They'll move it to Europe.  By the
time they finish building it, most of the patents they care about will have
expired; they'll bring it back and start selling it.

Two months from now I'll be giving a half_day tutorial at the Computer
Freedom and Privacy Conference sponsored by the ACM.  The tutorial will be
on election fraud.  In it we'll talk about some of the techniques that could
be used to prevent election fraud.  Many of them are cryptographic, and I
believe at this point all of them have restrictions based on patents that
would prevent them from being used, and all of them could be constructed
from information that was in the literature before any thought of patenting
those things came out.

Last year I gave a tutorial at the Computer Freedom and Privacy Conference
on privacy of data about individuals, and we talked about what could be done
cryptographically to better protect the data.  And again, the best
techniques are protected by patents and you can't license these patents to
use in good and strong ways.  The licensing restrictions are not just, we
want so much money, but we want to control the way you use the patents. So
that level of privacy, a level of a fair voting system, a level that will
allow people to speak in an anonymous, safe way, and to prove who they are
is held hostage to a patent system that will keep us from entering the next
level of participatory democracy, hold it up at least another three years,
perhaps another ten.

Lastly, I'd like to talk about what I see as the coming age of defensive
patent portfolios.  At this point companies get defensive patent portfolios
so that they can force other people to cross_license to them.  Individual
programmers like myself, I don't have such a portfolio.  I'll need to join
someone who has it so that I can cross_license everything I need so that I
can publish.  If I have an individual patent, it won't do me any good
because I can't build anything without that cross_licensing.  So the patent
will afford me as an individual developer no protection, but afford the
large companies, the companies IBM, AT&T, HP, with giant defensive
portfolios the ability to control the new technologies that come out,
whether they've invented it or not.

In perhaps a related issue, people in my area tend to think of patents and
the software look/feel copyrights at the same time.  We look at the history
of the look/feel copyright. It was validated by the court to protect video
games, a video game named "Scramble," that I enjoyed playing when it first
came out.  And it's been extended and extended.  If we look in the same
manner, the first software patent that was granted, Diamond vs. Diehr, it
was a computer system, part of process control. The idea was that a
statutory bar on numerical algorithms would not prevent it from being part
of a combination patent.  We have gone from computers stopping you from
getting patents, to the computer part being okay, to now anything once you
put a computer in it, it's a form.  Well, you couldn't copyright a form,
it's on a screen; now I can get a look/feel copyright.  It's an equation;
you can't patent an equation, it's part of a computer program with no
physical relation to the world.  The rules say you can't patent it, but 1400
such patents were granted.  The rules, as they were written, would provide a
valuable service; the rules, as they are executed, especially with the giant
defensive patent portfolios, do a disservice to developers and to the
American public as a whole.

Thank you.

COMMISSIONER LEHMAN:  Thank you very much for sharing that with us, Mr.
Brand.

Next, Willis Higgins, with the law firm of Cooley, Godward, Castro,
Huddleson & Tatum.


Back to The San Jose Index

Forward to Willis Higgins