MR. JORDAN;:  Mr. Commissioner, ladies and gentlemen, my name
is Richard Jordan.  I'm Patent Counsel with Thinking Machines
Corporation.  By way of background, Thinking Machines
Corporation was founded in 1983 to develop, manufacture and
sell massively parallel super computer systems.  Thinking
Machines products are an outgrowth of research undertaken
principally by its chief scientist, Danny Hillis while he was
a graduate student at MIT.
Since Thinking Machines announced its first product, the
Connection Machine, Model CM-1 super computer in 1986 the
company has had excellent revenue growth and revenue from
massively parallel super computers is believed to be the
largest of any company.
However, it should be noted that its revenue is much less than
that of a number of other companies in the computer industry,
both domestic and foreign, including companies in the
traditional super computer field as well as those principally
known for selling computers and more conventional
architectures, many of which I should say are developing
products that are competitive with Thinking Machines.
Over the past several years the computing power of massively
parallel computing technology has been emphasized by a number
of awards relating to Thinking Machines' technology.  Since
1990 the IEEE, the Institute of Electrical and Electronics
Engineers, has given its Gordon Novell Award for computing
speed to several teams, including Thinking Machines employees
for programs processed on a connection machine super computer
and for compiler technology.
The importance of massively parallel computing technology has
also been recognized by articles in journals such as the
Scientific American and newspapers such as the New York Times
and the Wall Street Journal.
A connection machine computer achieves its computing power
through a combination of hardware and software, unlike a
conventional computer which uses one or only a few powerful
data processors on masses of data, the hardware of a
connection machine computer includes tens, hundreds, or even
thousands of microprocessors which operate in parallel on
relatively small amounts of data that are distributed to them.
The individual microprocessors are interconnected by a data
routing network which allow them to share data as necessary
and the software effectively coordinates the operations of the
individual microprocessors and the routing network to achieve
tremendous computing power.
While the hardware is important to the computing power
achieved by a massively parallel computing system, at least as
important as the advancement in software techniques.  Many
advancements have come in the development of parallel
algorithms and computing techniques; the pattern of assignment
of data to processors to minimize processing time; techniques
for rapidly routing data through the routing network; compiler
techniques; the development of high level languages and
compilers to make massively parallel computers easy to use.
Thinking Machines currently has a staff in excess of 500, of
whom approximately one-third are involved in hardware and
software engineering development.  Of these engineers only
about 30 percent are involved in what might traditionally be
referred to as hardware development, while fully 70 percent
are involved in software development.
In addition, a number of other employees actively develop
software in Thinking Machine's large customer service group
developing software techniques specifically for or with
customers.  It is manifestly evident that Thinking Machines
software development effort represents a very significant
portion of its investment in massively parallel computing
technology and Thinking Machines believes that patent
protection provides an important tool to help protect this
Thinking Machines further has an ongoing program to encourage
its developers to publish papers and articles describing new
parallel processing techniques.  This provides information on
new uses for massively parallel computing technology and
techniques and may also help to enhance the professional
standing of its employee authors within their professions.
Published papers represent divulgation of technology for which
Thinking Machines has provided often considerable investment. 
And Thinking Machines believes that patent protection can be
an important tool to protect this investment as well,
particularly in view of the substantial degree of competition
that's developing in the marketplace.
Thinking Machines, unlike some larger companies, does not
require its employee's papers to be cleared, that is
scrutinized to determine whether they describe technology
which the company may wish to protect, before the papers can
be sent out for publication, but it does actively file for
patent protection on technology to be described in the papers.
Thinking Machines recognizes that computer software is also
protected by copyright, but it believes that copyright will
not provide the degree of protection required to protect its
investment.  First, the scope of copyright protection is far
from clear and has been made less clear in recent years in
view of the Second Circuit's opinion in the Computer
Associates case.  It's generally said that copyright protects
the expression of a work and not its idea.
While these words are easy to say, it's very difficult to
apply them in practice.  Furthermore, the application of 17
U.S.C. Section 102(b), which exempts from copyright protection
ideas, processes, methods and so forth regardless of the form
in which they're described, further renders uncertain the
degree of protection provided by copyright.
Much of the value in the program related techniques developed
by Thinking Machines is not in the detailed computer program
code, which is clearly protected by copyright, but in the
algorithms, programming techniques for which copyright
protection is far less clear.  Similar ambiguities are not
present in patent protection.
Furthermore, patent protection is important in view of the
publishing by Thinking Machine's employees, which disclosed
the algorithms and techniques to the world and particularly to
the competition and in view of the fact that copyright
protection may not protect against reverse engineering.
Accordingly, Thinking Machines believes that patent protection
for computer program related inventions is an important tool
to protect its investment.  That being said, Thinking Machines
believes it important that the patent system maximize the
likelihood that the patents issued are valid, that the claims
are directed to new, useful, and nonobvious technology in
accordance with the statutory mandate.
It does no one any service if patents are issued that do not
meet the statutory standard.  While no one can reasonably
expect that any institution run by human beings can be 100
percent perfect, we believe that enhancements can be made to
improve the system.
And I might mention that while we're here discussing patent
issues relating to computer software-related inventions, these
same problems can arise in connection with computer hardware
and indeed in any technology.  The problems may be exacerbated
somewhat in the software area since the PTO for a number of
years was reluctant to consider computer program related
inventions to be statutory subject matter -- a reluctance that
to some extent still continues -- which delayed its
development of a prior art database in this area.
However, there is no industry in which all of the technology
is patented or otherwise published.  Several changes to U.S.
procedure may be appropriate to provide for early publication
of the applications.  This would have two advantages.  First,
it would ensure more timely publication of the technology,
making it available to those working in the industry.  A
publication delay of one to one-and-a-half years after
submission to a paper is not atypical for engineering and
scientific journals, but for patents, particularly in this
subject matter, a much longer delay is more typical.
With developments in computer technology moving as rapidly as
they are, the patent disclosure after such a lengthy delay may
be somewhat less valuable as a source of technical
Second, early publication can also operate as early
notification to others working in the field of the potential
issuance of a patent, allowing them to favor that into their
business decisions.
The potentially lengthy delays to patent issuance under
current practice in the United States means that others
working in the field would not be notified that a patent
application is pending that may cover something they are
developing until the patent actually issues, which can be a
number of years after its original filing date and perhaps
after much time and money has been invested in the potentially
infringing enterprise.
Third, early publication followed by an examination in a more
contemplative environment than would be possible in the
current push for a reduction in the pendency period would
provide a better patent upon issuance.  For example, if a
patent issued on the original application with the
approximately 18 month pendency period as currently suggested
by the PTO, the prior art from foreign patent offices would
most likely not be available for consideration by the U.S.
Having this prior art is particularly important in the
computer area.  It makes for a more complete examination and
a higher degree of confidence in the validity of the patent. 
Typically such prior art is not available until around 18 to
24 months after the priority date and an 18 month pendency
time would mean that the art would not be available until just
around the time the U.S. patent would be issuing.
If the art were deemed material and the application still
pending, the applicant would likely have to file a
continuation application to get it considered, which could
delay issuance to the patent and publication of the technology
for an even greater amount of time and require additional
expenditure by the applicant of another filing fee.
On the other hand, if the U.S. patent had already issued the
only ways to have the art considered would be by reexamine or
reissue, both of which can be costly.  In addition, it puts
too much stress on the Patent Examining Corps which can have
problems with retention of examiners.
These problems can be alleviated by a few relatively simple
changes to the prosecution procedures and the PTO.  First,
they can publish the application 18 months from the priority
date, preferably with a search report so that the applicant
can have it and ideas to the likelihood of being able to get
a patent.
In addition, the public should be brought into the process at
some point, perhaps by way of an opposition proceeding just
before or after issuance.  It would alleviate the secrecy
problem, things going into the Patent Office.  But you have to
make sure that oppositions are conducted and restricted here
as to procedures and time frames, otherwise they can run on
Another way that the system can be improved is by holding
ongoing dialogue such as these hearings to get input from the
Bar and others who have interest in the patent system.
Wearing another hat, I am also Chairman of the Electronic and
Computer Law Committee of the AIPLA and our committee
leadership has for a number of years been meeting with the
group directors and others in the electrical examining groups
to discuss issues of mutual concern.  We expect to hold
another meeting in April, of which we hope to discuss among
other things, some of the issues raised by the notice for
these hearings.
An outgrowth of earlier meetings was a program held last
October in conjunction with the AIPLA's annual meeting that
was extremely well attended by examiners from the Examining
Corps and by members of the AIPLA.
At the program a number of problems and practice issues of
concern to the Examining Corps and to the Bar were discussed
in detail.  Each side, so to speak, learned quite a bit of the
problems and perspectives of the others and the program
received quite good reviews and we hope to have more of them. 
Thank you very much.
COMMISSIONER LEHMAN;:  Thank you very much.  Those were very
helpful comments and I appreciate your be willing to give them
in advance of your prepared time.
MR. JORDAN;:  Thank you very much.
COMMISSIONER LEHMAN;:  I'm going to try to call one more
person in the morning session.  Is Jonathan Band here?
(No audible response.)
COMMISSIONER LEHMAN;:  How about Vern Blanchard?
(No audible response.)
COMMISSIONER LEHMAN;:  Jeffrey Berkowitz?  Mr. Berkowitz,
great.  Mr. Berkowitz is an attorney with Finnegan, Henderson,
Farabow, Garrett & Dunner.
Yesterday we had a meeting with the unions at the PTO and we
big -- we were sitting on the Partnership Council and we had
a big discussion about who should get represented and how many
representatives they should have on this.
It's interesting that we have a disproportionate
representation from Finnegan, Henderson, Farabow, Garrett &
Dunner.  I think this is about the fifth witness that we've
had in the course of these four days of hearings.  It will be
interesting to see this other face of the firm.
Back to the index of speakers for Arlington

Forward to A. Jason Mirabito