PRESENTATION BY BERNARD GALLER
UNIVERSITY OF MICHIGAN SOFTWARE PATENT INSTITUTE
MR. GALLER;:  I'll introduce myself.  I'm Bernie Galler,
Professor of Computer Science at the University of Michigan,
and former president of the ACM.  But I'm here today as the
founder and chairman of the Software Patent Institute in Ann
Arbor, Michigan.  And I'm speaking here as Software Patent
Institute representative.
The history of inventions in the software area is not recorded
well.  There are a few formal journals, such as the Annals of
the History of Computing, and some textbooks.  But the prior
art that is needed by the PTO is not available in many of the
forms that more mature fields support.
For example, in the fields of chemistry or physics, in
addition to a large number of patents available to the PTO,
most researchers' results are published in a relatively few
journals.
This is not the case in the software community.  Not only are
the results and inventions not published in formal journals
most of the time, they usually described if at all, primarily
in informal conference reports or newsletters.  Add to that
the almost complete lack of issued patents before 1981 in this
field, and it is clear why PTO examiners have a difficult time
finding prior art, even when previous work that is relevant is
well-known in the field.
There are some repositories of program code, but it's very
difficult to extract, or abstract, the innovative and
nonobvious algorithms and ideas that are detailed there.
What is needed is not the detailed code, but some level of
description of what is in that code.  Unless the author
carefully documents the developing algorithm, the control flow
and the data structures, it's very difficult to discover these
concepts to understand the underlying process.
It is well known, however, that programmers are usually too
interested in moving on to the next task to take the time to
document the last one.
It isn't difficult to understand why software results are so
often not published in formal journals.  Most of the work in
this emerging field has been done outside academia, since
software is almost always immediately applicable to the
solution of problems that already exist in industry.
Of course, there is theoretical work in computer science and
compute engineering.  But the explosion of computing in our
society has led to a corresponding explosion in software
techniques in advance of the theory.  And in the rush to
exploit these techniques, relatively little effort has been
devoted to disseminating these results and techniques widely.
In fact, even when this kind of information is not regarded as
a trade secret, many companies are not particularly anxious to
have it made widely available.
During the years before 1980, there was much confusion as to
the kind of protection that might be available, if any, for
software inventions.  And there was little incentive for
programmers to try to publish their work.  Much of the
communication that did go on occurred at thematic conferences
and workshops.  The reports of such conferences constitute a
very valuable source of prior art, but they are not readily
available to the PTO.
Thus, the PTO has found it difficult to identify the relevant
sources for prior art, or to collect that prior art into a
usable database for the purpose of evaluating patent
applications.
What are the relevant sources for prior art in the software
area?  I already mentioned our conference and workshop
proceedings from both general and specialized conferences. 
These are usually sponsored by professional societies such as
the ACM and the IEEE, and special interest groups, the sigs,
or societies.  And the sigs publish newsletters also, often
containing nuggets describing new ideas and techniques which
eventually prove to be important prior art.
Universities such as Michigan and UCLA have for many years
offered short courses lasting one or two weeks in which
leading edge research results are presented, disclosing new
ideas, concepts, and techniques.  The notes which are
distributed to attendees contain valuable descriptions of such
work and in time prove to be important prior art publicly
disclosed.
Manuals for commercial systems and applications often contain
important descriptions of the techniques these systems and
applications embody, and are a valuable source of prior art. 
Such sources would not be readily available to PTO examiners
unless the PTO would have the funding to build an extensive
library with appropriate indexing for that purpose.
A number of software vendors publish internal reports and/or
research journals, which are made available to their
customers, and are thus publicly disclosed.  These reports and
journals and other materials used for the education and
training of customers often describe innovative ideas and
techniques which could be used as prior art if they were
available to the PTO examiners.
Government sponsored research is often documented in reports
generated by the principal investigators and published by the
sponsoring government agencies.  While these are public
documents, it's not easy to know where to look for them.  They
often contain the earliest reports of significant research and
applications in the software area.
Another source of material can be found in books published on
various subjects in computer science and computer engineering. 
These include textbooks for the more advanced courses, and
research publications from academic institutions.
It is not always easy to find the kinds of prior art that
examiners need in such books.  But if they were on-line
instead of only in printed form, it would be much easier to
discover which books contain material relevant to a particular
claimed invention.
Finally, corporate defense of disclosure publications can be
important sources of relevant prior art.  A company that wants
to make sure that a competitor does not obtain a patent
covering a process or technique that is essential to its own
business might publish a description of that process or
technique to have it publicly disclosed without taking the
additional step of applying for a patent.  And there are
well-known examples of this.
On the other hand, that company may not be particularly
anxious to advertise its discovery or use of that process or
technique, so the publication would not be very widely
disseminated.  There are also well-known examples of that.
If indeed a patent is later issued for that process or
technique, the company can point to the disclosed art during
litigation, but that is a very late stage in the cycle.
Companies that rely on defense of disclosure should be
encouraged to deposit their published disclosures in a
database available to the PTO so the controversial patent most
likely will not be granted at all.
Well, the Software Patent Institute is a nonprofit institution
dedicated to providing information to the public, to assisting
the PTO and others by providing technical support in the form
of educational and training programs, and to providing access
to information and retrieval sources.
The primary goal of the Software Patent Institute is to
provide the best available information as to prior art in the
software field for utilization by the public and the PTO.
We applaud the efforts by Dr. Dobb's journal of Miller
Friedman publications to make its articles available on CD
ROM.  And the efforts of Ziff/Davis Publications to put a
number of recent computer-related publications on CD ROM, as
well as the efforts by the IEEE and the ACM to make available
abstractive computer science articles.
We also applaud the efforts of those who are working to
identify, collect, and distribute copies of the patents they
consider software-related, especially since many of the
patents that have been identified come from a large number of
PTO classes.  These efforts are valuable contributions to the
overall effort to document the history of software technology,
and to make the results available in online form.
The Software Patent Institute, for its part, is tracking these
efforts carefully so that our collection supplements rather
than duplicates these other efforts.  To track the history of
an exploding industry with rapidly developing technology is a
massive undertaking that will require significant efforts by
a number of organizations.  We are committed to being one of
them.
The Software Patent Institute also provides an educational
resource from which the PTO and the public can obtain an
enhanced understanding of the nature of software, of software
engineering, and of the history of the discipline and its
relationship to the patent process.
Several lectures have already been given to the examiners of
the PTO on aspects of software history and techniques.  And
several more are scheduled during the next few weeks and the
coming months.  We will have a professor from Carnegie Mellon
there next week, and a professor from Michigan there the week
after that, lecturing to the examiners.  And we hope to
continue that.
We plan to offer our first one-day session on related topics
to patent professionals and the general public sometime this
spring.
Although there is a current debate on the overall desirability
of having software patents, the Software Patent Institute has
deliberately taken no position on that question.  We recognize
that the patent system is in place, and working, but that
there is currently a problem regarding software-related
patents.  We are dedicated to helping alleviate that problem
independent of longer-range considerations that must
eventually be resolved.
The Software Patent Institute has asked people throughout the
software industry, government, and academia, to contribute
descriptions of software techniques and processes to the
Software Patent Institute database.  These descriptions form
the content of the SPI database, and have now been made
available for computer-aided searching by the PTO, and by
members of the Software Patent Institute.  Access by the
general public will follow shortly.
The SPI database already contains many examples of each of the
kinds of relevant prior art outlined above, and it is growing
rapidly.
Our recommendation to this panel is to issue a strong
recognition and endorsement of this kind of activity by the
Software Patent Institute and by others, and to encourage the
PTO to take advantage of the services of the Software Patent
Institute as much as possible.
We strongly believe that the PTO can and will do a better job
than it has if it has the right tools and the right
information.
I thank you for being able to talk to you, and I certainly
would answer questions.
COMMISSIONER LEHMAN;:  Thank you very much, Professor Galler.
One of the issues that came up earlier today was the whole
question of the classification system that we have right now,
that it very rapidly gets out of date.  And this makes it very
difficult for examiners even to take advantage of the
information that's already in our patent files.
Obviously you're struggling with that, working with that, as
you try to organize this new database.  Maybe you could expand
on that, about, do we have a problem?  What's the nature of
the problem?  And maybe you have some suggestions about it.
MR. GALLER;:  Well, for the time being we're providing full
text search with whatever words the patent examiners know
about.
What's really needed down the line, though, is a thesaurus
kind of help, which says, if you're looking with this term,
you really ought to be looking for those, also, and here are
some additional suggestions.  Here are some related articles
or entries that you may not have thought about, but they might
be close to what you want.
There are an awful lot of database techniques that are
well-known here which we certainly will start to use once we
have a process that is working and bringing in the revenue
that we need to keep going.
But is this kind of -- well, two things.  One is, the database
service can provide such help.  Here are some suggestions for
what you want to do.
The other thing is, as we give these lectures and other people
give lectures, and the examiners become more
technology-knowledgeable, they themselves will expand their
knowledge of how to search.  What are the relevant terms? 
What are the relevant things they ought to be knowing about?
The classification that the PTO has doesn't help.  You know,
from the computer science point of view, it's not a very good
classification.  But it exists.  And we can hope to help map
it into more coherent, technology-based classifications.  And
we certainly plan to do that.
COMMISSIONER LEHMAN;:  Thank you very much.  We look forward
to cooperating with you and working with you.
MR. GALLER;:  Thank you.
COMMISSIONER GOFFNEY;:  Bernie, Jerry Goldberg, who is the
director of Group 2300, certainly endorses your activity, as
do we.  He wasn't here, wasn't able to get here today, but he
has told me a lot about your work, and it's certainly
appreciated.
MR. GALLER;:  Well, he's been very helpful to us in helping us
understand the problems of the Patent Office, absolutely.
COMMISSIONER GOFFNEY;:  Thank you.
COMMISSIONER LEHMAN;:  Thank you very much.
We're getting there.  Finally, unless Professor Fryer has
arrived -- he hasn't.  I know him personally, and I don't see
him.
Then finally, we're at Gregory Aharonian, who has waited very
patiently for two days now.  He was also in San Jose.

Forward to Robert Greene Sterne