MR. HOFSTADER;:  Thank you for giving me the opportunity to
testify before you today.  The League for Programming Freedom
is an organization of software developers opposed to software
patents and copyrights on user interfaces.
I would like to use this opportunity to clarify some of the
issues that were raised in an earlier round of hearings in San
Jose.  To evaluate the numerous conflicting arguments that
have been made, we must organize them within a systematic
framework.  Since the patent system is an economic system,
economics is the best framework.
What questions need to be answered?  What issues should be
confronted prior to determining whether software patents
should be granted?
The goal of the patent system is to provide science in the
useful arts.  Whether software should be patentable is
therefore a question of whether patents promote innovation and
progress in the software industry in the computer sciences. 
The economic interpretation of this question is whether
granting patents on software benefits the economy by making
the software industry more efficient.
The League for Programming Freedom asks:  Does the transfer of
economic resources, which software patents represent,
constitute a transfer whereby the resources are going to be
employed more productively?
As an example of how the patent system is dependent on
economic factors that will vary from one industry to another,
I will mention just one factor, the overall size of an
Let's imagine there are 5,000 people employed by the
candlemaking industry in the U.S. and that it has been
determined based on sound economic principles that the optimal
life for a patent in the candlemaking industry is 20 years. 
Suppose the demand for candles is twice what it actually is. 
The candle making industry would be almost twice its earlier
size, employing close to 10,000 people.
Under a set of economic assumptions reasonable for the
candlemaking or software industry, economics would then
dictate a cut in the length of patents for the candlemaking
industry.  Cutting the length of patents by one-half would
yield roughly the same incentive to invent, and thus the same
rate of progress that existed earlier.
Alternatively, we might consider cutting the length of patents
by one quarter.  In doing so, we're sending a signal to the
candlemaking industry regarding the increase net economic
value of improvements in the candlemaking process.
This signal, however, has to be effectively traded off against
the increased lack of competition.  When the size of an
industry increases, the optimal lifetime for patents needs to
be shortened.  Without knowing various factors relating to the
inventive process in the candlemaking industry, the new length
for patents is a matter of debate.
It isn't fair to directly compare the software industry to the
candlemaking industry.  The software is much larger, and it is
also much broader.  From the example of the candlemaking
industry, it should be possible to understand how the
traditional 17-year patent grant may in some industries
conceivably hurt progress by stifling competition more than it
helps progress by encouraging innovation.
The software industry employs some 6 million people.  A
significant fraction of them develop software.  More people
are probably engaged in the software development than in all
other branches of engineering combined.  As a result, in the
software industry reinvention has become commonplace, and
software patents seriously harm the competition.
In eliminating software patents, is it going to be possible to
legislatively define software?  This issue was raised
frequently at the San Jose hearings.  It is surprising that
such an argument can be to justify risking the future
efficiency of a $50 billion a year industry.
Since this argument is apparently one of the key arguments in
favor of the continued granting of software patents, the
League decided to subject it to intense scrutiny.
It is true that many things in this world form part of a
continuum.  Nonetheless, we are able to legislatively
differentiate between them.  The post office is able to
distinguish between a letter and a letter packet.  The FDA is
able to distinguish between a cheese spread and a
cheese-flavored spread.  There is no way to draw a perfect
line between drunk and sober, but the law does draw a line,
and it works.
On a larger scale, the IRS classifies capital goods into many
different categories, to determine depreciation rates, while
the Customs Service is able to classify things to apply
duties.  Considerable financial incentives exist to try to
circumvent these classification systems, yet they work.  There
is little problem with them being circumvented, or with their
complexities imposing great financial burdens.  The legal
system effectively handles disputes over occasional borderline
A legislative definition of software need not embody absolute
truth.  It need only work effectively and efficiently. 
Searching for absolute truth makes no more sense than
determining the exact definition the IRS should use for wood
pulping machinery.
The definition the League proposes is, "Software is composed
of an ideal infallible mathematical component whose outputs
are ineffective by the components they feed into."
I'm confident that the PTO and the courts would be able to
readily distinguish between software and hardware using this
definition.  The PTO is already skilled at administering a
classification system that deals with far more subtle
To show that it is possible to legislatively define software
patents, the League performed an ambitious experiment.  The
League examined 2,000 patents issued during a one-week period. 
We tediously analyzed the details of every software-related
patent granted in that week.  We found little difficulty
existed in identifying software-related patents.
The League then took each software-related patent and analyzed
its claim according to a number of different criteria.  These
criteria were chosen on the basis that their presence could be
used as a part of a test to identify software patents that
should not be granted.  The results of this research clearly
showed us that it would be relatively simple to legislatively
define and identify software patents.
This is not surprising, given that legislation already exists,
that it is able to successfully identify far more nebulous
concepts than the difference between software and hardware.
At the San Jose hearings, Tom Cronin of Taligent forcefully
suggested start up companies require software to attract
venture capital.  He described Taligent as a recent start up
that has succeeded in attracting a large amount of venture
capital, and for whom software patents were considered as
He failed to mention Taligent was an IBM-Apple joint venture
staffed by transferring surplus personnel from these two
companies.  Taligent is quite unique when compared to most
other startups.
The numerous two-kids-in-a-garage stories demonstrate that
successful software ventures require very little capital.  It
isn't necessary to attract large amounts of capital to produce
software, or at least it was not necessary.  Defending against
patent threats may increase this expense.
All the software companies spawned by the micro-computer
revolution gathered sufficient starting capital without any
software patents.  Microsoft, Oralent, Novelle, Adobe,
Systematic, Oracle, and WordPerfect are just a few examples.
The final prepared remark I have deals with why copyright is
the most suitable form of intellectual property protection for
the software industry.  Patents are used in other industries
to prevent companies from using, but not paying for, the
results of their rivals' research and development.  Permitting
this would be a serious disincentive against R&D investment.
Unlike every other industry subject to patents, the software
industry is unique in that its products are also subject to
copyrights.  Copyrights ensure that to be commercially
successful a company choosing to follow another must spend as
much to develop program as the original firm.  Indeed, the
history of spreadsheets, word processors, and virtually every
other software product suggests that it is actually more
expensive to follow than to lead.
A product that seeks to displace the market leader can only do
so by incorporating new features, thereby making it more
expensive to develop the original product.
Copyright is effective because it protects precisely the
product that has been developed.  It prevents other companies
from benefiting by copying your products, while at the same
time permitting them to reap the full benefits of anything
they develop.
Copyright is efficient because it enables firms to compete on
the basis of rival implementations.  This competition is vital
for the efficient allocation of economic resources.  The
traditional literal aspects of copyright doctrine is also
efficient because it has negligible administrative overhead
and presents no uncertainties.  A small start up has the
knowledge that they control what they create.
Given that copyright law effectively and efficiently achieves
the economic aims of the patent system, there is simply no
need for software patents.
This concludes the League for Programming Freedom's response
to issues raised at the San Jose hearings.  I would be happy
to take any questions you might have.
COMMISSIONER LEHMAN;:  First, are you a computer programmer
yourself, or are you just delivering Christian's --
MR. HOFSTADER;:  I'm not a programmer.  I've worked for a
high-tech firm in their legal department, though.
COMMISSIONER LEHMAN;:  So you are a lawyer?
MR. HOFSTADER;:  I'm not a lawyer, no.  I'm not a programmer,
though, either.
COMMISSIONER LEHMAN;:  You mentioned that the League had done
an analysis of 2,000 computer program patents.  Do you have
that analysis available that you could share with us?
MR. HOFSTADER;:  I don't have it with me right now.
COMMISSIONER LEHMAN;:  Does Christian have it?
MR. HOFSTADER;:  Yes.  What's happening and how I'm here right
now is that they were stuck in Boston during the snowstorm. 
So the speech got faxed to me.  The other materials are being
sent Federal Express.
COMMISSIONER LEHMAN;:  I'd be interested in following up on
that.  We have to proceed on the basis of facts and take a
look at these analyses.  We might have some of our people --
Mike Fleming and others -- might take a look at it.  And we
might even want to have some further dialogue with you, or
with the League, about that because it gets really into the
question of our prior art database and what's going on here.
So I think, rather than just sort of have a statement about
the results of this analysis, we'd really like to take a look
at it to see if we would come to the same conclusion.  If we
would, obviously it would have some impact on what we would
COMMISSIONER LEHMAN;:  So you can pass that back.  Thank you
very much.
Now we're ready for Mr. Scanlon, Tim Scanlon.  Would you
identify where you're from?
Back to the index of speakers for Arlington

Forward to Timothy Scanlon