Patent policy: RAND

I am sorry that I must write this to you in such haste, but I only found
out about the proposed change to a Reasonable and Non-Discriminatory
standard for patent licensing a few minutes ago.  As your comment period
closes very soon, a more careful composition is just not possible.

I am opposed to the change to RAND.  Software is a field unlike the
other endeavors covered by patents.  The exceedingly low barriers to
entry, and the large numbers of very talented people willing to make
contributions without recompense, simply do not exist in other fields.
Because of this, patents do more harm than good in the software world.
They should not be encouraged with the W3C's approval.


PATENTS HARM ALL TYPES OF SOFTWARE DEVELOPMENT

I have been a developer of computer software for over 20 years.  In that
time I have worked on commercial software for sale, proprietary software
intended only for internal use, and GPL-licensed Free Software.  The
proposed change to RAND harms all three of those efforts.

A very small part of GNU Diff is my code.  I was provided with a skeleton 
of the program, and a paper from molecular biology describing an algorithm
for comparing DNA sequences.  My task was to adapt that algorithm and
code it into the skeleton program.  

That program is today used all over the world, distributed with every
copy of Linux, and many other operating systems.  Microsoft even sells
it, as part of their POSIX utilities for Windows NT.  Had there been any
patents on this algorithm, even just patent applications, the work would
not have been done.  None of the Free Software developers could take the
risk of being sued for patent infringement.  I certainly would not have
written it, had there been a patent standing between me and full free
distribution.

Software developed within corporations for internal use also suffers
from patent limitations.  Internal software does not create any direct
revenue stream to support a large licensing fee.  A small fee is
pointless from the patent holder's perspective, because the actual cash
returned is dwarfed by the costs of contracting on both sides.
Companies thus adopt a mixed solution.  Don't do anything patented, and
if you do, certainly don't tell management about it.  

Commercial software would seem to make the best case for software
patents.  After all, a commercial product drives a revenue model, from
which royalty payments can be supported.  But in my experience,
producing software selling for thousands of dollars per copy, this is
not what happens.  My managers were quite eager to see my team develop
anything patentable, but they were very concerned that we not use
anything subject to anyone else's patents.  Although patent royalties
would seem to be a simple case of cost-benefit analysis, the actual
result was that _any_ cost was a complete bar to the use of a
technology.

The United States built the world's leading software industry during a
time when the accepted wisdom of our legal profession was that software
was not patentable.  The fortune of Microsoft is built on copyright, not
patent.  Free Software adherents happily produce and give away their
code.  Those who write for profit, and those who write simply as a gift,
are equally able to pursue their chosen path, and the world is richer
for it.  

The software world works very well indeed without the disruptions of 
patents.  It would be folly for us to encourage a change which upsets 
a system that works so well.  By permitting patented technologies in W3C
standards, the RAND rule upsets the high performance of the present
system.


THE PROPOSED RULE ON DEFAULT NON-DISCLOSURES IS FLAWED

Although I am opposed to any acceptance of patents within W3C standards,
the draft proposal's comment on RAND-by-default is particularly
worrisome.  The draft says that if a contributor, when asked about
patent claims, makes no response, the default assumption is the
contributor's patent holdings will be licensed under RAND terms.

As the world saw with the GIF standard, and again with Rambus and the
JEDEC committee's SDRAM standards, this approach does not work.  Once
adopted as a standard, developers commit enormous resouces to building
standard-compliant products.  Indeed, some will have committed
significant effort even to the further development and finalization of
the standard itself.  This work is then jeopardized by later patent
royalty demands.  

The response of the world's graphics community was the development of
the PNG format, a replacement for GIF.  PNG is now widely supported by
web browsers, but the cost has been a duplication of effort in
implementing both standards.  The response of the world's SDRAM
manufacturers to a "submarine" patent by Rambus has been years of
litigation.  I don't think any of us want to encourage that sort of
result.

The more effective rule is that failure of a contributor to disclose 
any patent claims promptly operates as a waiver of those patent claims
against all who implement the standardized technology.  A contributor
who wishes to have others help them standardize the contributor's
proprietary technology must plainly declare himself at the outset.  This
enables other working group members to make an informed decision as to
whether they will assist someone who may be their competitor.  It also
allows the public to make a reasoned analysis of the costs, as well as
the benefits, of adopting the standard.

I reiterate that I am opposed to any patented techology being permitted
in W3C standards.  However, if you are determined to proceed to RAND
nonetheless, then please at least give us RAND with full disclosure.


Thank you for considering my comments.

-- 
David Hayes
david@hayes-family.org

Received on Monday, 1 October 2001 00:00:41 UTC