response to the W3C Patent Policy Framework Working Draft


   We would like to take the time to thank the W3C for extending the
   comment period for this Working Draft.  It appeared on our radar
   only when much of the other free software community took notice at a
   very late time in the comment period.  The extension has given us at the time to review the draft and to mull over what the
   suggested changes would mean to our project and to the Web at large
   as seen through our eyes.  We do have some concerns about the
   direction that the W3C is going as expressed through this Working

The W3C Mission

   We feel that the existence of RAND-based standards endorsed by the
   W3C goes against the stated role of the W3C as defined in its
   mission statement: ( located at: )

     "3. Standardization: W3C contributes to efforts to standardize Web
     technologies by producing specifications (called
     "Recommendations") that describe the building blocks of the
     Web. W3C makes these Recommendations (and other technical reports
     freely available to all."

   This mission statement item does not make clear whether the use of a
   recommendation is free, as opposed to the recommendation being
   freely available, but its uses potentially encumbered.

   We feel that the existence of RAND-based standards endorsed by the
   W3C goes against the stated Design Principles of the Web in the W3C
   mission statement: ( )

     "1. Interoperability: Specifications for the Web's languages and
     protocols must be compatible with one another and allow (any)
     hardware and software used to access the Web to work together."


     "3. Decentralization: Decentralization is without a doubt the
     newest principle and most difficult to apply. To allow the Web to
     "scale" to worldwide proportions while resisting errors and
     breakdowns, the architecture(like the Internet) must limit or
     eliminate dependencies on central registries."

   The inclusion of patents into the W3C's standards moves the W3C from
   being an organization that "promotes and develops its vision of the
   future of the World Wide Web" to an industry group that defines
   standards to build revenue streams for its Members.  None of the
   goals of the W3C talk about making money from the infrastructure of
   the Web.  In fact, the discussion of commerce of any kind is
   noticeably absent.  The role of the W3C appears to be to help build
   technical standards for the Web so that all other forms of
   communication and commerce can take place on top of that

   If there needs to be a venue where companies can get together and
   create documents that describe their patent-encumbered standards,
   they should do that outside of the W3C.  The W3C should promote
   standards that are truly freely available.  This would promote truly
   interoperable software and standards and would put the resulting
   technologies into the hands of as many people as possible.

Selling Standards Compliance

   The direction and deployment of standards requiring the use of "RAND
   Licenses" will require a focus on the patent holder rather on the
   decentralized forces stated in the Design Principals.  Even worse, a
   standard including "RAND" Licensed technology could potentially be
   tied to the single implementation of that vendor's choosing.  (The
   RAND license guidelines don't rule this out as near as we can tell.
   It only specifies that the license must be available equally to all.
   Non-discriminatory licensing might dictate the use of a specific

   Vendors might offer to grant patent licenses for those creating
   alternative implementations, but might make it inconvenient to
   obtain such licenses as compared to licensing the vendor's software
   and patent as an integrated package, thus effectively discouraging
   the creation of alternative implementations and helping to ensure
   that the vendor's own implementation becomes the de facto "standard"
   in the marketplace.

   If you truly want to allow "(any) hardware and software" to
   participate in the Web then we believe that the use of RAND patent
   licensing significantly undermines that goal, since the use of
   patents that are not royalty free raises enormous, perhaps
   insurmountable barriers to broad adoption, particularly for software
   licensed under free software or open source software terms.

   A standard that is encumbered by RAND Licenses may not be evaluated
   in the same way that standards are usually evaluated.  In the case
   where the patent holder does not allow for more than one
   implementation or discourages the creation and use of alternative
   implementations in favor of their own, that standard can not be
   tested for interoperability.  Interoperability is demonstrated by
   proof that multiple existing implementations can operate together,
   following the standard.  The existence of multiple implementations
   is up to the licensing policies of the patent holders instead.  This
   allows licensing policies to affect the technical viability of the
   standard itself, which is a sure path to deteriorating standards.

Bait and Switch

   Section 5.3 of the Patent Policy Framework also lays out a method by
   which the licensing mode of an existing working group can be
   changed.  As proposed, the mode change renders any licensing
   commitments void that were made under the old charter for that
   Working Group.  This opens the door to a lot of possible abuse in
   our opinion and violates the good faith of the members of the
   Working Group who participated in a royalty free mode.

Default Licensing Mode

   We feel that the fact that any Essential Claims made by Members are
   automatically licensed under RAND terms instead of royalty free
   terms goes against the principle of least surprise and violates the
   spirit of a working group that works together in good faith.
   Essential Claims should be licensed under royalty free terms unless
   the Member opts out.  This is a simple choice.

Patents and the Growth of the Internet

   We feel that the Internet and the Web evolved into the mainstream
   without the use of patented standards and the statement that "The
   second decade of the Web has already demonstrated that patents will
   be a factor in the ongoing development of the Web infrastructure" is
   an inadequate foundation for the W3C's proposed actions.  By
   allowing "RAND" licensing, the W3C will eliminate the ability of
   individual programmers (except those who are very rich) to develop
   and implement standards-compliant software.  Standards compliance
   becomes possible only for corporate entities who have adequate
   patent licensing budgets.  This will dissipate the very energies
   that ought to be encouraged.

   If the W3C endorses the use of patents through the use of RAND
   licensing it means that companies participating in the
   standardization process will have one more reason to pursue patents
   in key Web technologies.  This will limit the number of players in
   those key Web technologies and could prevent their widespread
   adoption in many implementations.

Universal Participation in Standards

   A significant goal of many free software and open source projects is
   the creation of a freely available implementation of a standard.
   For example,'s licensing policy includes the use of the
   Mozilla Public License ("MPL") which includes a specific,
   royalty-free patent grant applicable to the code governed by the
   MPL.  The proposed policy will make such implementations far less
   likely.  The proposal would allow the W3C to promulgate standards
   that can be implemented only by those with a patent licensing
   budget, since even if the royalties for any single RAND license are
   relatively small, licensing the entire set of patents required for
   real-world products might require substantial amounts of money.
   Under this scenario, a W3C "standard" will be out of reach of much
   of the development community, and may well be worse than no

   A standard that can't be implemented by free or open source software
   projects, but can only be implemented by those few possessing enough
   revenue to pay for it, is not an effective standard.  The W3C should
   not seek to define the future of the Web in a way that inhibits the
   development of open source and free software.

Software Patents and International Law

   We are also concerned that the RAND licensing proposal threatens the
   development of effective, international standards.  The extent to
   which software can be patented varies a great deal between the
   United States and much of the rest of the world. Patents issued by
   the United States affect implementors outside the United States, but
   this does not make U.S. patent policy an accepted international
   policy.  Those companies which have the money and are interested in
   the U.S. market may well try to implement RAND-encumbered standards.
   But there is little reason for governments to endorse or adopt
   standards which require payments for a type of intellectual property
   recognized primarily in the United States (i.e., software patents).
   And those entities (governments, companies, open source and free
   software projects and individuals) whose primary interest in the Web
   is not focused on the United States will need to evaluate whether
   the imprimatur of a W3C standard is worth determining the degree to
   which the RAND Licensing purports to affect activities outside of
   the United States.  The international appeal of a W3C label may be
   reduced since (as noted above) there will be parts of the
   U.S. development community that will not be able to participate in
   the W3C RAND-encumbered standards as well.  Recent data suggests
   that a great deal of the growth of the Web in the upcoming years
   will occur outside the United States.  We believe it is unwise to
   try and impose a U.S.-centric view of software patents on this

Ambiguity of RAND Licensing Definition

   We are also concerned that the "reasonable" in RAND will prove to be
   a misnomer.  Patent licenses are often very complex documents.
   Trying to simplify the required terms, and develop requirements
   applicable to the range of settings in which a standard ought be
   implemented will be difficult, if not impossible.  A few examples
   are stated below, many more could be described.

   Example 1: Lack of process for determining "reasonable."

     We cannot tell if the terms of a RAND License will be reviewed by
     the W3C before a technology is included in a Recommendation.
     Either answer to this question leads to serious problems.  If the
     answer is yes, then the W3C will need to become a licensing body,
     reviewing terms and payment requirements of the proposed "RAND
     Licenses" before technology can be included in a Recommendation.
     This will slow the work of the W3C dramatically.

     If the answer is no, then we fear a W3C Recommendation will be
     significantly devalued.  Until the terms of each of the RAND
     Licenses required to implement a Recommendation are publicized, it
     will be impossible to know the cost of implementing a
     Recommendation, or the business practices that will be required to
     comply with the applicable set of RAND Licenses ( see Example 4
     below for details on this point).  It's hard to imagine having any
     idea of the value of a Recommendation in this setting; the
     Recommendation may be out of reach of much of the development
     community at the moment of its creation.  And the publication of
     the set of applicable RAND Licenses may simply confirm that the
     ability to implement a Recommendation is effectively limited to a
     small wealthy group.

   Example 2: Difficulties in determining "reasonable."

     It's nice to believe that "reasonable" can be an effective
     standard. But even people who share a common set of referents may
     find that they disagree vehemently over what is considered
     reasonable in a particular setting.  Here are some areas where we
     might expect disagreement:

     a. Who determines "reasonable?"  Does the patent holder
     unilaterally set a price?  Unilateral price setting could make
     sense if the price and terms are disclosed before the
     Recommendation is complete, and factored into the determination of
     whether to include something in a Recommendation.  But
     promulgating a Recommendation that requires a patent license to
     implement, and then allowing the patent holder to unilaterally set
     the price is a recipe for disaster.

     b. What criteria go into determining a "reasonable" price?  Is a
     reasonable amount that which would allow the licensor to recover
     its costs in a set period of time?  Or perhaps an amount that
     would allow the licensor to obtain a certain return on investment
     in a set period of time?  If so, would this be the type of return
     the venture capital community generally looks for, or a return on
     investment found among more mature companies?  Taking a different
     tack, might a "reasonable" amount be that which approximates the
     value of the technology to the Recommendation?  Who determines
     this?  What happens if the cost of developing the patented claims
     is high, but the patent affects only a small piece of the

     c. If a reciprocal patent license is required, does its scope
     affect the amount which would be "reasonable?"

     d. Is the "reasonableness" of the payment amount affected by the
     more limited scope of software patents outside of the United
     States?  If so, how is it applied in a non-discriminatory manner?

   Example 3: Limits on incremental improvements to a Recommendation.

     Point 3 of the definition of "RAND Licenses" states that such a

       "3. may be limited to implementations of the Recommendation, and
       to what is required by the Recommendation;"

     This sounds simple and "reasonable" but leads to problems.  Assume
     Company A pays Company B for a "RAND License" to a set of
     Essential Claims.  Company A then develops an "improvement" to the
     Essential Claims which have been licensed.  That is, Company A
     creates an invention that builds upon, but is separable from the
     Essential Claims it licensed.  Does Company A have the ability to
     implement its improvement?  Perhaps not.  It may be that "what is
     required by the Recommendation" doesn't include the improvement,
     so the "RAND License" doesn't include it, and nothing in the
     proposal requires Company B to be reasonable in its response.
     Company A is paying for the right to implement a Recommendation,
     and has still been effectively precluded from innovation.  Now
     assume Company B develops this improvement.  As far as we can
     tell, company B has no obligations regarding the improvement.
     Company B is now in the setting where its "RAND License" is
     required to implement a standard, and where improvements to that
     technology can be subject to unreasonable, discriminatory
     licensing practices.  We could be heading for a situation where
     the W3C promulgates standards that when they are included in
     future products, even those created by other people, can only be
     implemented only through additional payments to the initial patent

   Example 4: Potential effects on business and distribution models.

     Point 6 of the definition of a "RAND License" states that such a
     license "...may include reasonable, customary terms relating to
     operation or maintenance of the license relationship such as the
     following: audit (when relevant to fees), choice of law, and
     dispute resolution."

     This may sound non-controversial, but raises a set of problems,
     which we'll illustrate using the audit provision.  Audits are used
     to determine that the correct amount of fees have been paid.  For
     an audit to make sense, there must be some soft of tracking
     requirement in place.  A payment requirement of $XXXXX.xx per
     quarter might not require an audit.  But payment terms of $X.xx
     per copy would require knowing a number of copies, and an audit
     requirement would make more sense here.  So if an audit is
     permissible in a RAND License, we can assume that a requirement to
     track the number of copies of a product may be required.  This
     sort of tracking obligation may impose restrictions on the
     behavior of a licensee.  For example, suppose the patent holder
     determines that a "reasonable" fee is a per-copy fee, of $X.xx per
     copy distributed.  Not all implementors of a standard count the
     number of copies that are distributed.  This might be true of
     those that permit anonymous downloading of software from mirror
     sites, permit copying after download, and also those open source
     and free software projects that provide freely downloadable source
     code.  The "reasonable" per copy fee could have the effect of
     requiring such entities to change their entire method of
     distribution. This required change of business and distribution
     model might be fine in settings where people individually
     determine that a technology is worth it to them.  We feel strongly
     that it is inappropriate to be forced into a predetermined
     business and distribution model in order to implement a W3C

The Fast Pace of Technology or "Standing on the Shoulders of Giants":

   This statement is made in Section 2.2 of the Working Draft of the
   W3C Patent Policy Framework

     "* Importance of interoperability for core infrastructure, lower
     down the stack: Preservation of interoperability and global
     consensus on core Web infrastructure is of critical importance. So
     it is especially important that the Recommendations covering
     lower-layer infrastructure be implementable on an RF
     basis. Recommendations addressing higher-level services toward the
     application layer may have a higher tolerance for RAND terms."

   The policy framework does not define where the division points lie
   on the stack.  Are HTTP and HTML at the top or at the bottom of the
   stack?  Depending on where you stand those could be at different
   levels of "the stack."

   How a piece of technology is looked at on that stack is only a
   function of time.  What is considered basic infrastructure at the
   bottom of the stack at one point in time might have been considered
   at the top of the stack years or even months previously.  For
   example, we would consider a browser to be pretty low on the stack
   considering what people have built on top of just a browser.
   However, a more strict interpretation of this statement in the frame
   of the OSI model would include a web browser at the top of the

W3C Recognition of Patents

   Despite the concerns we've expressed above, we feel that there must
   be some good things that can come out of this patent work that the
   W3C is exploring.  For example it's good to see a written
   requirement that patent claims are to be stated up front by Members
   before a Working Group is created.  The Members of that Working
   Group are giving rights by participating in that working group.
   This is a good step in the right direction, whether or not the W3C
   chooses to support RAND licenses.  Even if the W3C chooses not to
   support RAND licenses we would personally love to see the W3C
   require that any Members grant royalty free licenses before working
   in a Working Group.  The Patent Policy Framework states that there
   have been royalty free patents in the past with regard to W3C
   standards.  Why not make this doctrine instead of opening the door
   to non-free patents that restrict development and deployment?


   Simply put, we feel that the proposed changes in the way that the
   W3C operates threaten the organization that we have turned to in the
   past as the arbiter of standards for the Web.  The W3C is a resource
   for those groups who want to create a standard that will allow
   multiple implementations to interoperate.  With the addition of
   patents into that mix, the standard's merit is no longer the
   important factor in determining how that standard is used and who
   can use it.  Instead, the effective power to make that determination
   moves to the patent holder.  This limits the creative ways in which
   existing standards can be reused in new and interesting
   implementations.  This means that the effective power to determine
   the direction of the Web and the Internet is in the hands of the
   patent holders instead of the standards bodies and the implementors
   of those standards.  This fact itself undermines the entire purpose
   of having a standards body.  The alternative is having corporations
   choosing a direction for the Web for their own benefit at the
   expense of its users: the general public.

   We hope that the W3C Working Group and the Director will take these
   comments into consideration when considering this Patent Policy
   Framework.  Thank you.

The views expressed herein are the view of the named individuals and
do not necessarily represent the view of the employer of any

Mitchell Baker <>
Christopher Blizzard <>
Frank Hecker <>
Daniel Nunes <>
Dan Mosedale <>
Gervase Markham <>
Myk Melez <>
Brendan Eich <>
Dawn Endico <>
Asa Dotzler <>
Mike Shaver <>

Christopher Blizzard - we're on a mission from God.  Still.

Received on Tuesday, 9 October 2001 21:54:01 UTC