Web Standards should avoid patents

I'm writing to express my opinion about the W3C's emerging policy on
patents.

W3C representatives have said that the W3C can't continue to operate as if
patents don't exist.  They point to the IETF and their policy on patents,
and argue that RAND-style patent licenses are common among standards
organizations, and the W3C should join in in this practice.

To the first point, I heartily agree: the W3C should be very mindful of
patents.  But they should be mindful of them mostly for the purposes of
avoiding them or knocking them down.  A fine example of the W3C's patent
policy gone awry is the SVG proposal.  SVG uses utterly standard
masking/compositing techniques, which are the subject of an absurd patent
held by Apple.  There are a number of good solutions to this problem:

  1. Knock down the patent, as was done in the P3P case.  It takes time, but
     the world is then a better, more open place, now free of one more
     ridiculous patent.

  2. Convince Apple to hand out an RF license to the patent

  3. Exclude the algorithm from the SVG standard.

The one solution that is not acceptable is to start producing
patent-encumbered standards for the web.

The only thing worse than a standard encumbered by patents is a standard
encumbered by submarine patents.  Yes, the RAND policy requires participants
to state all patents they know about.  But other patents may surface later
and be enforced nonetheless.  Even if the W3C allows for patents to be
included in standards, participants should be required to agree to RF
licenses of any patents that are undisclosed at the time of the standard
being set.  Otherwise, the incentives are all wrong: participants have no
incentive to find out about patents they may own that may impinge on the
standard.  More critically, even if participants knowingly insert encumbered
technology, it may be quite hard to prove that this was done, making the
disclosure rules that exist unenforceable.

Finally, RAND licensing terms make free software for implementing RAND-based
standards impossible.  This is simply unacceptable.  The web was built in
large part on free software, and to exclude free software now is both
ungrateful and self-destructive.  At a bare minimum, RAND licensing should
be changed in such a way as to allow RF licenses for free software and
non-commercial implementations.

A final note:  the IETF has not produced many patent-encumbered standards,
and is moving towards more RF licensing.  The W3C, to its detriment, is
moving in the opposite direction.  If such changes are made, then I believe
the "moral authority" of the W3C will be compromised, and the web may become
more fractured than ever.

Sincerely,
Yaron Minsky

-- 
|--------/            Yaron M. Minsky              \--------|
|--------\ http://www.cs.cornell.edu/home/yminsky/ /--------|

Open PGP --- KeyID 0x91F2C47C
Fingerprint: 6838 FF4E 256C BD24 453E 6EEA 0FBE 6CF8 91F2 C47C

Received on Thursday, 4 October 2001 11:01:46 UTC