RE: BOCU-1 spec

From: Philippe Verdy (
Date: Mon Feb 19 2007 - 02:34:56 CST

  • Next message: Lisa Moore: "Re: BOCU-1 spec"

    De Asmus Freytag:
    > Well, that pretty much describes the process used by Unicode as well,
    > but it turned out to be not so easy. The conditions under which the IP
    > license was offered appeared to leave it to the discretion of the IP
    > owner as to who to issue a license to, and there has been at least one
    > documented issue with a licence refusal. The membership of the
    > consortium concluded that such issues prevented this specification from
    > being published under a Unicode imprimatur and withdrew the draft.
    > If IETF or any other body publishes such specifications with these kinds
    > of IPR issues unresolved, that only shows that such standards have lower
    > levels of guarantees for implementers than Unicode is willing to accept
    > for its standards. Perhaps it is a hallmark of "real" standards to be
    > defective in that regard.

    The current rule at ISO (and also ITU) is that IPR licence requests must be
    honoured on a non-discriminatory base, but this does not void the IPR
    claims: nothing forbids the IPR to be licenced against a fee, provided that
    it is at a "reasonable" cost. This term is quite fuzzy, because there are
    many IPR claims that were submitted to ISO and accepted within their
    published standards, despite the licence fees are really expensive (notably
    all those that belong to the MPEG series, and DSP in general.

    So not only you must pay to get a copy of the published standard (well it's
    generally not very expensive to get such copy, except when a standard
    implementation requires many cross-related documents sold separately), but
    we must also pay the licence fee and royaltee fees on our own
    implementations. The worst of them being the royaltee fees that are based on
    expected estimated future sales, with an initial fee covering unreasonable
    amounts of future sold units. This is an accepted practice that really
    favors only the largest organizations, that can monetize the system with
    their own patents (so large companies like IBM or Microsoft are not affected
    by the large standardization bodies and can get those related standards for
    almost free).

    So non-discrimination of licence grants is not the only issue ; there's also
    the risk of future claims to continue to use a patent that was initially
    licenced for free. We've seen much enough companies pushing others to adopt
    their standard with initial adoption plans without even saying what would be
    the future fees claims for those patents, and that suddenly come back years
    later to claim millions of dollars to the adopters. As this risk can't be
    eliminated, this also prevents the smallest organizations to adopt those
    standards, because they could not support the risk of future claims.

    The risk is then even higher when patent claims are directed to the
    customers of the early standard adopters: this is unexpected indirect sale,
    that customers did not initially accept directly. So they can sue their
    provider, and a provider that would be a small organization cannot take this
    risk of covering all possible possible future claims by a third party.

    For this reason, the base of all future fees should be explained, and
    nothing should be claimed to early standard adopters before a grace period
    where they can also evaluate the cost of changing of solution: either
    continue to use the IPR, or remove the too costly patent-covered features
    from their implementation.

    Due to all these risks, no standard can be really widely and safely adopted
    if there's not also a strict delimitation about how licence fees can be
    claimed later; if there's too much freedom left to the standard proposer
    that integrates IPR in their proposal, such proposal should be rejected also
    because it will effectively be discriminatory (only the largest
    organizations can provide the future patent litigation protection for their
    customers, especially when it concerns products that will be distributed in
    millions of units, like TV set boxes, DVD players, and so on).

    Unicode and ISO/IEC 10646 standards are intended to be distributed and used
    in billions devices and software packages, because they are intended to be
    universal and be there for very long (decennials or more), without any easy
    replacement (except some other existing national standards like Chinese
    GB-18030 which only covers the repertoire but not really any algorithm like
    Unicode does). The risk is much too high for customers and distributors (no
    one can take the risks of covering future claims). So this is an excellent
    reason, for ISO and Unicode, to reject proposals that are not completely
    protected from future IPR claims.

    There's no reason to make the Unicode standard licensed against any fee, or
    used after claimed IPR licences have been payed by every adopter: it is
    enough that this standard will be used, studied, and enhanced by lots of
    member participants, who also have the opportunity to benefit of early
    experimentations and get profit from the early marketability of their
    products implementing it.

    If only this could be the same thing in the MPEG standard series, because
    image and audio are the other two important medias, needed in addition of
    text, to transport information. Thanks, ITU's H.264 is coming now, let's
    hope it will remain patent-free (the last issue in the MPEG series was in
    the JPEG format for the quantization algorithm, and the patent is now in the
    public domain since last year).

    This archive was generated by hypermail 2.1.5 : Mon Feb 19 2007 - 02:38:44 CST