🚀 go-pugleaf

Forwarded-For: rms@ai.mit.edu

I've heard that some people disbelieve that compress is patented.  So
I am sending this letter, which reportedly was printed in Dr. Dobbs,
March 1990.

>  Dear DDJ,
>    In the "Letters" column of your December 1989 issue, Mark Nelson
>  discusses U.S. Patent 4,558,302 entitled ``High Speed Data Compression and
>  Decompression Apparatus and Method.''  The patent was developed by Terry
>  Welch, a former Unisys employee, and is owned by Unisys.  According to Mr.
>  Nelson, I have been quoted as saying that Unisys will ``license the
>  algorithm for a one time fee of $20,000.''  As a concession to the modem
>  industry, Unisys has agreed to license the patent to modem manufacturers
>  for use in modems conforming to the V42.bis data compression standard
>  promulgated by CCITT, for a one-time fee of $20,000.  This $20,000 license,
>  however, is not a general license under all applications of our patent but
>  is limited to the specific application discussed above.
>    Responding to the second paragraph of Nelson's remarks, Unisys is
>  actively looking into the possibility that a large number of software
>  developers may be infringing one or more of our data compression patents.
>  We have only recently become aware of these potential infringers and the
>  process of taking action will take some time.
>    Unisys is happy to accept inquiries from persons interested in acquiring
>  a license to U.S. Patent 4,558,302.  If your readers have any further
>  questions, they should contact Mr. Edmund Chung of our licensing office, at
>  [Phone number deleted]
>  					Robert S. Bramson
>  					Unisys

(I did not include the phone number because I would hate to be
responsible for helping them get customers.)

I first found out about the problem in the draft of the POSIX user
portability extension standard (P1003.2a).  They propose to make
compress a requirement for POSIX despite the fact that it might be
covered by this patent.  I have quoted the relevant passage below.
(Note that there are rumours that the POSIX committee is negotiating
for royalty-free use of compress in POSIX systems, but Hal Jesperson,
the chairman of the committee, has told me this is not so.)

There is some disagreement as to whether the patent actually is valid,
and whether it actually covers compress.  Unisys says it does.  The
POSIX committee (in another part of 1003.2a) says the patent "seems
to" apply.  James Wood thinks it does not.  Reportedly some computer
companies think it does not.  I cannot have an opinion--I am not the
right kind of expert.  But even if I were, I couldn't be certain.
There's no certainty in what a judge will rule.

The only thing we can be sure of is that there is a reasonable chance
of being sued and losing.  If Unisys doesn't sue us this year, they
could sue us next year.  And fighting in court would cost hundreds
of thousands, even if we win.

Therefore, I think it is best to kill off compress as a standard (even
a de facto one) as soon as possible, since it will get harder with
time.  The more we become dependent on it, the more we are at a
disadvantage in any negotiation or maneuvering.

To those who argue the invalidity or inapplicability of the Unisys
patent under the current legal system, I wish success.  However, even
supposing we are lucky this time, it is just a taste of what will
happen over and over again.  We have to solve the overall problem of
software patents, or we will have to fight one patent after another,
each at great expense, for the rest of our careers.


>From POSIX 1003.2a:

5.5 compress -- Compress data

Editor's Note:  There are patent issues associated with
@code{compress}, @code{uncompress}, and @code{zcat}.  The
Lempel-Ziv-Welch algorithm described here may be covered by US patent
4 558 302, assigned to Unisys Corporation.  We are in the process of
obtaining a letter from the Unisys legal department that acknowledges
POSIX.2's reference to the patented algorithm and commits to license
the use of the algorithm to all POSIX.2a system providers on an
unrestricted, equal-access basis for a "reasonable" fee.  Such a
letter would meet the IEEE's requirements for specifying patented
technology in a standard.  If this letter cannot be obtained, these
three utilities will be dropped from the next draft.

It is not possible for the P1003 group or the IEEE to negotiate with
the patent holder on the specific terms of the license agreement or on
the size of the reasonable fee; this might be considered a violation
of US antitrust laws.  It will be the responsibility of those
providing POSIX.2a systems to deal directly with the patent holder.

The IEEE allows reference to patented items if it is the consensus of
the developers of a standards project to require those items for
technical purposes within a standard.  The IEEE Standards Manual
states, in Section 7, that "there is no objection in principle to
drafting a proposed IEEE standard in terms that include the use of a
patented item, if it is considered that technical reasons justify this
approach."

The technical experts of the P1003.2a balloting group are asked to
evaluate this technology on its technical suitability alone--does it
properly fit within the scope of the proposed standard and meet all
technical objectives?  Since all issues concerning patent licensing
are being resolved to the IEEE's satisfaction, ballot objections
outside the technical domain will not be considered responsive.