Subject: Re: software patents in the wild
From: "Stephen J. Turnbull" <>
Date: Fri, 19 Aug 2005 15:31:48 +0900

>>>>> "Santiago" == Santiago Gala <> writes:

    Santiago> What can be patented is ways to manufacture those
    Santiago> [distributable] goods, or improved ways to perform the
    Santiago> function in them.

If the patented way to perform the function is embodied in the good
that is distributed, distribution of the good without a license is
illegal under _U.S._ law.  And that is the important point.  The
U.S. has never hesitated to strongarm foreign entities on U.S. soil
for acting against U.S. interests (national or private) elsewhere.

As for the rest of your statements, while IANAL, I view them with
extreme suspicion.  I would not want to try to run a business based on
them, although I don't hesitate to participate in hobbist development
activities without legal advice.  ;-)

    Santiago> I'm not sure if distributing the source code of a
    Santiago> software implementing patents would be illegal.

Distributing it is not; that is the whole point of having a disclosure
requirement in patent law---to get the knowledge widely
distributed.[1] However, without a license, compiling and running it,
for the purposes claimed in the patent, is.  If you were to try to
make a business of distributing the source code making any claim that
it was useful in ways claimed in the patent, you will lose (i.e.,
defending yourself against a suit will be nontrivial and very
expensive, even if you win).  You might even lose the case even if you
could prove there were uses not claimed in the patent, at least in the
U.S., if the plaintiff could show that your customers were buying it
primarily for the purpose of avoiding the patented versions.

Thus, there is no reason not to write a textbook about patented
methods.  That requires somebody else to do the hard work of actually
writing the program, handling the edge cases, etc.  But distributing
source code that can be compiled and run by anybody undermines the
patentholder's monopoly---which is what patents are all about, trading
monopoly for publication.  I think it unlikely that this loophole
would be allowed, although patent-holders might choose not to enforce
for reasons of laziness, public relations, or (gasp!) public
spiritedness.  Doesn't mean you're safe, cf. LZW.

    Santiago> In fact, the patent office publishes (hence distributes)
    Santiago> the source code of those patents.

Aside: This is not necessarily true in the U.S., where core dumps have
been admitted as a "specification" of the "device."  ;-)

    Santiago> At least in old Europe, what the patent covers is the
    Santiago> use of the method or improved technique to manufacture
    Santiago> goods. So, for instance, Thomson is not even trying to
    Santiago> enforce mp3 patents on individuals using OS mp3
    Santiago> players. They could try to do this on people
    Santiago> distributing linux binaries, for instance, but I highly
    Santiago> doubt they could do anything on people distributing just
    Santiago> sources (say, a university), no matter how many patents
    Santiago> are there.

Excuse me, but in most cases the vendors do not want to enforce a
patent on a _decoder_.  Giving away decoders is like giving away
cameras: they make the money on content sales, just as Polaroid and
Kodak made their fortunes on selling and developing film.  

And it's possible that they _can't_, for the simple reason that there
probably aren't any such patents!  Decoders are often non-patentable;
they're typically "obvious to any competent practitioner"
unmarshalling operations.  It is quite possible that MP3 _decoding_ is
_not_ patent-encumbered.  (Parallel to LZW: GIF decoders and
uncompress were not covered by the Unisys patents.)

On the other hand, you have seen the DeCSS debacle and threats
(action? I didn't follow it that closely) against the developers of
Blade and LAME, which are precisely the encoding side of MP3.  What
does Thomson say about them?

Perhaps nothing, because they don't need to?  At least Debian sid
doesn't seem to have either packaged.  In general, it looks to me like
MP3 encoders are completely unrepresented in the Debian distribution
after a quick look through the sound category---all of the packages
that say they do MP3 encoding actually require separately distributed
applications, which are not available in Debian (by the given names,
anyway).  On the other hand, the toolame MPEG layer 2 encoder, which
Debian's blurb claims is not patent-encumbered, is included.  I don't
have time to review the MPEG specs, but that strengthens my suspicion
that MP3 _decoding_ is not encumbered.

My conclusion is that Thomson is at least as skilled at politics and
public relations as it is at engineering (they are the famous French
national champion, no?), and they are not half bad at engineering.

[1]  In practice, it has failed sadly, of course.

School of Systems and Information Engineering
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.