Subject: Re: IC's patent-pending technology
From: <stephen@xemacs.org>
Date: Tue, 26 Sep 2006 19:45:46 +0900

Ben Tilly writes:

 > On 9/25/06, Lawrence Rosen <lrosen@rosenlaw.com> wrote:
 > [...]
 > > FSB might instead help us understand how such patent claims (assuming they
 > > are valid) might affect free software businesses.
 > [...]
 > 
 > Let me just address this point and this point only.
 > 
 > With the license grant as I understand it, this patent cannot be
 > included in any open source software that wishes to be possibly used
 > commercially.  That might include being pre-installed on a computer,
 > sold on a commercial CD, be embedded in a commercial OS, etc.

It can't be distributed together the commercially usable portion.  It
could, however, be distributed separately (in your leading example of
Debian, in "non-free" along with Aladdin Ghostscript).  Am I missing
something?  (Of course Debian might choose not to do so, but I don't
see any "can't".)

 > by projects such as Perl, Mozilla, Apache or Open Office.  Any
 > proprietary competitor to those projects would, of course, be able to
 > negotiate a license and practice the patent.

 > Could it actually be used by any free software?  Yes, it could.  For
 > instance while Perl can't use it, one could put up a CPAN module to do
 > fast Unicode processing.  However it can't be used by any "big"
 > projects,

In what sense is CPAN not "a big project"?  If it's in CPAN, in what
sense can't Perl use it?  Ask Russ Nelson about the usability of
various patches with qmail---a CPAN module would be much easier than
that, no?

 > To the extent that rapid XML processing is valuable, the result is a
 > resounding negative for free software.  And is therefore negative for
 > any business that is built on free software.

Not so fast.  "To the extent that a faster CPU is valuable, faster
proprietary CPUs are a resounding negative for free software."  What's
the difference?  (I'm serious.)

 > In my opinion it would be far more useful for free software and for
 > free software businesses if you, say, gave a license for pure software
 > implementations in GPLed software.

They don't want to do that, and I don't blame them:

 > Sure, some people you want money from could do an end run around
 > you.

In Ghostscript's case, that was just about the whole market, no?
While the scope of this patent is much less than the scope of
Ghostscript in a printer or fax, hardware manufacturers might be
willing to free their second-best firmware for a competitive benchmark
on low-end products, while licensing the patent for the top-of-line
product---or simply specializing in the low-end, and cobbling together
their firmware from freely available or cheap modules.

 > And, speaking practically, the makers of a proprietary product will
 > have more of an incentive to license from you if they find that they
 > are losing benchmarks against someone that can practice the license
 > when they can't.

I'm sure it'll be available from CPAN, I imagine there will be a
modapache_icxml in short order, etc.

So they will be losing benchmarks, or will be shown similar benchmark
comparisons; International Characters marketroids will see to that.

 > If nobody else is using the technique and their software is already
 > seen as "fast enough", then they have little incentive to spend
 > serious money on you.  Sure, you might sell them a license, but
 > you'll probably get less for it.

You have no way to guess at probability.  The more competition there
is, the less profit there will be in IC's client markets.  So they may
get a bigger share of the proprietary profit if there's better free
competition, but it will be of a smaller pie.