Subject: Re: [ Patent Cross-License Agreement]
From: "Karsten M. Self" <>
Date: Wed, 29 Sep 1999 06:50:33 +0000

I'm taking this online (to FSB) as this is getting beyond a few short

Mark Shewmaker wrote:
> On Tue, Sep 28, 1999 at 08:02:25AM +0000, Karsten M. Self wrote:
> > Mark Shewmaker wrote:
> > >
> >
> >  - There are a few general solutions posed to the patent problem.  Pools
> > and patent licenses being one.  I find patents and tools to commit them
> > to Open Source complex, hard to gather and use, and not particularly
> > prevelent among the developer community.
> Patents representing a threat to progress, something has to give at
> some point.  Either someone figures out a good way to let Open Source
> projects easily use patents, through legislative, judicial, 

Legislation is extremely unlikely:  legislation tends to reflect the
current power balance.  It's not us.

Judicial judgement is possible, though unpredictable.  It's also bound
by the limits of the law and Constitution.  Recent history doesn't
suggest a sudden change of heart against software patents.

You omitted regulatory reform -- essentially the way in which government
agencies are run.  Similar to legislation -- power interests would tend
to influence decisions, and there's a lot of vested interest in the
status quo.

...which leaves us

> or licensing arrangements, 

> (Which you more or less say in your patent vs. copyright
> poison pill article.)

We're in agreement on ends.  Means is to be decided.   Mind you I don't
think a pool is bad.  Just a lot of work to build and maintain.  I'm

> I don't think ignoring the problem is a viable option.  

It's not, if free software is to extend beyond hobby and fringe use. 
Large companies and user bases *are* potential infringement suit
targets, and can't scoff at the law or pack up and move offshore (which
might not even protect them) on a whim.

> On the easy-to-understand issue, L. Peter Deutsch said that my license
> is way too long, which I initially agreed with.  But now, I'm thinking
> that even though my version could be far shorter, that it's likely that
> by the time enough lawyers have looked at it, that it will grow back to
> being fairly large.  

Bigger isn't necessarily better.

> (Looking back at a simple apartment rental agreement
> has made me think that even 14 printed pages isn't too large in comparison.
> I'm guessing that by the time lawyers for companies with serious patent
> portfolios look at the agreement, that it will have grown in size again
> to address their concerns.)

Look at IBM's IBM Public Source License
(  Compare it to its
predecessors and the Sun Community Source License.  The IPSL has a more
uniform goal -- preserve public access to software, offer use of IBM
(and other contributors') patents, and use IBM's almighty patent
portfolio as a damned big stick.  It's a pretty short and sweet
document.  It was written by a company with a heck of a legal
department.  The first two Jikes licenses were much more confusing,
possibly longer, and _didn't get_ a few key elements of free software
(or how to keep a handle on patents).  The SCSL is confounded by Sun's
desire to look open source but not really be the same.  The document
reflects the confusion of goals.
> > There are a number of
> > potential strategic problems with a pool.
> True.  I think the trick is to make an agreement in which it is far
> more cost-effective for companies to participate and contribute their
> patents, than to cross-license patents with individual companies, so
> that the advantage appears all over their accounting books.

That's the obvious method.  You end up with a boot-strapping problem
though -- how to get the pool started, and how to grow it, before it
reaches its fully effective stage.  

If you try to jumpstart this process by bringing in a patent
heavy-hitter like IBM, you end up with problems -- mostly of perception 

 - of trust -- can the organization be trusted, is it a pawn of its
major players?

 - of anti-trust -- is a patent pool based in large part on IBM going to
be well received by the DoJ?

 - of strategy -- IBM might be more useful as a seperate, "pool unto
itself".  With its stated committments to free software, IBM hasn't
really given itself a choice of whether it will use its patents in
defense of free software or not -- it simply will, whether it's part of
an Open Patent Pool or not.  Witness the IPSL.

> Most companies get far more of a benefit through cross-licensing
> than through royalties anyway, even IBM.  (There's a quote on that
> on the LPF pages somewhere.)

IBM made US$1.5 billion in patent royalties last year (or is on target
to do so this year, I forget exactly).  That's real dinero.
> > Also anti-trust.
> I'm not quite as worried about anti-trust myself.

Do be.  It's a major concern of any patent pooling proposal,
traditionally.  The DoJ has revised its opinion of pools in some
instances, but tends to look at them unfavorably.  Patents (like
copyright) offer an artificial monopoly, the exchange is supposed to be
incentive for further invention.  Piling up the monopoly in one place
starts to raise Sherman-Act hackles.  I'm not saying the pool will be
anti-competitive, but there should be a prepared response to the
question, and an eye given to minimizing any such aspects of the pool. 
An informed (legal and economic) opinion would be very helpful. 

> I'm proposing only to push openness, and as I understand it, you're
> proposing to additionally allowing some ascpa (?) type licensing as
> well--convenient one-stop shopping, as it were.

The issue I ran into was that the pool would need a revenue stream to
maintain itself.  Ergo licensing, from which a percentage could be
skimmed.  Much of this used to identify and patent new technologies from
free software sources.
> >  - I was glazing over reading the whole multi-tiered pooling scheme.
> > Take a look at my version of same (granted I wasn't incorporating legal
> > language).  Know that IBM basically has three classes of patents:
> > license freely, license -- for a price, and don't license at all.  The
> > first is quite large, the last quite small.
> I've got to add more background info to the web page on the goals of
> the license, as opposed to just having the license there.  It's useful
> as-is if you're having trouble falling asleep, though.  :-)

No comment. <g>
> >  - Take a look at my copyright v. patent poison pill.  There's a
> > web-based sketch of it at
> > (  If you can find
> > the FSB archives, we've been discussing it over the past two-three
> > weeks.

> I'm waffling on my stance on the poison pill approach.  It may be
> because I'm obviously in favor of my own approach and am biased on
> the issue.  The main thing I dislike about it, I think is it's single-
> issue stance.  The license is basicly only useful for that one software
> work, and doesn't do anything in general to stop the threat of software
> patents.

No, it specifically addresses that issue.

The poison pill creates a trigger on *one* work based on actions against
any of *many* works.  If the classes are defined such that the many
works and one work are licensed under the same terms, you have a license
which effectively terminates rights to the entire class.  Because of how
a license works, however, each specific copy of the license addresses
only the work it is attached to.  I can't say "I am license to FOO, but
if you do BAR I will terminate your rights to FOO and BAZ".  The license
doesn't grant you rights to BAZ, it can't take them away.  The web of
licenses + works creates a general deterrent.

Mind you that the objections, particularly Stephen Turnbull's (companies
will avoid free software with such a cluase), are significant.

> One question:  I didn't know about your web page until LPD pointed it
> out to me in email Friday.  I had thought of adding it as a reference
> on my page, but you also seem not to want to widely publicize things
> yet--I'm not sure.  Would you mind if I added it to
> , or would you prefer I didn't?

Go ahead.  It's a piece of history at the moment.  There was a time when
the patent threat seemed imminent (Halloween Documents had just
surfaced) and I wasn't sure how much of a hand I wanted to tip.  Turns
out the idea's been around for a while, and I've reason to believe that
flagrant and widespread attacks against free software aren't terribly
likely, though we'll see harassment moves such as the Unisys gif (LZW)
patent enforcement from time to time.  The growing concern IMO is
coverage for the many companies growing up around free software.  This
is a major issue in the proliferation of free software licenses --
companies are looking for coverage not included in existing licenses
such as the GNU GPL.  Pity, as this leads to code balkanization.

Karsten M. Self (
    What part of "Gestalt" don't you understand?

SAS for Linux:
Mailing list:  "subscribe sas-linux" to    
 11:10pm  up 6 days,  2:09,  1 user,  load average: 0.04, 0.28, 0.50