Subject: Re: New angle on the patent problem
From: "Karsten M. Self" <>
Date: Mon, 13 Sep 1999 06:15:29 +0000

Assaf Arkin wrote:

> > but it's similar to the licenses you mention.  BTW, what's IPSL?
> > Haven't heard of it.
> IBM Public Source License. Their acclaimed Jikes (Java compiler) is
> distributed under it. It's open source compliant.

Familiar with it as "the Jikes license".  Note the rather awkward patent
construct though -- patents as used in the original source if your
program includes at least 60% of the original source in a way that it's
reachable in normal execution of the program.  I was one of the folks
who pointed out that the original wording (excluding the "reachable in
normal execution" language) allowed patent capture by putting original
code in a comment or no-op loop.  The Jikes patent language is still too
awkward for my taste.
> > > I do not see how this clause deals with the issue.
> >
> > It doesn't.
> >
> > This does:  the clause doesn't specify _one_ program.  It defines a
> > termination condition based on action against any of a _class_ of
> > programs.  Examples:  Any GNU project.  Any software licensed
> > under GNU
> > GPL (this is where the patent-grab aspect starts to enter,
> > and becomes a
> > bit frightening).  A select list of projects (say:  Linux,
> > GCC and libs,
> > BIND, and Apache).
> GNU covers a lot of software, but surprisingly I can get through the day
> without using any GNU code. Unfortunately, many of the companies we are
> afraid will come with a patent claim are not well known GNU users.

These are examples.  A key IMO is finding the right class.

> The BSD or Apache licenses cover a narrow class of software. If I don't
> ever use an Apache server, or only use it through my ISP, I can freely
> sue the Apache group for patent infrignment. If I use Linux (GNU) I am
> free to sue patents used in BSD.
 < ** snip ** >
> So, even if we are talking about a class of software, the solution is in
> no way complete.

Not necessary, perhaps not desireable, depending on costs.  The
threatened group has just diminished markedly in size.  I see resolving
the patent problem as an exercise in risk reduction -- much the same as
security or other business risk management practices.  Total resolution
is an ideal, partial containment is a reasonable achievement.

> What happens when there is no class of software. For example, many open
> source projects out there do not fall under the GPL, they fall under a
> BSD-like license. Who offers them protection?

To be determined.

 - The license could include software covered under other licenses,
either globally (all softare under this license or license X) or
selectively (software based on A, B, C,...).

 - There are other avenues of protection.  IBM has taken a major stake
in Apache.  Though they won't say they will do so, IBM is in a good
position to protect its own interests through its own patent portfolio. 
It might even (and IMO probably would) act on behalf of its customers,
Apache users as a class, or independent developers.  They're simply too
large to mess with.  I call this the "patent patron" model.  The GIMP
faces more significant problems WRT patents, lacking a similar patron. 
Maybe Kodak or Nikon would want to take them on (electronic photography,
camera sales cheap 'lectronic darkroom).  There's no need to seek a
single strategy.

 - License-based protection may provide incentive to unify free software
licensing.  I dislike the current trend to YAPL (Yet Another Public
License) -- it's balkanizing -- and see the unification incentive as a
Good Thing.

> > The termination clause specifies that infringing action allows _any_
> > developer (not just the developer(s) affected by action) to
> > enforce the
> > termination clause, and negotiate a royalty term for use of covered
> > code, past, present, and future (life plus 70 years).
> Once again, the issue is the ability to go in patent claim against a
> group of users that is matched with a group of developers that do not
> produce any software you happen to use. 

No.  Covered as action against one set of developers activates
termination option for other developers in class.  Big, broad stick.

> And "Abandon use of software and
> select non-free alternatives" seems to be a very popular choice these
> days. It will be quite a while before essential software falls under a
> broad license, and then it can spell trouble in other ways (as you
> predicted).

This is the sticking point.  It will probably affect how far the concept
can be taken, and how quickly.  However there are compelling arguments
to go with free software, and for a very large class of users, little
downside impacts of doing so (there are more non-holders of software
patents than holders).  If, as appears the case, free software offers
compelling business advantages then you'll see market selection toward
firms who find the patents for free software trade to be acceptable. 
Note also that there is no affect on the ability of a company to enforce
its patents for _non-free_ use.

> My second case expresses the concern a company might have against using
> or contributing to open source software to begin with. Until you can
> guarantee that all possible patent claims will arise from companies
> using said open source license, you are better off not using said open
> source license, which reduces the possibility of that happening to begin
> with.

See Turnbull's objection.  Same.  Significant issue.

> arkin

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

SAS for Linux:
Mailing list:  "subscribe sas-linux" to    
 10:52pm  up 3 days,  2:32,  0 users,  load average: 0.17, 0.08, 0.09