Subject: Re: New angle on the patent problem
From: "Karsten M. Self" <>
Date: Mon, 13 Sep 1999 17:24:08 +0000 wrote:
> Oops -- meant to send this to the list:
> >    Brian> Perhaps the license change could be done as an experiment.
> >    Brian> If the change killed the growth of GPL acceptance, it could
> >    Brian> be revoked.
> >
> >No.  The "patent-poison-pill" license is only good if it is (a) viral
> >and (b) widespread.  That combination makes experiments essentially
> >impossible: people who adopt this kind of license are betting the
> >future of their software.
> Hang on.
> The *current* state of affairs is that GPL'd software (I'll use GPL here, but I
> think any widely used viral license --free or not -- will do) can be killed by a
> patent.  There exists, at present, no protection against this.
> In practice, those companies engaging in patent extortion (er, um, I mean
> "enforcement" of course) do not appear to be free software companies.

Depending on how you want to define your terms, I'd suggest looking
close to home.  IBM has been known to, ah, aggressively assert its
portfolio.  Business Week ran an article in 1995 quoting a US$1b/yr
royalty target "Big Blue Is Out To Collar Software Scofflaws".
(  The article suggests
that the software and computer industries could be payin a 5-10% "tax"
to IBM.

Greg Aharonian, anti-SW patent crusader, regularly rails on IBM.  At the
same time, I'm inclined to see IBM as very friendly to certian free
software projects:  Linux, Apache, Jikes.  I don't know that I'd call
IBM a free software company in a pure sense [1], but free software
appears to work to the interests of IBM's services and hardware

> Indeed,
> wielding patents -- except as a means to enforce the terms of the free software
> license -- would be counterproductive for free software companies. [One might
> want to word the poison pill to permit patent enforcement where the remedy is to
> honor the free software license rules retroactively.]

I don't understand the remark in braces.

> Adding the "patent poison pill" provides some protection, but as far as I can
> see has no downside to the producers of free software.

Insofar as it might affect adoption, and hence services and support
revenues, I see a downside.
> If incorporated, however, I would suggest that it be tried in GPL.  It's
> particularly powerful there, as *everybody* uses gcc, and a lot of people are
> using Linux.

This is a goal of mine.

> Somehow, in the course of the discussion, I managed to miss the specific wording
> proposal.  Could some kind soul email it to me?

There is no specific language, I'm hammering out a mechanism.  RMS and I
have discussed this in terms similar to those of the Sun Community
Source License (SCSL) termination clause:

RMS wrote:
> Karsten M. Self wrote:
>     Your rights to copy, modify, and distribute the Covered Work
>     terminate in the event you seek patent infringement action on any
>     developer (original or subsequent contributor) of the Covered Work
>     for code contained within the work,...
> I am thinking of putting a clause something like this
> in version 3 of the GPL.  Perhaps it could be generalized
> to cover patent misbehavior against someone other than the
> developers of the program.

The SCSL language is

> 6.      TERMINATION.  
> [...] This License will terminate immediately as 
> contemplated in Sections 7.1 and 8.4 or at Developer's 
> discretion upon any action initiated in the first instance by 
> You alleging that use or distribution of Original Code, 
> Upgraded Code, Your Shared Modifications, Your Error Corrections 
> or Specifications by Developer or any other Contributor or 
> Licensee infringes a patent owned or controlled by You. 

There are other licences which have extended the concept to include a
negotation involving copyright-for-patent among the claimants.  The
novel aspect of my proposal is to extend this negotiation beyond the
immediate parties of any particular patent dispute.

 - Patent for copyright swap -- requesting patent royalty on 
   the one hand opens the door for a developer to request a 
   copyright royalty.

 - Joint enforcement -- any developer -- Original or Modifier,
   can seek enforcement of license terms.

 - Extension to class -- the patent enforcement termination 
   clause is applied not only to the Covered Work, but to 
   a class of software. that any patent enforcement claim against any software within the
covered class opens the door to a copyright royalty claim against the
patent enforcer, by any developer of software within the class.

The language I'm envisioning is something like this:

    Your Rights to the Covered Work, but not your Obligations, 
    terminate in the event you seek patent infringement action 
    on any Developer (Original or Subsequent Contributor) of 
    the Covered Work, or other works in the Collective Patent 
    Defense Class unless the following conditions are met:

    (i) You withdraw your litigation claim.

    (ii) You agree to pay any Developer of this work who files
      claim, a mutually aggreeable royalty for past and 
      future use of Covered Work contributed by the 
      Developer, should such a claim be filed within (specified 
      time -- one calendar year?) of the date of your 
      litigation claim.

(Capitalized terms would be defined within the license.)

I'm not sold on the specific language, and would welcome criticisms or
suggested improvements.

The effect appears to me to be the one that RMS desires:  of liberating
free software from patent enforcement, at least for patents held by
licensees of free software.  The positive spin is to see this as
tit-for-tat:  copyright for patent rights.

> Jonathan S. Shapiro, Ph. D.
> IBM T.J. Watson Research Center
> Email:
> Phone: +1 914 784 7085  (Tieline: 863)
> Fax: +1 914 784 7595

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

SAS for Linux:
Mailing list:  "subscribe sas-linux" to    
  9:03am  up 3 days, 12:43,  0 users,  load average: 0.32, 0.20, 0.15

[1] Showing it's possible to use "IBM" and "pure" in the same sentence.