The Copyright Act preempts the GPL

[Russell McOrmond]

On Wed, 28 Jan 2004, Ken Brown wrote:

> Well, if everything else is a derivative...then how can anyone claim to
> be the original owner?  I mean how many original owners can you have?
> There can only be one, whether the license says you can transfer it to
> 10,000 people...right?  

  You have to ask the question:  what is this "it" that you are looking
at?

  Is it a "program" made up of many modules, subroutines and lines of 
code?
  Is "it" one of those lines of code or a subroutine, or a source (.c, .h
or .py) file?


  Linux is made up of many modules where some "first owner" of copyright
created that module and then linked it together with other modules.  
Other contributions involved changing these modules with the permission
(granted by the license) to create derivative modules.

  There may be a "first owner" of the Makefile used to link everything
together, and in the case of Linux that Makefile is likely under the
copyright of Linus Torvalds.  That is not certain, however, as someone
else could easily have started from scratch at any point and replaced
Linus' Makefile.  The ability to easily start from scratch and replace
entire modules and relink without having to get lawyers involved is part
of what allows popular Open Source projects to mature so quickly.

  Each module may have its own first owner of copyright and many creators
of derivatives from that original work.  These modules are then linked
together to create a much larger thing called a "program".

> So help me understand your earlier point.  To charge that someone has
> violated a copyright, doesn't the original owner have to make the
> charge? Otherwise, we don't know where the true property rights
> ----started right?

  The creator of the derivative work has copyright on their own work as
well.  Every copyright holder -- of original works or of an authorized
derivative works -- can sue for infringement of their copyright.

  In a complex linking of modules such as Linux there are many original
copyright holders and many copyright holders of derivatives so there are
many people who can make the charge of copyright infringement in relation
to the part of Linux that they hold copyright over.

  It is interesting, but peer production opens up a whole new possibility 
of "class action" copyright infringement cases.

  This is an important thing to understand with new models of knowledge
production such as FLOSS.  We are no longer talking about simplistic cases
where whole programs are created by or otherwise under the copyright of
one entity (whether it be a natural person or a corporation).  We are
dealing with situations where a program is made up of modules with each
module possibly having its own paternity and history that is entirely
different from all other modules that are then linked together.

  In the case of some of the recent examples of corporate contributions to
Linux these modules were actually written for other kernels and then, as
the legitimate copyright holder, they ported these modules to work with
Linux.  SCO claims that if the original kernel was the UNIX kernel that
they now (possibly?) hold copyright over that the copyright holder is not
allowed to links this with Linux.  For those parts that were modifications
to original UNIX code that were 'derivative works' they have a case but
that code is to my understanding not under discussion.  For new modules
that were authored SCO have no case at all (at least under copyright law)
to claim the copyright holder of this new code is not allowed to license
the code in as many different licenses and whatever license they wish.



  Copyright supports the concepts required for "commons-based peer
production", even though having to understand the complexities involved
wasn't something that was as important in the past as it is today.

  When you have "software manufacturing" you still have many situations
where there isn't a single copyright holder.  Microsoft isn't the sole
copyright holder of Internet Explorer as this is a derivative "program"
(both derivatives of existing code and new code) from the Spyglass
licensed versions of Mosaic that they started with.  Microsoft is going to
be the copyright holder of the (authorized) changes they made to the code,
and new modules that Microsoft employees were first creators of.


  There may be a major copyright holder that then licenses modules from a
third party and bundles it together to create a whole program.  That
program may have modules with tens of separate paternities being linked
together and marketed.  With a relatively small number of separate
paternities the haphazard licensing arrangements of "software
manufacturing" made with each copyright holder was still practical.


  With FLOSS and peer production you then step that up a few orders of
magnitude to where you are talking about single "programs" and "projects"  
that have hundreds and soon likely thousands of separate paternities being
linked together.

  With this many contributors the concept of making individual contractual
and monopoly-rent-seeking (royalty payment) arrangements with copyright
holders does not make sense.  Common/compatible monopoly-rent-free
licensing becomes the only reasonable way to manage the complexity.  It
turns out that not only do common monopoly-rent-free licensing agreements
simplify the complexity for these large number of contributions, it scales
perfectly the other way and simplifies license negotiation such that it is
still far easier than the "software manufacturing" way of doing things
even when there are only two copyright holders.


---
 Russell McOrmond, Internet Consultant: <http://www.flora.ca/> 
 Governance software that controls ICT, automates government policy, or
 electronically counts votes, shouldn't be bought any more than 
 politicians should be bought.  -- http://www.flora.ca/russell/

--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3