[rms at gnu.org: Some problems on gnucash.org]

[Chris Shoemaker]

On Fri, Aug 11, 2006 at 03:13:17PM -0500, Linas Vepstas wrote:
> On Thu, Aug 10, 2006 at 05:16:32PM -0400, Chris Shoemaker wrote:
> > Hi Linas,
> >         If we're talking about the same "standard" FSF assignment
> > form, it's designed to assign copyright of the entire existing
> > Work, not just future changes.  That's why they require that the
> > Assigner(s) be sole copyright holder(s) for the Work.  Was that
> > requirement met?
> 
> I'm going to have trouble answering that, because that is not how
> I remember the assignments working.  The way I remember/understood it
> was that the assignment simply allows the assignee to claim copyright. 
> The code is essentially forked at that time, with the original author
> retaining copyright in one forked branch, and assignee is allowed to 
> claim copyright on the other. Insofar as these two copies are the 
> same, the net effect is that the original author retains copyright,
> and FSF can claim copyright as well.  I do not remember any 
> provisions for sole authorship.
> 
> I am not a lawyer; the above is my recollection of what I thought 
> I'd signed six years ago.

Ok, I'm starting to understand how this looks from your perspective.
Of course I have no idea what you signed with the FSF, but I suspect
it was intended to do something very different from what you've
described.  The FSF has used various assignment contracts at different
times and for different purposes, but AFAIK (the FSF copyright clerk
would be authoritative) they've all had a few things in common:

1. By "Assignment" they mean "Transfer of entire right, title and
interest", not "Relicense".  That means the FSF becomes the copyright
holder and the Assigner is retains _no_ copyright.  This is usually
made very clear by statements such as "hereby transfers to the [FSF]
its entire right, title, and interest (INCLUDING ALL RIGHTS UNDER
COPYRIGHT) ..."  This is pretty essential to what a copyright
assignment is, so I'd be surprised if your contract differed much.

 [ Just to emphasize that first property, the contract usually
specifies that FSF grants specific rights (not copyright) back to the
Assigner. ]

 [Note2: In a legal context, "transfer" is a "conveyance of right,
title, or property, either real or personal, from one person to
another, whether by sale, by gift, or otherwise." [Webster, 1913] It
doesn't allow for "here, you have this right, and I'll keep it for
myself, too." ]

2. They apply to all past _and future_ changes to the Work.  It would
be pretty pointless not to include this in the contract, since
otherwise, future changes by the Assigner would create a derived work
of which the FSF would no longer be the exclusive copyright holder.
This is also why you don't have to continually "reassign" copyright to
the FSF, if you already have an assignment contract "on file".  It
covers future changes.

and most importantly...

3. They are designed to make the FSF the copyright holder of the
_entire_ Work.  They accomplish this my requiring the Assigner to
warrant that he/she (I could imagine an assignment contract designed
for a group of copyright-holders, but I've never seen one) is the
_sole_ copyright holder for the assigned Work.  Without that part of
the contract, the assignment would lose one of its key benefits -
enforcement of copyright against infringement.  (See
http://www.gnu.org/licenses/why-assign.html ) In U.S. courts, holding
copyright to only some portion of the infringed Work, doesn't really
help.  (Indeed, this is one chief motivation for copyright assignment
and it applies to the FSF no less than to any other entity.)  This is,
of course, exactly why the FSF _requires_ copyright assignment for
code contributed to projects that FSF already holds the copyright for.
The inclusion of non-FSF copyrighted code would jeopardize the very
purpose of the assignment.

> >         Robert Merkel claimed that no copyright had been assigned on
> > July 6, 2001.  (His last contribution was in May, 2001.)  Was that
> > true?
> 
> I would have to dig through a file cabinet in the garage. Although
> different people signed these at different times, and some had to be
> nagged, I was under the impression that everyone signed them. But 
> maybe not.

Well, neither one of you seems to have a firm recollection of the
event, but I suspect it doesn't make much difference anyway...

> >         More to the point, if the FSF had been assigned copyright to
> > all of GnuCash in 2000, why did you and others continue to mark files
> > as copyright held by yourselves?  
> 
> Because that is my understanding of how the assignment works.

That does explain it.

> > The whole thing seems very messy,
> > especially since GnuCash has been actively developed since then by
> > developers that never signed an assignment contract.
> 
> Why should this matter?

Well, usually, FSF copyright assignment is followed by rather strict
policy enforcement designed to ensure that FSF remains the sole
copyright holder, usually by having contributors file their own
assignment contracts.  GnuCash obviously never did this, so FSF is
clearly not the sole copyright holder now, even if they ever were
(which seems very unlikely.)

> >         I'm pretty confused by the implication that FSF became the
> > copyright holder for GnuCash in 2000, 
> 
> They would not be the sole copyright holder, they would be one among
> many. There are maybe 20 or 30 or more copyright holders, depending 
> on how you wish to treat small patches.

That's become very clear.  What's less clear is why the FSF would want
non-exclusive copyright to a Work.  (See again
http://www.gnu.org/licenses/why-assign.html ) I would guess they
either expected that they would later become the sole copyright
holder, or that they thought they were doing so immediately and just
weren't very careful.

It seems to be of very little practical difference to GnuCash today,
but the remaining open question in my mind is what did the assignment
contracts actually accomplish?  If they had no severability clause and
warranted something about exclusive copyrights, I suspect they
accomplished nothing, (disclaimer: IANAL, blahblah), which is exactly
how everyone has behaved.  Otherwise, it's possible that FSF retains
the copyrights to some portion of GnuCash, which would seem to be of
no practical benefit to either the FSF, the assigners nor anyone else.

Bottom Line:

Given the insignificance of the practical difference it would make,
I'm inclined to take the existing copyright notices at face value, and
to not change any existing copyright notices unless the actual
copyright holder (whomever or whatever that may be) asks for incorrect
copyright notices to be corrected.  At the present, that's equivalent
to assuming that the FSF holds no copyright to any part of GnuCash,
which is exactly what I thought a week ago.  :)

-chris

<snip> 
> --linas
>