The position RMS takes...

John Cowan cowan at
Wed Mar 29 04:34:31 UTC 2000

W. Yip scripsit:

> Please correct me if I am wrong, but I have the impression that only one
> who owns the copyright can issue a license, hence in the case of
> conventional licenses, the licensee cannot sublicense without express
> permission from the licensor, since the licensee does not own the
> copyright?

Conventional licenses are in no way non-exclusive copyright licenses.
I have just read over several Microsoft EULAs, and I think they are typical.
They expressly forbid copying, distributing, and the making of derivative
works.  They also forbid many other acts while permitting others.

The EULA has all the earmarks of contract, except perhaps for actual
agreement, and that is the defect which UCITA is intended to cure.

The GNU GPL, however, is a non-exclusive copyright license; as such,
no one but the copyright owner can issue it, so there are no sublicenses.
However, since the GPL is a general permission to everyone, there is no
need for any sublicensing either.

> Not so if there is liability for defective software, particularly if the
> software was sold as fit for a particular purpose.

Again, all licenses known to me (free and proprietary) disclaim all warranties,
including those of merchantability and fitness for any purpose.

> >> Thus, I
> >> would challenge the wording of the OSD on the grounds that it seems to
> >> claim that rights subsist 'in rem' , that is, in the program itself.
> >I think clause 7 (to which I suppose you refer) *is* badly drafted.
> On the contrary, I think it was intentionally drafted this way by Perens to
> directly challenge conventional understandings of copyright licensing. As I
> said, this should get interesting if it ever reaches a court. The notion of
> 'rights attaching to a program' (as opposed to a person) is a novel one in
> law, but is entirely consistent with the Open Source creed.

I guess we won't know until Perens tells us.

> Is it still 'in personam' when the license is unilateral, made to all? I
> question what practical difference there is between a unilateral license
> and public domain.

Plenty.  An *unconditional* public license like "Anyone may copy, distribute,
publicly perform, and make derivative works, without restriction" might be
taken as a dedication to the public domain.  But the various free licenses
are conditional: they permit anyone to copy, distribute, etc. etc. only
if certain conditions are met.  It is not clear to me if these conditions are
to be taken as conditions precedent or not.

> 'lawful acquisition' is another troublesome phrase. This seems circular
> because what is 'lawful' almost certainly depends on the license being
> valid?

In part.  But acquisition of a copy by way of a crime, for example,
would clearly not be lawful acquisition, so the circle can be broken.

> There are two possibilities to the 'automatic license'. The license either 
> (in a case where A[copyright holder] licenses to B licenses to C)
> (i) automatically issues from the licensee(B) to the sublicensee(C)
> (ii) automatically issues from the licensor (A) to the sublicensee (C)

Has to be (ii) for a non-exclusive copyright license, since B can't
license anything by way of copyright.

> The GPL claims (ii) following s. 6 'Each time you distribute...the
> recipient automatically receives a license from the original licensor
> ....'. This is troublesome because A is entirely *outside* contractual
> privity when B distributes to C? How does the automatic license emanate
> from A who is entirely outside the picture where there is distribution from
> B to C?

A is the copyright owner, the only one who can issue such a license.
The GPL, when adopted by A as a license for his copyright property, says
that C gets such a license.

> s.6 GPL entirely removes all control from A (original licensor and
> copyright holder). This is because HIS ASSENT IS NO LONGER REQUIRED FOR
> LICENSING. Thus there cannot be a contract between A and C, even though the
> GPL mandates that A will give the license automatically. But there can be a
> contract + license between B and C.

A contract, certainly (e.g. the contract of sale between Red Hat and me
for GPL-licensed software copyrighted by the FSF).  A license of some
sort, perhaps.  But a non-exclusive copyright license?  No.
Red Hat (B) has no right to issue such a license.  The FSF (A) licensed me
(C) to copy/distribute/etc. as a consequence of Red Hat distributing to me.

> So there are potentially concurrent licenses involved in the GPL?

I'm not sure: concurrent in what way?

> Don't you find it circular that the GPL allows distribution but yet enables
> infringement through distribution?

Distribution that violates the condition is infringing.  Since you
don't like my previous example, suppose I license you to distribute
my book, on condition that you include an erratum slip with it.
Distribution without the erratum slip would be infringing.

> I'm not quite so sure myself, actually. It seems to me that one is either
> given permission to distribute or not to distribute. Having been given
> permission to distribute, I think it impossible to infringe copyright,
> particularly if the license can be seen in isolation from the act of
> distribution of a copy.

Why can't permission to distribute be conditional?

John Cowan                                   cowan at
       I am a member of a civilization. --David Brin

More information about the License-discuss mailing list