ATT SOURCE CODE AGREEMENT Version 1.2C

[John Cowan]

Bruce Perens scripsit:

> On Fri, Sep 10, 1999 at 12:14:45AM -0400, John Cowan wrote:
> > Unfree-1: The ASCA is a "shrink wrap" license
> 
> I think the words "shrink wrap" and "tear-open" refer to the fact that the
> license is entered into by performing some sort of action, such as opening
> a package, installing the program on a computer, or making _use_ of the
> source code to build a program.

Just so.  This is quite different from the GPL, LGPL, BSD, MIT, and QPL, which
are unilateral licenses of the form: "I, the author, grant you certain
of the rights I have under copyright law, conditional on your doing
such and such."  You need do nothing and agree to nothing, but if you
violate the conditions, the copyright laws prevent you from modifying,
distributing, etc. etc.

But you still have many other rights as an owner of a copy of the copyrighted
work.  For example, you can study it to learn things.  But the ASCA
keeps ownership (not merely copyright-ownership, but full ownership)
of every copy in AT&T's hands, and gives you *only* four enumerated rights:

o	to distribute the original bits

o	to compile the source code and execute it on a computer

o	to package and distribute patches and additional bits with
	the original bits

o	to compile and execute the above package

The right to print, the right to display on a monitor, the right to read,
the right to study: these rights are *not* granted.  This is unfree.

>  All Open Source licenses that I know of are "shrink wrap" or
> "tear-open" licenses. That appears to be essential for OSD compliance.

In the sense that they require no agreement, yes.  In the sense that they
retain ownership (not merely copyright-ownership) of the copy you have,
and license you to use it only in specified ways, no.

> > that is, it purports to restrict the *use* of code, rather than restricting
> > only copyright rights such as distribution and modification.
> 
> Can you restate this in terms of the "fields of endeavor" clause of the OSD,
> where the field of endeavor is not in itself a violation of copyright
> law? Please note that copyright law allows restriction of public performance.

I don't think this is a "fields of endeavor" issue.  Rather, it is an issue
of who owns the copy.  When you have a copy of typical open-source software,
you own it and have unlimited dominion over it, except for those rights
that are 1) reserved by law to the copyright-owner and 2) not granted to
you by the owner.  When you have a copy of typical proprietary software,
the license carefully tells you that you do *not* own it, and you have
only the specific rights granted by the license.

I did not intend my list of the rights that copyright owners control to
be exhaustive.  I merely mean to indicate the fundamental difference
between a conditional *copyright* license such as most OSI licenses
are (I'm not clear about the MPL), and a conditional *use* license that
retains ownership of all copies in the hands of the creator, such as the
ASCA does.

> > In particular, you may not reformat the Capsule from "tarball" format
> > to another generic or platform-specific format.
> 
> I think this is OK with ATT after the last round of changes to the license,

Well, not in the version you posted, which only allows distributing the Capsule,
rather than the unmodified Source Code (as the QPL does).

> which made it clear that a tear-open license could be used on the source code.
> One of them commented to me today that they were no longer restricted to the
> odd capsule format (which probably presents a license before you extract it,
> and that's why it was necessary). I'll bring this up with them.

Good.  The file mentioned in your posting of the license had the extension
tgz, suggesting that it is an ordinary gzipped tarball.

> > Likewise, you may not print out the source code and distribute the printed
> > version to a colleague for his or her inspection.

You omit this point, which is a variant of the right to copy the unmodified
source code and distribute it, this time in printed form.

> > Unfree-2: Because the ASCA restricts use, the user may not take
> > advantage of the statutory rights of someone who owns a copy of a
> > copyrighted work.
>  
> The OSD does not require that fair use as defined in copyright law be
> preserved. I have some question regarding whether other Open Source licenses
> preserve it, and whether such preservation is always in our interest, as it
> impinges on our ability to provide "viral" protection as in the GPL in some
> cases.

Fair use of a published work is a public right that supersedes any licenses.
The license for a typical book (not an FSF book) says "no copying at all",
but fair use permits copying small portions with the intent of public comment,
for example.

The ASCA, however, unfairly denies this right of the public by purporting to
impose a contractual obligation against it.  As I have said before, I believe
such a contract to be unenforceable, and all but one of the court cases
agrees with me.

> Their restriction on how others use their web site seems to be outside of the
> scope of the OSD. I remain unconvinced that presenting someone else's web site
> in a frame is fair use, anyway, and do not see that you would have this right
> had you not entered into this license.

I won't dispute the point.  As I said in my preamble, unfairness and
GPL-incompatibility certainly do not disqualify a license as Open Source, but I
think they are of interest to the Open Source community, especially
when they are not obvious (as in the "strict construction" clause).

> > Unfair-2: If you obtain a Derived Work from a third party incorporating
> > the licensed code, and the third party breaches the license either
> > as explained in Unfree-1 or otherwise, then your license to use the
> > Derived Work is void, even if you neither knew nor should have known
> > of the third party's breach, and even if the breach had nothing
> > to do with the product you are using.  This is a consequence of Section 8.5,
> > and is manifestly unfair.
>  
> Can you suggest fair remedies?

The GPL preserves the rights of innocent third parties in this case.
Why can't the ASCA?

> > Unfair-3: The statement in Section 8.6 that the ASCA is to be
> > interpreted "without any strict construction in favor of or against
> > either AT&T or you" looks even-handed, but it is not.  The ASCA is
> > what is called a "contract of adhesion", one which is drafted solely
> > by one party to it who then imposes take-it-or-leave-it terms on
> > the other party.
> > 
> > The ordinary rule of law requires that such contracts are construed
> > strictly against the party who drafted them: they may not say "We
> > didn't mean what we said".  This clause attempts to unfairly compel
> > licensees to sacrifice this right.
>  
> It sounds fair for me to ask them if this is their intent.

I suspect that this clause is in any event void by reason of being against
public policy, but I know of no relevant case law.

> > Un-GPL-1: The ASCA is incompatible with the GPL,
> 
> Of course GPL compatibility, much as _I_ would like it, is not a stated goal
> of the OSD. In fact, the OSD would not have been necessary were
> GPL-compatibility a goal, we could have just said "it's free software only
> if it's compatible with the GPL". In this case, GPL compatibility wasn't
> ATT's goal, either. Thus, while I'd _like_ these changes, I don't see much
> chance for their acceptance and can't sincerely state that they are
> essential.

I agree that this point is not essential, but the un-GPLness is a direct
consequence of the attempt to control use.  If this unfree misfeature
was removed from the license in favor of QPL-style terms, the problem
with the GPL would be greatly lessened, perhaps to zero.
BTW, we still don't have Appendix A.

> Removing the viral nature of a license in order to achieve GPL compatibility.
> Surely you see the irony here.

Yes and no.  The GPL's virus compels the creator of a derivative work
to make it free.  The ASCA's virus compels the creator of a derivative
work to make it unfree (because the creator must impose a use license
on the end user).

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