BSD-like licenses and the OSI approval process

[Chris Travers]

On 10/12/07, Lawrence Rosen <lrosen at rosenlaw.com> wrote:

> That sort of a silly abstract question. For one thing, I could trivially
> avoid the whole issue in the world of permissive software licenses by making
> a non-trivial expressive change (such as affixing my beautiful logo to your
> BSD-licensed work) to create a derivative work.

And in that case, provided that the combination is creative rather
than functional (I don't know if a court would let one get away with
that if the sole purpose was changing a license, but you could still
enforce your own copyrights and trademarks).

> Furthermore, the fact that I
> enable someone to drill down to and copy a specific page in my collective
> work doesn't make the collection itself any the less a copyrightable
> collective work.

Nor does it make that page subject to your copyright license.  In
fact, I am not sure it says *anything* about the license of any given
component.

Consider the following example:

Suppose I create your hypothetical collected work.  Suppose I include
in it GNU Readline.

WAIT I hear you cry-- Readline is under the GNU GPL, you can't do
that!  I believe I can and I will tell you why.

The GNU GPL (any version) does not apply to collected works absent
some derivative element (mere aggregation clause).  Since Readline
would be distributed as its own independant component, the GPL doesn't
yet apply to the collected work provided that the my license of the
collected work does not force you to violate the GPL.

Secondly, even works which link to GPL libraries don't have to extend
the restrictions of the GPL to their own elements-- licenses can
comply and grant a superset of rights.  For example, see PostgreSQL's
use (as a BSDL program) of GNU Readline.  The catch as far as
PostgreSQL goes is that you can't include that support in your
closed-source spinoff (you can use any other code you like, however).
Since the AFL doesn't prevent further restrictions on the parts of
dependencies or components, there is little reason one couldn't add
closed-source elements as well provided that they could be
redistributed in the form you chose to distribute them on the site.

> That person's copy of the individual page was copied from
> my work that he received under the terms of my collective work license, AFL
> 3.0 (to which he assented!).

I.e. contractually agreed to.  Which would be entirely *separate* from
the question of copyright license.

> Finally, in my case I'm not intending to assert
> any copyright interest specifically in the page he copied by drilling down,
> because it remains under the BSD license--as well as, under the authority of
> the BSD license--under AFL 3.0.

I am confused by your phrase "under the authority of the BSD
license--under AFL 3.0."  Earlier you suggested that the primary
authority was that of contractual agreement.  If it is dual-licensed
under the authority of the BSD License, that would seem to assume that
the BSD License is divisible (i.e. an individual is free to sublicense
only some of the rights granted to downstream recipients).  I can find
no reason to make such an assumption given the relevant case law on
divisibility of nonexclusive licenses and the fact that, unlike the
AFL, the BSD License is not explicitly divisible.

Note that indivisibility wouldn't preclude an implied or stated
sublicense right (i.e. a right to enter into a sublicensing
relationship with a sublicensee could still exist if the sublicensor
was required to provide the software under the exact terms of the
original license).  But it would mean that all rights licensed must be
passed on as a bundle.  I believe that the MIT license by addressing
all downstream users and requiring the permission grant to be
reproduced verbatem essentially allows for sublicensing but is
indivisible.

>
> My copyright interest, as you suggest, does not extend to your BSD-licensed
> work. But I read your BSD license as giving me permission to distribute your
> work under any license I choose, including AFL 3.0, as long as I copy the
> text of your BSD license in the source code.

Ok, so you are interpreting the BSD License to be implicitly
divisible.  I.e. you are suggesting that a verbatim copy can be
distributed with only a subset of the license rights granted to
downstream recipients.  IANAL so I may be missing something but it
seems to me that:

1)  Most BSD-style licenses address *all* downstream recipients and
2)  Case law in the US seems to side consistently against divisibility
as regards nonexcluive licenses, and the 9th circuit even ruled a few
years back (Gardner v. Nike) that exclusive licenses were indivisible
by default.

If there is something which overrides these, I sure don't see it.

> Where's the beef in your entirely hypothetical question? Legal questions, no
> matter how "essential" to ponder, don't exist in a vacuum. Would you take
> such a question to court? What's the legal dispute?

My problem is that I don;t think your system actually says anything
about the license.  For example, I see no reason why a GPL component
couldn't be merely aggregated into such a collection.

Secondly, this sort of question *has* come up in public before (the
recent Linux/OpenBSD flap), so I am not at all sure that there aren't
people who would take it to court.

Best Wishes,
Chris Travers