[License-review] Sublicensing

[Engel Nyst]

Jim,
Your argument on strong copyleft seems entirely backwards.

Disclaimer: the following is personal opinion. I think I have
statistically contributed more to permissive software projects, and I
license share-alike in writing. I don't make comparative statements of
value on free/open licenses, by personal policy. However it is fair to
say that I have a preference for strong copyleft.

Regarding these statements:

On 09/12/2014 11:48 AM, Jim Wright wrote:
> I understand the perspective that as long as you're distributing the
> source and notices to everything, few will complain, but that doesn't
> mean no one will - and beyond that, if we foster an environment where
> copyleft requirements are not subject to strict compliance as long as
> all the source is released, maybe that's ok, but it seems to me this
> may cultivate a culture which weakens the strength of how copyleft
> licenses are interpreted and implemented generally, which will bleed
> over into commercial use as well to the extent it hasn't already. And
> if you're a fan of strong copyleft, this might not be a preferred
> outcome.

On 09/12/2014 03:20 PM, Jim Wright wrote:
> the reason *I* want to expressly permit this is to be crystal clear
> that the license facilitates use in both (i) strong copyleft code
> (potentially requiring licensing originally UPLed code under the
> copyleft terms)

I use in the following 'copyleft' in the sense of strong copyleft,
copyleft to the extent of copyright in a work, or close to that extent.

Copyleft exists to counteract copyright, it's a tool to restore the
liberty of the public to deal in works of authorship, as if copyright
exclusivity wouldn't be there, except strictly in the measure necessary
to keep it that way to any recipient of works based on them. To put this
in perspective, I think of it in the context where copyright law is, in
a sense, in flux: courts interpretations change what was understood and
relied upon before their new decisions, and the direction seems to be
that copyright and related rights will be more expansive. It has
happened before. It has happened in 2014.

If, in 2020, copyright will restrict *more* the freedom of people to
deal with works of authorship, than it does in 2014, then copyleft will
need to at least contemplate addressing it, in its implementation and/or
its interpretation.

*If* copyright can and will destroy the common sense understanding in
dealing with unmodified copies of BSD/MIT/AL2.0 works, by some future
court decision, then copyleft may contemplate addressing it, in its
implementation and/or its interpretation.

But it's not the other way around.

To use copyleft to argue that copyright law 'must' be more invasive than
it is, that it 'must' be interpreted to erase FLOSS licenses on
*unmodified copies*, is entirely backwards. That's simply *not* common
sense, not what GPLv2 strictly *says*, not the way people usually work
in free and open software projects, and not helpful to anyone except a
traditional proprietary middleman.

Copyleft to the extent of copyright needs to guarantee that each and
every recipient has all rights to use every part of the covered works,
without proprietary restrictions, no matter if the parts are derivative
at their turn or not. While I don't think GPLv2 or A/GPLv3 texts are
ideal implementations of this concept, none of them relies on
'sublicensing' / 'relicensing'.

On the other hand, I subscribe to the view of other commentators in this
discussion, that the argument for sublicensing in FLOSS licensing or
that all non-copyleft FLOSS licenses can be 'erased', is only to the
advantage of a traditional distributor of others' works. I note
proprietary licensing doesn't need it either. You don't need to
'sublicense' in order to make proprietary derivatives of a permissive
work. Only some assumption of the traditional publisher that they 'must'
claim to be the ONLY licensor for authors' works to subsequent
recipients, seems to me implied here.

Copyleft has no need for that. On the contrary. Most free and open
source projects, copyleft or not, work on this basis: each author gives
a license to the other authors and everyone else in the world, a license
which gives them all legal rights to copy, modify, build upon, transform
and adapt the work, use it in other works or have it used by them, as
long as the recipients fulfill its conditions.

Perhaps it's only me, but I don't see anything right now, in my layman
understanding of copyright law, that could possibly 'stop' the flow of
rights from BSD or MIT or AL2.0 or MPL2.0 licensors on an *unmodified
copy* of their work[1] to all recipients, purely under copyright. None
can be exactly 'erased'. That seems correct about most or all currently
approved OSI licenses.

In this context, you're proposing a license which claims to be 'clear'
that... it's not clear. That it can get to people and not mean what it
says. That it depends on something else, some other license, external
to the software and copyright in the software, *by design*, which other
license will 'really be' the license people get. For unmodified copies
in source form, even.

For OSI approval, I think this doesn't pass OSD 7.

Arguably, it doesn't pass the other criteria either, since the rights it
seems to give don't depend on it.


In the larger context, my concern is what can happen, is that some
interpretation of this odd concept of sublicensing is steered along a
way that may, one day, try to make cases like C and specially D[2] into
copyright infringement cases. Or throw more doubt on people's rights to
use an *unmodified copy* of a free and open work, a copy licensed
sufficiently clear for them to read and know their rights.

If copyright can really go there, copyleft may be able to cope with it,
but I don't think this is an academic question, and I don't see how is
it a good thing in the first place.


[1] assuming everything else is equal, i.e. source code available, no
proprietary parts in a binary, no calls to otherwise-licensed code, etc.

[2] C and D from the hypo in a previous email, which I copy below for
convenience:

> Lets say A makes a software work, which consists of three long files
> implementing some complex algorithms. A licenses it as BSD, and
> distributes it.

> B takes the three files, doesn't modify them, adds a file interacting
> with them, and distributes them together in a proprietary package.
> Lets assume they're all written in python or javascript, and B
> distributes them in source form.

> C downloads B's package and accepts the license agreement. It says
> the user accepts they can't make derivative works, distribute, in
> original or modified form.

> One day, C opens and reads A's BSD-licensed files from their machine,
> and remembers their friend D is interested in those algorithms. C
> gives D a copy of the BSD files.

> (1) Did C do something wrong *under copyright law*?

> (2) Did C break the agreement with B?

> (3) D copies one of A's files on their machine, makes modifications,
> complies with BSD and distributes the result publicly. Does D
> infringe copyright?

> (4) Does D break the agreement with B?


-- 
  "Excuse me, Professor Lessig, may I ask you to sign this CLA, so we can
*legally* have your permission to remix and distribute your CC-licensed
works?"
  ~ Permission culture, take two.