For Approval: GPLv3

Donovan Hawkins hawkins at
Tue Aug 28 03:53:54 UTC 2007

On Mon, 27 Aug 2007, Chris Travers wrote:

> In the US, and many other countries, we have an originality standard for 
> copyright.  This means that original expression may be copyrighted, but that 
> functional elements (ideas, facts) may not.   Furthermore in derivative or 
> collected works, any copyrights for the work as a whole only extend to the 
> original expressive elements added by the new author.  Thus, if I include an 
> excerpt of a translation of poetry in a book on Norse mythology, I need the 
> translator's permission to do this, but if I am granted that, I do not gain 
> copyright ownership over that excerpt as included (I ran into this, couldn't 
> get permission, and did my own translation as a result).

That's my understanding of copyright law as well. If you did get 
permission, you would have copyright over the rest of the book but not 
over the quote you used from that translator. You could stop anyone from 
reprinting the book without permission, but could not stop anyone from 
stealing that quote because it isn't yours.

> I don't think anyone will argue that the same applies to software.  If I 
> include BSDL code into my program, I do so under the permission fromthe BSD 
> License.  I do not acquire any copyrights to the code I include by doing so.

How is this different? If you take some code (the quote) from a BSDL 
program and put it in your GPL program (the book), you have copyright on 
your code but not the BSDL code. You can enforce how your code is used but 
not how the BSDL code is used.

But then, you don't need enforcement power to issue permissions. You can 
tell your readers "You may use this book" when it contains another 
person's quote because that person gave you permission to do that. 
Similarly, you can tell people "You may use this program under GPL v3" 
when it contains BSDL code because that person gave you permission to 
distribute that (and most any other) way.

Keep in mind that all licenses are permissions. In no way does the 
existence of a license place restrictions on you...if it did, you would 
have to consent before the license could be issued. GPL v3, for example, 
is the permission to use the program in a very specific way. That specific 
way is phrased in terms of a ton of "restrictions", but they are 
restricting you from doing things you couldn't have done anyways without 
the license. On the whole, the existence of GPL v3 on the program simply 
grants you a complex permission. BSDL grants you a greater (and less 
complex) permission.

So if you distribute your derived work under GPL v3, you aren't telling 
anyone "you are not allowed to release binary-only copies of this". You 
are telling people "here is a permission to use this which is contigent on 
not releasing binary-only copies". GPL v3 specifically allows other 
permissions to exist and does not limit them, regardless of whether they 
are "Additional Permissions" included with the GPL v3 license or simply 
other permissions that happen to exist (such as you might get by 
contacting a GPL author directly and negotiating a deal for usage in a 
closed source program).

To modify the earlier example we had going: you told people "You have 
permission to dance and/or sing". I told them "you have permission to 
dance while not singing." I technically granted a permission to dance a 
certain way, not a restriction to not sing.

(Of course, that recent court case involving the Artistic License being 
treated as a contract mucks my whole argument up, but let's set that aside 
at least until the appeal is over. Speaking of which, has the EFF or SFLC 
been asked to help out with that yet? I hope they don't let this go down 
to defeat without filing a few briefs at least.)

> Now in either case, the work as a whole needs to be licensed under the GPL. 
> We can all agree on this.  However, the basic matter is that I can only 
> enforce my own copyrights-- I cannot enforce copyright infringement on 
> elements that are not original to me or have had copyright transferred 
> appropriately.  Hence even if I "only leave the notice of the license and 
> relicense under the GPL" the original elements of the code which have not 
> been substantively altered do not become, as a practical matter, encumbered 
> by the GPL simply because I do not have rights to enforce it.

I agree completely. Practically speaking, nothing has changed in terms of 
what someone can do.

> For this reason, I have *no* idea what it means to, in the process of 
> conveying, remove additional permissions (section 7, second paragraph).  Such 
> a removal of permissions would either be meaningless (I try never to read a 
> contract or license that way if I can help it, and if this *is* the correct 
> reading, one runs into truth-in-advertising issues), it could be beyond the 
> scope of the copyright license (in which case it might be compatible with the 
> MS-PL), or it could be in conflict with every license out there by mandating 
> controls over copyright permissions that nobody else has the right to 
> mandate.

I forget if GPL has a severability clause, but if it does then the latter 
interpretation wouldn't hurt anything because that would simply not occur.

For the most part, the only affect that removing an Additional Permission 
really has is to do a bit of housecleaning on LICENSE.TXT, prevent 
possible confusion from people who think the permission applies to all the 
code, or allow someone to not have to mention by name a license they don't 
approve of.

I think the FSF was concerned with people using Additional Permissions and 
Further Restrictions to mutate the GPL into a different license in a 
permanent way. The ability to remove Additional Permissions and ignore 
Further Restrictions ensures that any copy of GPL v3, no matter how 
mutated, can be restored to the original GPL v3 in the next generation. 
They were afraid of someone coming in and releasing their derived works 
under GPL v3 plus a bunch of absurd permissions/restrictions that would 
keep any other developer from touching it, making it a worthless 
contribution to the community. Keep in mind that the author of that 
derived work would NOT be bound by those additional terms, so he could 
poison the well without drinking from it himself.

As a good example, imagine the "Additional Permission" to use all future 
derived works under BSDL. Would it prevent others from enforcing GPL on 
downstream derived works? Even if it didn't, it would probably scare off 
most developers from trying if not from the trivial ability to nuke the 

Donovan Hawkins, PhD                 "The study of physics will always be
Software Engineer                     safer than biology, for while the
hawkins at                   hazards of physics drop off as 1/r^2,                biological ones grow exponentially."

More information about the License-discuss mailing list