[You didn't honor my M-F-T so I guess this will continue to go to both
lists.]
On Thu, Feb 07, 2008 at 12:29:29PM -0800, Russ Allbery wrote:
> Branden Robinson <branden@debian.org> writes:
>
> > I believe your reasoning is faulty, because it is based on incomplete
> > information. There was more than one "BSD" license in use well before
> > USB's Office of Technology Licensing withdrew the 4-clause version.
>
> [snip]
>
> While this is very interesting (I was aware of some of this, but not all
> of it), and I appreciate the time that you took to write it up, I think
> that:
>
> http://web.archive.org/web/19990210065944/http://www.debian.org/misc/bsd.license
>
> shows that indeed the original BSD license to which the DFSG was linked
> was the four-clause version. (Thanks to Charles Plessey for uncovering
> that.)
>
> The version in /usr/share/common-licenses/BSD is very specifically the UCB
> version,
A major point of this whole discussion is that there is no "the UCB
version". There have been multiple BSD licenses, even promulgated by the
single source we call the University of California at Berkeley.
> not any of the other versions, and my assumption was that that had
> historically also been the case (since it wouldn't make sense to me to
> move from a less specific copyright holder to a more specific one).
[...]
> Certainly agreed; however, I was specifically talking about the UCB
> version as seen in /usr/share/common-licenses, so I was really being
> inaccurate with my original statement.
The copyright line in /usr/share/common-licenses should be made generic, or
better yet, not even be present. Much of the benefit of the
common-licenses directory is lost if it can serve as a stand-in for
particular licenses *as applied by particular copyright holders*.
> > (I have heard rumors that the OTL was in large part persuaded to drop
> > the advertising clause because of threatened counter-litigation by a
> > party that was violating it, who made an apparently strong argument that
> > the clause was unenforceable under U.S. law. Unfortunately, despite
> > poking around for this over the years and talking to some luminaries who
> > might have been aware of it--though not William Hoskins himself--I have
> > been unable to substantiate it. If this turns out to be true, Debian
> > should not be recommending as a best practice licensing provisions which
> > are legally void significant jurisdictions like the United States.)
>
> Note that I have never argued that Debian should be recommending the
> four-clause BSD license as best licensing practice. It manifestly isn't.
> Only that it is and has been DFSG-free since the beginning of the concept.
First, I think you are reading far more deliberation into where the Debian
Project has pointed web links in the past, and what it's put into
/usr/share/common-licenses/BSD, than is warranted.
If I were to write some code and license it under the BSD license (in the
terms spelled out in /usr/share/common-licenses/BSD), package it, and have
my debian/copyright file refer to /usr/share/common-licenses/BSD, that
would not mean that the Regents hold the copyright on my code, nor would
such an action on my part transfer the copyright to them.
Secondly, phraseology like "is and has been" ("and will be for all time!"
usually follows in arguments like this), denies the very real phenomenon
that humans learn over time.
It would not surprise me if a majority of Debian Developers in 1997, if
surveyed on the subject, would hold the 4-clause BSD license to be
DFSG-free (with degrees of passion ranging from "yeah, I guess so" to
"hell, yeah! It's way better than that GPL crap![1]").
I would suggest that our experiences with the GNU FDL, and with the
XFree86's projects relicensing of its code base, have taught us just how
onerous mandatory invariant testimonials can be. While some folks may feel
that Debian was an outlier with respect to our dissent on the GNU FDL
front, it's pretty difficult to make that argument about the revised
XFree86 license, whose resemblance to the 4-clause BSD license is much more
clear. (In fact, that was one of David Dawes's ultimately futile arguments
for trying to get the community to accept his license as free.)
If I'm not mistaken, I have argued on -legal in the past that having
section 10 of the DFSG has turned out to be a bad idea, because people
misread "examples" as "paragons".
I think it is instructive that every single license we identified in 1997
as a good example of a free software license has seen significant revision.
The 4-clause BSD license has evolved into 3-clause and 2-clause variants,
dropping various restrictions; the Perl folks came up with a Clarified
Artistic license several years ago, and of course there is the case of the
GNU GPL v3.
Moreover, these license exemplars have been revised *by their original
promulgators*. Consequently, I do not think you can argue that the
supersession of the licenses we originally identified as examples in 1997
is the work of upstarts who don't understand the meaning of Free Software,
or who are trying to subvert or pervert it the concept[2].
Larry Lessig once told me (more or less) that there's more to interpreting
the U.S. Constitution than putting oneself in the mindset of Virgina
plantation farmer with a powdered wig, a muzzle-loader, and access to
high-speed horse-drawn or wind-powered transports. The interpretation we
bring to its words today also matters.
The community has demonstrably learned something about Free Software
licensing over the past ten years. I'd like to believe that we have, too.
[1] Whatever happened to Alex Yukhimets?
[2] Granted, people say essentially that about RMS and no doubt will
continue to do so long after the man is in his grave, but you *don't*
hear them saying that about Jordan Hubbard, William Hoskins, or Allison
Randal.
--
G. Branden Robinson | To believe is very dull.
Debian GNU/Linux | To doubt is intensely engrossing.
branden@debian.org | -- Oscar Wilde
http://people.debian.org/~branden/ |
Attachment:
signature.asc
Description: Digital signature