On Sun, 2005-06-12 at 15:47 -0400, Lex Spoon wrote:
> Jim Gettys <Jim.Gettys at hp.com> wrote:
> > Lex wrotes:
> > Thus, I'm much more interested nowadays in what
> > > the likes of Lawrence Lessig have to say than in what Debian,
> > > OSI, or FSF have to say.  If these latter organizations want to stay 
> > > relevant in the area of open-source ideas, then they need to adjust to 
> > > the changing times.  Like it or not, real law hacking requires real law
> > > hackers, not computer scientists who have taken a class or two on law.
> > 
> > Hmmm...  I will note that Lawrence Lessig is on the board of directors
> > of the FSF.  There is some serious legal talent on board.
> 
> Glad to hear, and I hope this goes somewhere.  RMS is great at advocacy,
> but Lessig knows law and has a great reputation in intellectual circles.
>  It could be a great team if they find a way to work together.
> 
> debian-legal is an entirely different story.  The "consensus" agreements
> it comes up with are driven by amateurs.
> 
> >And at least in the case of the GFDL, I think Debian has made the right
> > call (I'm not very familiar with the other cases).  The interaction of
> > trademark law with the "invariant sections" part of the GFDL can cause a
> > document to become unusable in other contexts, presenting a raft of
> > problems, either for reuse of code, or when forking a project, which
> > sometimes becomes necessary.  
> 
> I agree on a strictly theoretical level, but these differences don't
> have enough practical impact that Debian should toss all of that stuff
> into its non-free section.   Most people just aren't going to load up
> their documents with invariant sections, so Debian's policy means that a
> great deal of free documentation is getting tossed into "non-free". 
> Similarly, while I agree that all the GNU licenses cause problems with
> intermixing, that was an intentional feature that RMS added.
> 
> >From a wider perspective, notice that there is a vast difference between
> Microsoft-L and any vaguely free license you can name.  Despite this
> vast difference, however, groups like FSF, Debian, and OSI are spending
> tremendous efforts quibbling about border cases among different
> open-source licenses.
> 
> 
> > One of the biggest issues right now is the fact that there are too many
> > licenses, many/most of which are not well thought through.  The
> > combinatorial explosion when mixing code is becoming intractable, and
> > many people are working to reduce the number of open source licenses in
> > use.
> 
> 
> This is truly  a problem.  On the other hand, it is not the end of the
> world even in the extreme case that code from different projects cannot
> be intermixed at all.  The individual projects can always continue to
> move forward.  And usually, it is allowed for components to speak over
> socket connections no matter what kind of licenses they have.  While
> this is not a nice solution, it means that incompatible licenses are not
> usually a complete show stopper.

It causes major waste of effort, and the license complexity actively
inhibits progress in many circumstances.

> 
> Thus, I would say a bigger problem is that the *legal systems* aren't
> set up to handle open source software.  There is such a thing as public
> domain in copyright law, but nothing AFAIK corresponding to open source,
> where you have zillions of authors and the goods are given away for
> free.  If I want to public-domain something, I can just say it; even if
> I say it incorrectly, the courts can figure out what I meant, and I have
> a good chance of causing the legal effect I was going for.  For the long
> term, we ideally need the same thing for open-source software.

No one will touch code without a copyright declaration.  If you want to
give software away, choose the MIT license as the shortest and sweetest
declaration of that sentiment (the BSD license is a bit more verbose,
and the old versions of the license have the infamous advertising
clause, so I recommend the MIT license).  That's what we drafted it for
(I was somewhat involved in its drafting).


> 
> So I think more open-source activists should be working on things like:
> publicizing the issues to wide audiences, showing why the general public
> should care, and ultimately getting legislators to look at it and move
> towards a new international copyright treaty.  Digital copyright is a
> hot topic right now around the world, so now is a good time to strike.

We are, every change we get.

> 
> In the short term, I have started public-domaining my code and
> documentation.   It makes no difference for me in practice to using a
> specialized open-source license, and I avoid a lot of headaches and
> possible future problems.  Even though open-source licenses have some
> nice strings included that I might like to enforce, they make no
> practical difference because:
> 
> 	1. There are loopholes, and I don't know where they are until these
> licenses get vetted in court -- which isn't going to happen in the near
> future.  Witness the ICS chess software, many branches of which are
> closed source despite descending directly from GPL code.
> 
> 	2. It's all a bluff anyway, because I will never bother to sue. 
> Witness Interval's hacking of Squeak without releasing source
> code--blatantly illegal (so I'm told) but no one cares enough to sue.

Yes, if there is no interested party, you have no one with standing to
sue.

In fact, in the GPL case, there have been many enforcement actions  by
the FSF and others.  As the MIT/BSD license permits any additions you
like, it doesn't end up in court; about the only thing that might come
up is violation of people's use of their names and/or organization's
name without permission.

Only a few violations required going to court to get resolved.  Most are
resolved amicably behind the scenes with no publicity.

So far, the GPL/LGPL has stood up cleanly enough there has not been any
serious litigation; the licenses are in fact holding up much better than
many people expected.

Personally, independent of license, I think a situation where there are
more than one copyright holder in a larger set of code is more likely to
end up with enforcement (since there are more people with standing to
sue). 

> 
> 
> So, I think that hackers may as well avoid open-source licenses all
> together for now, and that activists should spend a larger portion of
> their time trying to adjust copyright law itself.  Obsessing about our
> differences is causing more harm than good at this time.
> 

No, here I *strongly* disagree. Code without copyright no one will
*touch*.

Remember, copyright law is at the basis and serves as the protection of
open source and free software.

Many people are actively trying to reform copyright law.  So far,
Hollywood and the RIAA are winning (at least in lengthening copyright
terms toward infinity).  The big issue is around orphan works: works
that are not being marketed, and for which it can be impossible to even
identify the owner any more.  In most ways, copyright law is much less
broken than patent law.

As the consequences of interactions between licenses are *not* well
understood particularly by most programmers, I do strongly urge anyone
(depending on circumstance and inclinations), to choose one of the big 3
licenses: MIT or BSD rev 2, LGPL, GPL.  Note I leave the GFDL off that
list.

These licenses are well known, and interoperable. Some other open source
licenses have real problems.  The OSI is working on both reducing the
number of recommended licenses, and resolving problems between them, as
best they can (sometimes licenses have terms that will prevent mixing of
code in various ways).


				Regards,
					- Jim Gettys