David Chan <[EMAIL PROTECTED]> writes:
> On 2 Mar 2000, Mark Seaborn wrote:
>
> > If someone loads a Mudela file into Sibelius, modifies it there,
> > and distributes it, in your scheme they only have to release the
> > Sibelius file. But the Sibelius file is of no use whatsoever to
> > most people.
>
> This is true. However, somebody somewhere with a copy of Sibelius
> will probably save the new file back into Mudela (or equivalent),
> and stick it up on the Web.
But you could equally well say that someone with a music recognition
program will probably scan the music in and make the Mudela output
public. In fact, once there exists free music recognition software,
this is more likely than the Sibelius case.
> > What if someone used a completely secret program to produce their
> > derivative work? They could release the machine-readable source
> > to the music, but it would be meaningless.
>
> Yes. However, this argument works with software too. I can write a
> secret compiler which compiles a secret language which is an
> extension of C. Then I can take a GPLed C program, modify it using
> my language's extensions and then distribute binaries; the new
> source code is now useless to everyone apart from me.
I'm not sure you can. The GPL says:
``For an executable work, complete source code means all the source
code for all modules it contains, plus any associated interface
definition files, plus the scripts used to control compilation and
installation of the executable. However, as a special exception, the
source code distributed need not include anything that is normally
distributed (in either source or binary form) with the major
components (compiler, kernel, and so on) of the operating system on
which the executable runs, unless that component itself accompanies
the executable.''
The use of the word `scripts' is rather vague, but I would contend
that that phrase *does* cover the compiler. Otherwise there would be
no need to add the `special exception' that exempts C compilers, etc.
But in your example, the compiler for your secret language is secret,
so it can't fall under the special exception.
> I suspect that for either music or software, it would take quite a
> lot of effort to design a whole new language which was complicated
> enough that people couldn't "crack" it easily, and write a program
> to compile it, so hopefully people wouldn't bother.
I disagree. There is no inherent meaning in information.
[snip]
> > And sheet music *is* machine-readable anyway. There are music
> > recognition programs around.
>
> But the same applies to printed program code; it's still not as good
> as having the electronic version in the first place. OCR scans are
> not infallible yet, and it would take quite a lot of effort to
> obtain a source file as good as the one which was used to create the
> printout.
But this is just an incidental problem, not an inherent problem. Some
day OMR will be good enough that it hardly gives any errors. A free
music licence will have to last, and not contain outdated
restrictions.
The effort in OMR'ing some music will, in general, only have to be put
in once if the person who does it makes the Mudela code public. And
it's certainly easier than reverse engineering some random format.
Don't forget that the availability of the original Mudela code on the
Internet through Mutopia will be a substantial disincentive against
producing derivative works without source. A sourceless derivative
work would get ignored in favour of the original, so why would the
author of the derivative work deliberately not release the source,
particularly when he's not allowed to make the derivative work
proprietary?
> > Consider someone who buys some nicely-bound *free* music from a
> > publisher. She does not have a computer.
>
> I hope this will happen a lot in future!
So do I!
> > She received a disc containing the Mudela source code for the
> > music, but it's totally useless to her.
>
> Hold on, I'm not suggesting that all users should *automatically*
> get the disc, just that they should have the *right* to obtain it if
> they want. This is exactly how the GPL works; boxed Red Hat doesn't
> come with source code, just a note mentioning the ftp site.
Yes, and this is still a problem with music.
Say a teacher photocopies a load of free music for her class. If the
music has a distribute-the-source condition, and she doesn't have a
computer, she can't do this for fear that the students will become
awkward and demand the source.
> > What happens if she wants to photocopy it for a friend? She
> > can't, because she can't copy the source code.
>
> This could work like section 3c of the GPL; a noncommercial copier
> could just show her friend the small print which says how the source
> code can be obtained.
Why should commercial copiers and non-commercial copiers be treated
differently? Why should commercial music copiers be required to have
computers? You become a commercial copier even if you sell
photocopied sheets at a few pennies each.
> > And why should the publisher distribute the source on a disc? Why
> > not a CD-ROM, or punched card? Obviously with computer software
> > the source can come on the same medium as the binaries; this makes
> > sense. But this can't apply to music.
>
> Well, the GPL doesn't require that the source is available on the
> same medium as the binaries.
But it does effectively require that the source come on a medium that
you can read. If the source came on 8 inch floppy discs, you could
reasonably claim that this was a violation of the GPL, because no-one
can read 8 inch floppy discs any more.
> It has to be "on a medium customarily used for software
> interchange". I think we could use that definition, but with
> "electronic data" instead of "software".
But the analagous term, ``on a medium customarily used for music
exchange'', is paper!
[snip]
> > > One of the biggest practical advantages Mutopia users will have
> > > is the access to the source code.
> >
> > People will still be able to get the Mudela source code from the
> > Internet.
>
> They'll be able to get our original source, but not derivatives.
> E.g. they might have access to the source for my new quartet, but
> not to the source for the brilliant trio arrangement that someone is
> selling in the shops. (This example is hypothetical, and will
> probably remain so!)
But anyone can copy the arrangement that is being sold in the shops if
the music is copylefted.
> > If the piece is copylefted, people will still be able to scan the
> > sheet music it, run it through a music recognition program, and do
> > whatever they like with it.
>
> Yes but music recognition is not in a fit state to do this effectively at
> the moment. Maybe one day it will be trivial to do it, but right now you
> have to spend a lot of time making corrections to the interpreted version.
>
> Also, If I had (say) a score of Beethoven's Ninth Symphony, the
> scanning process could take hours. If I just wanted to print out a
> transposed version of the Bb clarinet part then I could do it in
> minutes with the source code, so it can be significantly more
> convenient.
You're talking about convenience, but a licence should be based on
principles.
The GPL doesn't require that software be *easy* to fix, just that you
have access to the same information required to fix it as everyone
else. For music, a licence shouldn't require that music be easy to
tranpose (for instance), just that it be possible to do so.
Binary-only software isn't merely inconvenient to fix, it's
impossible, because essential information is missing. But scanning
music is just inconvenient, a mere incidental difficulty.
What incentive is there for a creator of a derivative work to not
release source when he knows that all it takes is one person to make
that source freely available, if the music is copylefted?
--
Mark Seaborn
- [EMAIL PROTECTED] - http://members.xoom.com/mseaborn/ -
``She never loses a match. If she loses a match, it's because
her opponent beats her'' -- Pam Shriver
