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. > 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 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. > A program is not much use without a computer. [...] > Music is different. You don't need a machine to process it. I agree, that does make a difference. > 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. > 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! > 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. > 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. > 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. 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". > > We could specify something like this, e.g. : > > If the work is generated by a machine, then you must [offer to] > > provide, at a nominal fee, the source code which the machine uses to > > generate the work. > > How do you determine whether the derivative work was created with a > machine? Well, first of all, anything in printed format will be machine-created. Nobody does manual music typesetting any more. For manuscripts, we must assume a machine was not involved. I doubt anyone would type out a mudela file, only to write the whole thing out manually afterwards just to hoard the source. Few people would pay money for music written out by hand. > > 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!) > 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. Regards, David
