On Mon, Mar 26, 2012 at 09:38:30AM -0400, Clark C. Evans wrote: > On Sun, Mar 25, 2012, at 01:36 PM, Ben Finney wrote: > > > I think this is a false assumption, the service itself required > > > creativity to implement, and the specific choice of word associations > > > in specific contexts is not algorithmic nor factual, but individual > > > calls by translation submitters who have granted the translation > > > service license to use their work. > > By the same argument, the GCC copyright holders can claim to hold > > copyright in every program compiled using GCC, and the copyright holders > > in PHP can claim copyright in every web page that program renders. I > > think that's exactly as unsound as the argument you present. > I don't think your comparison is sound. A compiler is going from > one synthetic anguage with well defined semantics to another. A > human language translation isn't simmilar at all. > If you insist on assuming that the output is indeed the result of > a mechanical compilation and if you presume the final result is > GPLv2+, then the process must comply with Clause #6 of the GPLv3. > This requires the corresponding *source code* needed to _generate_ > the work from its source code be compatibly licensed Hence, if > you want to compare this process to the GCC case... the translator > itself must be released under the GPLv3. Not in the least. Releasing something under GPLv2+ means the recipient gets to *choose* which version of the GPL they're complying with, including when they create derivative works. In the GPLv3 only case, I think there's also still room to maneuver; even though the translation is initially a mechanical translation, once done, doesn't this translation then become a new part of the *source*, subject to hand editing and revision? If so, I don't think it falls under section 6. > I think presuming the translation output isn't copyrighted is > wishful thinking, valuing *your* copyrights over the copyrights of > others. Copyright is defined as attaching to creative expressions. If there's nothing creative in the way the translation is *expressed*, it's not copyrightable; nor is there recognition under copyright law that the output of a machine is covered by any copyright other than that of the input. (BTW, if you're claiming that there's creativity in the machine translation itself and that copyright attaches, that's pretty much mutually exclusive with claiming that it's "object code" under GPLv3 section 6.) > For example, if the chunks arn't copyrightable, and assembling chunks is > just an automated process... why is the source document copyrightable, > it's just a series of uncopyrighted facts (e.g. words). If the sequence > matters, then you're not just sending chunks to the service, you're > sending the entire document. US copyright law recognizes that there may be creative expression in the selection and organization of factual information. This is why a phonebook (or a timezone database!), which has a trivial structure of organization and is intended to be exhaustive, is not recognized as having a copyright, but an anthology of selected short stories has an editor's copyright in addition to the copyrights of the individual authors. So when you choose what bits to feed into the machine and how to assemble the output, the new copyright that attaches is yours, assuming that the choosing and assembling is non-trivially creative; the machine doesn't hold any copyright. -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developer http://www.debian.org/ slangasek@ubuntu.com vorlon@debian.org
Attachment:
signature.asc
Description: Digital signature