[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Critique of draft GNU Free Documentation License v1.0



On Sat, Jan 08, 2000 at 12:28:33AM -0700,
	Richard Stallman wrote [quoted by > and indented 1 space]:

      [Sections indented 5 spaces were written by David Lawyer]
>     INVARIANT SECTIONS ETC.
>     I think that the "Invariant Sections" and "Cover Texts" parts should
>     be eliminated.  The problem is that they can be abused.  One may put
>     any non-technical material here and be assured that it will appear
>     years later in modified versions.
> 
> That is the whole point.  We use invariant sections like these in GNU
> manuals already--for example, the GNU Manifesto in the Emacs
> Manual--so any license we write has to provide for them.

Why not then have this as an option in the license?  One option would
insure that such sections remain invariant and would be used for the
GNU documents that need them, etc.  LDP might suggest use of the other
option which would make no mention of invariant sections.  LDP authors
would be able to use the "invariant" option if they wished since it
conforms to the requirements set forth in the LDP Manifesto.

> 
> Where this license differs from our previous approach is that it
> permits people who modify the manual to add their own invariant
> sections if they wish.  I think it is not right that only the first
> author can do that.
> 

If someone makes only minor modifications, I don't think they have a
right to an invariant section, especially if they use it for
advertising.  I think the simplest solution is just not to have them.
Perhaps there is a way to word the license so as to prevent abuse but
it will not be easy to do.

>     In order to make a license contained in a document very small in
>     size I would suggest that each document contain a very short
>     license that points to a url for the full text.  Such a license
>     might read:
> 
> That is a very unwise idea.  The license for a work should always be
> completely included in the work; otherwise, there is a danger that
> they will get separated, and people who have copies will not be able
> to be sure what she can legally do with them.
> 
> Referring to a URL can sounds like a good idea if you are thinking
> of a year's time span.  But we have to think of the long term (some
> GNU manuals have been in use for over a decade).  We cannot count on
> any URL to remain valid for that long.  We cannot count on URLs to
> *exist* a decade from now.

Most documents that are of current interest will hopefully be
maintained.  The maintainer of the document should change the URL if
that should become necessary.  There is a danger of URLs becoming
invalid for unmaintained documents.  To resolve this problem I think
the license should require that the license would be attached to the
document whenever the document becomes unmaintained.

Since the part of the license always attached to the document (as I
proposed) would allow copying, it could be copied as a historical
document say 50 years from now even if the full license never did get
attached to it.  Another way to search for the license would be to
look for those (historical ?) documents that did get the license
attached.

> We will certainly not recommend that people separate the license from
> the document, and I hope that the LDP will not recommend it either.
> 
>       For any modified version I would require
>     that a transparent copy be placed at a well known Internet site (such
>     as LDP's with a large number of mirrors) for free distribution.
> 
> Part of the criteria for free software is that the license should not
> require people to notify any specific place, or publish in any
> specific place, when they exercise their rights.  I think of that as a
> matter of privacy, a rejection of central monitoring.

I think that a free document that anyone can distribute to the world
in in no sense private.  The original author probably wants any
modifications that are distributed to be widely distributed.  For the
case of modifications that are for personal or internal use only
within an organization, then I think that they should be exempt from
having to post it on the Internet. 

> Thus, we will certainly not include such a requirement in a license
> for the GNU Project to use.  I hope that LDP authors writing about
> parts of the GNU/Linux system will not use such requirements either.

>     E. Preserve all the copyright notices of the Manual.
> 
>     My comment:
>     If only a small portion of the "Manual" has been used in the 
>     "modification" the previous copyright notices should not need to
>     be included.
> 
> If you use only a small portion, that is fair use, so the license is
> irrelevant anyway.  You always have fair use rights, no matter what a
> license may say.

But suppose you want to use an entire chapter (which may be a small
portion).  Fair use would not allow this.  In fact just lifting a few
paragraphs would not be fair use unless you quoted them and attributed
them to their author.
> 
>     My comment:
>     As a document is modified, it goes through various versions and in
>     some cases the document could be modified weekly.  It would be
>     impractical to have pointers to all the old versions.
> 
> The license does not require you to include the pointers for *your
> own* old versions.  So in practice, when a team releases versions
> frequently, it will not cause this kind of practical problem.

I think you're right since this would be the case if the 2nd
modification you do is considered to be based not on the first
modification but on the original that you started with.  I don't think
this concept is clear to most readers of the license and probably
should somehow be explained in the license.

> I added a statement that the publisher of a version can give you
> permission to remove the pointer to that version.

I'm still not clear who the "publisher" is for documents that are not
printed on paper but are just posted at websites.  Most of the HOWTOs
are like that.  Even when a HOWTO is printed, there will likely be no
pointer to the published version since it will likely be identical to
a certain electronic version not controlled by the publisher.  So the
publisher in this case has no say in the matter.


--  
To UNSUBSCRIBE, email to ldp-discuss-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org