Re: Copyright notices. Was Re: Some general questions!

Karl Auerbach (karl@cavebear.com)
Tue, 2 Aug 94 13:02:29 PDT


Thanks for your comments. The following is a fairly long reply because
your questions triggered some ideas.

> >No problem. I'll just do it my way. You will do it yours. Fred will do
> >it yet another way.
>
> That's why the registration of a MIME type (which would probably be
> more of a "license" than a "copyright", but whatever) would be a
> useful way to standardize it. Such a system should be generalizable,
> of course; if it only works with HTTP and HTML it's not very useful.

The model I have in my head is possibly (probably? ;-) different from
yours.

I think of a server which has a published "work", be it a graphic or
text or sound or whatever. Since not all "work"s can contain an
embedded copyright, my mental model is that the server will somehow
contain an ancillary block of information about the use of that
"work". When the server sends the "work", the server will make sure to
send the ancillary information as well.

This isn't really so different than what is done today with http in
the object metainformation.

I'm rather wary of having this kind of information stored in another
document and reached by a link because of the chance that the link
will eventually be broken or at the time the link needs to be used,
it's server will be off-line.

Rather, I think that this kind of "right to use" information is of
sufficient importance that it needs to be available at all times
and the best way to do that is to put it in the meta information.

I'm not saying that this is easy. I would hardly want to put the GNU
copyleft into the meta-information. (If I did, I would be wrong. And
I think the copyleft is itself copyrighted with copy/modification
restrictions.)

Whatever we come up with (if anything) ought to occupy no more than a
few lines of meta-header.

I'm trying to figure if there is a way we can come up with some kinds
of classes. Most of the stuff we have will go under something which
would be classed as "freely available and usable, but don't claim
authorship for yourself" category.

Another class might be "freely available, but distribute no copies"

What I am struggling with at the moment is the distinction between
the original copy being taken and any subsequent copies which may be
made from that first copy. (I'm thinking [perhaps incorrectly] that
a copy made via a cache is a first-level copy.)

It makes sense to have a URL which points to a license document so
that the user can know what he/she/it is getting into if he/she/it
redistributes. But I don't think it is realistic that a user is going
to read a license document before getting the first copy.

So I'm thinking of a two tiered meta-information expression:
- Stuff the client ought to know when taking the first copy. This
information should be readily available in complete form in the
meta information.
- Stuff the client ought to know if he/she/it wants to use,
republish, or modify the "work." This information can be obtained
via a reference -- there is no need for instantanous access.

> >2. One can't put the copyright into the link because anybody can create a link
>
> This is dependent on how things are configured. It would be quite
> trivial to design things so that not just anybody can create a link
> with the relationship value of "copyright."

Ah another place where, as usual, I didn't really way what I meant.

I was thinking of people putting copyright information in the URL used
to reference a work rather than a link from the work itself.

But your comment did trigger an idea that perhaps an alternative
solution is that publishers of restricted works ought to publish a URL
to a cgi script or something that forces the client to read and agree
to any limitations before passing on "the work".

> >5. Information that needs to be conveyed to express the copyright:
> > a) owner and date
> > b) terms (i.e. the license) under which the user can view and possibly
> > further copy the document.
> > c) fees (possibly many classes depending on the status of the user)
> > d) agent for collection of the fees
>
> And potentially other stuff, which may grow arbitrarily complex and
> arbitrarily large. Do we really want big documents like the GPL or
> whatever other licenses will be used being passed around in HTTP
> headers? No, we just want to pass around pointers to them. Much more
> efficient, as well as more general.

Agreed, big things need to be handled by reference. But I think I can
make the argument that most of those sort of heavy licenses really apply
to the copies which you might make to your original copy. I.e. what
I was calling second tier access.

(And, of course, I'm not dealing with the fact that copyrights and
licenses are handed differently in each jurisdiction.)

--karl--