Message-Id: <199310180008.AA22022@rock.west.ora.com>
From: Terry Allen <terry@ora.com>
Date: Sun, 17 Oct 1993 17:08:01 PDT
To: uri@bunyip.com
Subject: Re URNs & dead publishers
Robert Raisch writes:
> As I mentioned in private mail to Larry, I do not think that a document
> whose publisher in no longer valid should be supported, unless that
> document is placed in the public domain. A document without a publisher
> (owner) cannot be guarenteed in any sense, since the guarentee of identity
> is a function of the publisher.
Unworkable. I may have an archive of material published by people who
vanish. Am I then to throw out this material because I'm not the
authoritative source of its meta? I grant there is a very difficult
problem here, but this is not its solution.
I must have some way to verify that a document instance is the same as that
provided by the publisher of the document. What that way is, I don't know.
But I know I need it.
>So, if the publisher is the maintainer of the meta-information associated with
> the document, (where does it live, who maintains it, etc.), and the
> publisher goes out of business, then the document is no longer maintained
> by anyone.
> Who is the authority of the information of and about the document if it
> has no publisher? Without any authority, there can be no guarentee of
> identity.
> I believe that this presupposes that there is some authority which maintains
> the meta-information for documents in the public-domain. Volunteers?
I don't think public domain has anything to do with it.
Take the model of libraries. There is no authority that provides
ISBNs for books published before the ISBN was invented, yet they can
still be determined to be legitimate, and may still be under
copyright. Such books are legitimated through their publication
characteristics (t.p. info, number of
pages and plates, binding, dimensions), as recorded by an army of
bibliographers over the years and published in bibliographies and
catalogues. Detecting forgeries as opposed
to variant editions is harder, and more a matter of connoisseurship,
given that books are physical objects. But---here's the important
point---it doesn't require a central authority to do this, only
widely published information about the books concerned.
What can we do with electronic files to develop legitimating
information that does not depend on the publisher vouching for a
specific instance?
> I do not think that there was ever any assumption that the URN be a
> permanent identifier. Rather, a URN must never be reused.
> </rr>
>
> On Sun, 17 Oct 1993, Larry Masinter wrote:
>
> > It had been my impression that one of the design requirements for URNs
> > was that that names of objects have to be PERMANENT -- that you can
> > make a reference in one document to another document that will
> > `stick', and not change even though the second document moves, its'
> > publisher goes out of business or whatever.
> >
> > Perhaps I misunderstood? This hasn't been in the list of requirements
> > or differences between URNs and URLs that have been discussed so far.
>
-- Terry Allen (terry@ora.com) Editor, Digital Media Group O'Reilly & Associates, Inc. Sebastopol, Calif., 95472