---------------------- Forwarded by Cliff Morgan/Chichester/Wiley on
10/02/98 11:27 ---------------------------


Cliff Morgan
10/02/98 11:18

To:   [log in to unmask]
cc:
Subject:  Re: Relation Position: Revision 1  (Document link not converted)

Comments below (indicated by double angle brackets).





[log in to unmask] wrote:
>
> Re reference to "globally-unique identifier" below: shouldn't it also be
> persistent/permanent? For example, a URL is a URI but it's not permanent.
> Should we specifically refer to URNs or PURLs instead? Or, to be non-Web
> specific, DOIs?

I think the above shows a misunderstanding of DOIs that it would be best to
nip in the bud.  At the moment, DOIs, despite the generic-sounding name
(digital object identifier) are anything but generic.  They are part of an
identifier/resolution system initially developed under the auspices of the
Association of American Publishers and now run by an internation membership
organization.  This system is not open to anyone, and there are charges for
obtaining prefixes and for registering quantities of specific identifiers.
The system was developed as a tool to aid in rights management and
e-commerce, and there is some discussion of limiting the scope of
assignment to tradable content objects (that is, loosely, to things for
which publishers may want to manage rights).  DOIs are also, currently,
closely tied to the Handle System, and can only be resolved by the a single
resolver run by the International DOI Foundation.

None of which is a problem for that particular application. However, it is
a mistake to think of the DOI as a universally applicable name.

Priscilla Caplan

>>All that you say is fair enough, but I don't see what "misunderstanding"
of mine you are
correcting. I didn't claim that the DOI is "universally applicable", in the
sense that everybody can use it to apply to any object. All I would claim
for the DOI is that it is a unique and persistent number - it's not the
only number that necessarily has these properties. I did mention URNs and
PURLs as well. The only point I was trying to make is that the URL, or any
other location-dependent and temporary identifier, is likely to give
interoperability problems in the future, and that the RFC should at least
refer to (if not express a preference for) permanent numbers. ISBNs were
also referred to in the RFC, but these aren't permanent because they change
with change of ownership. Although the DOI contains a Registrant Prefix,
which can, for example, identify the publisher at the time of registration,
the number is effectively "dumb" since the object keeps the prefix even if
ownership changes.


>>There needs to be some way of identifying objects, and the DOI system
provides a mechanism for assigning, registering and managing DOIs. There
also needs to be a way of pointing to the current location of the object,
or of information about the object, and CNRI's Handle system (which was
developed to deal with resolving URNs to current URLs) gives us this. As
publishers, we want a well-regulated system, and we are prepared to pay for
this through registration and loading charges - the fact that not everyone
will want to pay for the maintenance and development of such a system
doesn't invalidate what publishers are trying to do.

>>I can quite understand that Dublin Core doesn't necessarily want to get
embroiled in what Stu Weibel has called the "morass" of digital
identifiers. I just think the Resource Identifier piece ought to recognise
some of the options, even if it's not going to lobby on behalf of any one
system.

>>Cliff Morgan