Date: Sat, 16 Oct 1993 17:19:45 -0700 (PDT)
From: "Rob Raisch, The Internet Company" <raisch@internet.com>
Subject: URNs and Meta-Information - The Value of ISBN
To: uri@bunyip.com
In-Reply-To: <199310162356.AA17193@rock.west.ora.com>
Message-Id: <Pine.3.03.9310161744.A11995-c100000@hmmm.internet.com>
I was thinking about this again this evening on my way back from The
Middle East -- a marvelous Arabic style restaurant in Cambridge.
What value is there in the ISBN (International Standard Book Number)?
(Did I get the acronym right?)
Correct me if I am wrong, but I understand ISBNs to work thusly:
The ISBN Authority assigns a set of numbers to a publisher called ISBNs.
The ISBN, itself, has a Publisher Identifier and a Product Identifier.
The Publisher Identifier is assigned and recorded by the ISBN Authority.
The Product Identifier is assigned by the Publisher and recorded by
the ISBN Authority.
And I assume that the ISBN Authority also records meta-information
about the publisher and its products.
So, we have a number of roles here:
The ISBN Authority guarentees the uniqueness of the Publisher Ident.
The ISBN Authority caches meta-information about the Publisher
and its products.
The Publisher guarentees the uniqueness of the Product Identifier.
The Publisher caches the authoritative meta-information about itself
and its products.
So, The ISBN Authority does really three things, it:
1. assigns a unique identifier to the publisher
2. guarentees that all ISBNs are in a defined format
3. acts as a central repository for certain meta-information
about the publisher and its products.
Now, let's look at this in terms of the Internet:
There is already an authoritative source of unique publisher
identifiers: The Domain Registration Authority.
By defining interoperable standards which are used to communicate
with information services, we guarentee the format of the identifier.
The only reason for a central authority for meta-information is
if there is a varying cost in retrieving the information from a
variety of sources. The cost of retrieval is equalized when there
is a point to point communications infrastructure like the Internet.
For the above reasons, I believe that ISBNs are a VERY BAD model for what
we wish to accomplish with URNs. The only value which the ISBN brings to
the discussion is that it is a unique identifier for a publisher's product,
but the management infrastructure which ISBNs use doesn't fit.
I strongly believe that the publisher is the only authoritative source for
the meta-information we need, and that even in the ISBN model, it is the
publisher who is responsible for the uniqueness of the ISBN, not the ISBN
Authority.
I believe that this suggests the following model for URNs:
URN -- This is a URN
':'
location of the server which contains the meta-information
':'
unique identifier which when provided to the server, retrieves
the meta-information
as in:
urn:publisher.com:0010929292
I do not believe that multiple name spaces buy us anything. There is
already a namespace which guarentees the identity of the publisher (The
NIC) and how the product_identifier is interpreted should be entirely the
responsibility of the publisher.
Supporting ISBNs implies some centralized ISBN Authority. Supporting IANA
registered products implies that we need to contact IANA for information
which is the responsibility of the publisher.
Perhaps I still don't fully understand the problem.
</rr>