Date: Tue, 22 Feb 1994 09:38:37 -0800 (PST)
From: David Robison <robison@nwnet.net>
Subject: Re: Unresolvable URNs --> ISBNs (discussion, not resolution)
To: timbl@www0.cern.ch
In-Reply-To: <9402211453.AA01922@ptpc00.cern.ch>
Message-Id: <Pine.3.89.9402220944.H27291-0100000@norman.nwnet.net>
On Mon, 21 Feb 1994, Tim Berners-Lee wrote:
>
> David Robison <robison@nwnet.net> says:
> > A correction is in order here: ISBNs are not opaque. They are made up of
> > 4 discrete sections: the first represents language/country; the second is
> > the publisher number; the third is the book number (simply and ordinal);
> > the last digit is a check digit.
>
> What do you mean my "opaque"? They are used by most people without the
> knowledge that you have. They are pretty opaque to me :-)
Are you implying that not all people are librarians? ;-)
> When you order a book by ISBN, does the bookshop look up
> the list of publisher prefixes? Is there a resolution algorithm
> well defined?
There are directories of publishers with ISBN indexes (GK Sauer, Books in
Print), but I suspect many bookstores simply manage their stock using
ISBNs and don't care about the publisher until it comes time to reorder.
I think the only algorithm involved is the one that produces the check digit.
> An interesting point is that you can't determine where the boundaries
> lie between the fields. So the ISBN authorities can cook up
> flexible (classless a la CIDR) allocation spaces without worrying the man
> in the street. But as you say, they dropped it when it came to
> the fixed length. (They just chose 10**9 instead of 2**32 :-)
> I guess variable-length bar code technology wasn't there at the time.
>
> Tim BL
Perhaps if librarians, rather than pubishers, had cooked up the scheme it
would be (more) scalable.
David