On 3/20/07, Andrew Gray shimgray@gmail.com wrote:
[This is mostly me blowing off steam, but there's an actual serious proposal in here too...]
On 20/03/07, John Vandenberg jayvdb@gmail.com wrote:
<snip>
Having the ability to verify works means volunteers can chase down the
ones that look a little sketchy. And rapidly double check the ones that do have the linkies.
But the thing is, once we've verified the bibliographic data is correct, we're done. End of story. We no longer need the catalogue reference, in the same way we wouldn't need a note saying "this book is borrowed by Sally, remember to add the publication details when she brings it back".
Basically, this is my objection. The Amazon link, or the LoC link, or any other link, is *not a source*; I mean, really, what are we verifying? That a work by this name and with this metadata appears in an online catalogue. Nothing more. It does not verify the assertion that this is a relevant work, or that it is one we used. It is simply what we use to sanity-check the metadata in our own list of sources. Once we've done that check, we've confirmed we've spelled the name right and got the year correct, we don't need to keep that link or that catalogue code; it's an internal editorial reference, and it can be kept on the talk page or junked, but it's unneccesary and somewhat misleading to keep it with the citation.
ISBNs are a special case - we keep them around because MediaWiki can do cunning magic with them, and we faintly hope we can hack something similar together later for LCCNs. But it would not materially harm the intellectual integrity of our listed references to quietly lose every ISBN overnight, because they're icing and not cake.
Hmm. I strongly disagree with this; keeping the ISBNs around means that it's easier for other people to find the work in question. In a variety of ways, yes, through mediawiki magic; but even without the link, people could search for the book with an ISBN manually in all the places listed on special:booksources.
The point of listing sources is *not just* to verify facts in an article. It's also to give readers a list of other places to look for more information on a topic. ISBNs, or any other unique identifier, help tremendously with that. They help you verify that the edition or translation of the work that was meant in the article is, in fact, the one that is likely to be found by readers. It is often very difficult to search online library catalogs properly and find what you're looking for, especially when you don't have familiarity with the work in question; ISBNs help.
In addition, listing the ISBN makes it easier to verify and re-verify a reference if you suspect vandalism. It is, in fact, important to continue to have the quick sanity check available that a work was cataloged somewhere, that it actually exists under the title claimed, etc., and that the subject headings match what the topic claims. Library catalogs can also tell you fun things like the promising-sounding work quoted is actually a work of fiction. Thus I'd argue that being able to find a work in a library catalog is important; and ISBNs help with finding. Unfortunately, of course, they also don't exist for all works.
<snip>
...which prompts a thought. Could we expand the citation templates and have default-non-displaying fields, like the hack we have with persondata? This would allow us to silently include all the catalogue identifiers we could wish for, BNB codes or BNF references or LCCNs or OCLC codes or, hell, even LibraryThing work IDs - without the objectionable issues of including and displaying unhelpful links (or the limitation of only including one ID number). We could then fiddle it with different div IDs so that people can select in CSS to display ISBNs or ESTC numbers or whatever their little heart desires...
(ESTC is probably one of the rare counterexamples to "no comprehensive identifiers before 1960s", but I don't see us citing too many of its works...)
An interesting idea... it'd be nice to improve the citation templates; but it would also be nice to get them in wider use. They're pretty complicated as it is.
<snip>
If we end up with a large number of ASINs that cant be replaced with a
better identifier, a repository like WorldCat may come along and create verified records for them all, simply because they are used on Wikipedia.
This isn't how cataloguing works, sad to say. You can't create a verified record without the book in your hand; if they had the book in their hand (well, stock), it would have been catalogued eventually regardless...
(WorldCat isn't a repository. It's a union catalogue of thousands of libraries. I suspect none of them are the Institute for Philanthropic Bibliography...)
Indeed. WorldCat is a merged catalog of most [not all] of the works in its member libraries, which are worldwide, but in general the consortium is prettily heavily US-biased.
WorldCat also has problems.
I have strong views against WorldCat as well, but that's another rant
- vast quanties of very sloppy cataloguing, and a misleading
presentation as being substantially more comprehensive than it is. (Remember, you have to pay to join the consortium...) I would argue (almost) as strongly against arguing we should routinely include OCLC id codes, or whatever it is they use; I got very het up recently when someone wanted to spam the Worldcat "author pages" across Wikipedia.
Now... also as a librarian, I am actually strongly in favor of using WorldCat links and OCLC numbers, when available, *especially* as a substitute for ASINs and other commercial unique identifiers. It's only recently that the entire WorldCat database was made freely available at www.worldcat.org, but we should take advantage of it. True, member libraries still have to pay to join OCLC, and it does appear more comprehensive than it is, but it's also, to my knowledge, the best free source of global cataloging data we have -- and the motives behind the OCLC projects are far more in line with our own that those of commercial booksellers.
And while the cataloging may be sloppy, for older books the cataloging is still tons better than what you'll find on most commercial sites, including Amazon. For non-ISBN books, the OCLC record will at least a) tell you what the book is; b) show you where it's held in the world; and c) link you to the LCCN, for an extra bonus -- all of which is better than you'd get with no number at all, if you are a regular reader who doesn't know that worldcat exists. And for all of the books with OCLC records, worldcat will tell you the nearest free place that has a copy of the work -- in other words, your public library. Even if it's not perfect, this seems to me a tremendous service that we should promote, in line with distributing knowledge to all people and all that.
<snip>
I would discourage calling it a "source", then... "further reading"
can cover a multitude of sins.
I'm in favor of strict source sections and liberal further reading sections, myself... and I worry when the two get confused :P
-- phoebe