On 23/10/2012 19:45, J. McRee Elrod wrote:
Thomas Brenndorfer said:Perhaps the biggest frustration I get in these discussion is the conflation of issues. A discussion of controlled vocabulary terms shouldn't be bogged down by display issues.Display issues? The function of 245$h or 33X would seem to me to facilitate discovery of desired resources. It would seem to me that how they are displayed is central to their serving their purpose. The major problems we see with 33X is that some terms are too long for convenient display, are redundant, or are obscure (e,g,,"tactile three dimensional object"; "object" would suffice). Relator terms have the same problems. It is not necessary to include "film" or "sound film" in relator terms; it is clear from the record what is directed or composed; "director" or "composer" would suffice.
This has turned into an interesting thread. In an ideal world, display can be rather unimportant so long as the information is input consistently. Information that is consistent in a computer can display in almost any way someone would want. So, if the text says “mediated” or whatever is beside the point. It is similar to arguing whether a computer code in the 008 field should be “1” “9” “z” or “§”. It really doesn’t matter. It’s only a code.
The moment inconsistency is introduced, the task of display becomes far more complex. So for me, the question of what a cataloger actually enters into a 33x field is rather unimportant: the computer can display it–or not–however you want. Yet, we should not ignore that this also concerns consistency with what is in the *totality* of the database, that is: what the public works with every day–not only the newest records–and this in turn brings up the issue of the incorrectly termed “legacy data”. This however, is a topic few catalogers seem to want to discuss, although the public will see it in *every single search* until the end of time. Not a minor concern, I think.
At the same time, from a theoretical point of view, the traditional GMDs really have conflated different aspects of an item, and this can be demonstrated clearly, as has been shown with particular clarity in the examples of [electronic resource]. I am sure we have all wrestled with this in our own practice.
An auxiliary point is the idea of turning our “text” into “data”. Here, we have an assumption that in the linked data universe, people will *not* be looking at entire records, so that someone will not even be able to examine an entire record to learn that the relationship of John Huston to Moby Dick is that he was director and not an actor. They may only see the name (perhaps through a URI) “John Huston”. If the other fields of a record are not readily seen because of linked data, then it can be argued that the information for roles (or whatever) must be carried within the data, itself (in this case, along with John Huston’s heading).
In my own opinion, the real question is: is all of this a problem only from the theoretical point of view, or is it a problem for the actual public? Unfortunately, we don’t have any research and have only anecdotal evidence.
My experience has shown that fewer and fewer people even understand what it means to search by author, even less by subject, and with very few exceptions, a search by title, other than a few major keywords of the item, is too weird for them even to imagine. To focus on practical considerations, and going back to a recent discussion on Autocat discussing Eric Miller’s talk at LC about the new Bibliographic Framework, he said that what needs to be made is something simple because if what catalogers make is too complicated, no web master will ever be able to implement it. (I also wonder if regular catalogers can either) His advice makes perfect sense to me. RDA is anything but simple.