On Wed, Jan 11, 2012 at 12:22 PM, Roy Tennant <[log in to unmask]> wrote:
The thread below has been in the back of my mind for a while and today my concerns with it crystallized. If we truly want to move ahead in a new way, I think it is absolutely necessary to question our assumptions. An assumption I see below is that provenance matters. Itís quite possible that it does, but allow me to question it. In all of my 37 years working in libraries Iíve never encountered a situation where it was necessary to know where the title came from to do useful work with bibliographic data. In what situations is this necessary, and why?

It's rather important if you want to do things like merge records from different sources to which you may attach different levels of credence.

Some sources of bibliographic data are more prone to errors than others, and often the same sources are prone to the same errors. Information about the source of a field is very useful when doing things like overlaying †records.†

†Provenance, and context more broadly †includes the record (utterance) in which a statement was made. If there are several records purporting to describe a resource, and there is disagreement between them on the value of a field for which there which should be identical, knowing that one record is used by 5 organizations, and one is used by 100,000 ought to increase one's degree of belief that the assertion made in the second record is true.

Utterance context is also necessary for preserving some aspects of semantics; for example, one cannot directly merge FAST headings from multiple sources into a single description. †