29.05.2013 06:25, J. McRee Elrod:
> Jeff said:
>> In some ways, I think we worry too much about consistency and normal forms.
> One thing we learned in a failed expensive attempt to use ONIX, as
> well as publisher produced spread sheets, is that consistency is
> *very* important.

Consistency is not hugely important for purely descriptive data,
though for the general reader, inconsistency there can be
irritating and may cast an unfavorable light on the library's

Consistency is of utmost importance for access-related data. That's
why authority control was invented. For how else can a catalog achieve
reliable results in known-item searches and collocation?
Mind that those data not (yet) authority controlled can nonetheless
be relevant for search functions, like publishers' names, dates
and place of publication names. THe rules say nothing about what
elements may become relevant for search functions, they leave all of
that to implementers. Who are eager to provide all sorts of access
options the rules never bothered with.

Consistency must suffer when rules are changed, no matter what the
qualities of the new rules may be. New and better access functions
will work only for new records and those that can be upgraded
automatically to conform with the new rules. So, reliability must
deteriorate when large numbers of legacy records cannot be