Adrian Pohl has a very informative blog post about the use of 
identifiers in authorities, stimulated by his reading of Rob Sanderson's 
document:
http://www.uebertext.org/2014/07/name-authority-files-linked-data.htm

The use of authorities to represent "preferred name" strings (as opposed 
to representing identified *entities* who have a preferred display form 
for human usage) appears to be an American anomaly - at least among the 
libraries he mentions.

One question we could ask ourselves, because we do have identifiers for 
the things described in authority records, is how feasible it would be 
to transform LC authorities to identify an entity  rather than a string. 
Is this a few tweaks, a total re-do, or something in between? My gut 
feeling is that it would be a behind-the-scenes change that wouldn't 
affect cataloger usage. Does anyone else see it that way?

I also wonder if we really must resign ourselves to using strings in our 
future data because "that's what's in the MARC records." There are 
authority control services that match the strings in bibliographic 
records to authority records in order to update the bibliographic data. 
That same process should be able to add authority identifiers to the 
bibliographic records for the same matches. In fact, there are vendors 
who will do this today. I'm sure that matching isn't 100%, but I suspect 
that having identifiers for name authorities in our bibliographic 
records is much less of an effort than converting our data and systems 
to BIBFRAME. It seems short-sighted not to begin this process today, 
before the death of MARC, rather than carrying the strings forward to a 
new bibliographic model.

kc

-- 
Karen Coyle
[log in to unmask] http://kcoyle.net
m: 1-510-435-8234
skype: kcoylenet