> Stuart - can you give the examples you were going to give now VIAF is
 > back up?

I believe that, as librarians, we can all see a semantic difference 
between the documents pointed to by these two URIs:

[I'm not interested in the exact semantic differences or the terminology 
we use to express them, that's a kettle of fish to be savoured separately].

I believe that we can agree that future BIBFRAME-based library systems / 
ecosystems are likely to need to make distinctions based on similar 
semantic differences.

I posit that the distinction between those documents is the same as (or 
at least parallel to) the distinction between the 'thing' and the 
'authority' in the lightweight abstraction layer.

I posit further that the lightweight abstraction layer is needed to 
express the difference between these documents:

Some of us may choose to banish some of those documents from the systems 
under our purview, but without the ability to express the differences 
and reason about them, we'll be unable to do so in an automated fashion 
based on BIBFRAME.

For example, we have a subject classification called Ngā Ūpoko Tukutuku It's entirely possible 
that we'd accept NŪT metadata from sources which we'd not normally 
accept LCSH metadata from.  Being able to express that in BIBFRAME opens 
the door to automated rule-based acceptance of such metadata, across a 
group of libraries.

Such a rule might be "When considering an authority for import, IF the 
authority refers to an entity that we already have AND we don't have NŪT 
for the entity AND the entity being considered does have NŪT for the 
entity, THEN import the authority." Such rules are hard / impossible to 
conceptualise without an 'authority' separate from the 'entity'.

[Note that I'm not suggesting we increase the scope of BIBFRAME to 
include rule-based workflows, just attempting to ensure that we can 
represent the condition clause of a rule in BIBFRAME]

ngā mihi
Stuart Yeates
Library Technology Services