Posting quickly since I'm scrambling to get out of town tomorrow - I'll do my best to follow along over the next few days but please be patient if I'm slow in responding...
I did see the RDA-L discussion, but caught up on it too late to weigh in usefully. I've always found the notion of an "aggregate" work in FRBR to be minimally useful. The music community in general has absolutely no problem with two Expressed Works appearing together on a single Manifestation, without having to call those two an aggregate Work because they happen to appear together. I've heard (even as far back as that OCLC FRBR Workshop a number of years ago) people argue that contributions to the Manifestation in the sense of deciding which Works go on it are somehow less important than if we called that pulling together a Work, but I don't really buy that. Just because you contribute to a different entity doesn't mean your contribution is less meaningful in my opinion.
I think the music community is comfortable with the one-many relationship between Expressions and Manifestations (remember, this one/many setup is *explicitly* part of the core FRBR model, right in that first Group 1 entity diagram!), and doesn't feel the need to call Works that appear together to be aggregations for a pretty simple reason: sound recordings in particular so often have more than one Work on them. Calling something an aggregate seems to me to be a workaround to cope with uncomfortableness with that one/many relationship, but I don't find myself uncomfortable with that at all. I know the FRBR report says you can model things as aggregates, but as with a lot in the FRBR report it's only helpful to use it if it gets you functionality you need, and for the materials I work with there's not really a benefit in treating three symphonies on a single CD (by different composers!) an aggregate Work. If you need to call all Works that appeared together an aggregate Work, then there would be no need for the one/many relationship between Expressions and Manifestations. With that thinking an anthology of poems is just lots of Expressed Works, no aggregation needed. I'd be inclined (personally) to only say something is an aggregate work if the creator of the original work had a hand in bringing them together, but even then I might be hesitant.
If I recall correctly, the RDA-L discussion focused more on "less important" (ok, that's probably unfair) content like forwards, is that correct? I see that as fundamentally different than the capital-W Works like symphonies that appear together, but even here I'm inclined to either call those their own Expressed Works that happen to appear on the Manifestation, or just pretend they don't exist because it's not worth our time to model them that fully, rather than calling the whole thing an aggregate Work. And I think the RDA-L discussion got complicated because RDA itself muddles FRBR a bit in this area (I forget the specifics but Karen I know you pointed out one case where they seemed to contradict each other), so I figured it was best to just stay out of it. :-)
This is all just my opinion mostly, but in general in the music community FRBR discussions the notion of an aggregate doesn't come up all that much. The Music Library Association had a 2008 task force on Work records for music, which explicitly considered aggregates to be out of scope <http://www.musiclibraryassoc.org/BCC/BCC-Historical/BCC2008/BCC2008WGWRM1.pdf>. I wasn't on that group, but I suspect one reason for that was internal disagreement on whether aggregates were useful at all in this context. Andrew has done some thinking in this area as well, I know!
> -----Original Message-----
> From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of
> Karen Coyle
> Sent: Tuesday, March 16, 2010 8:39 PM
> To: [log in to unmask]
> Subject: Re: [CODE4LIB] Variations/FRBR project relases FRBR XML
> Quoting Andrew Hankinson <[log in to unmask]>:
> > An album with Beethoven's 7, 8 & 9th Symphonies performed by the
> > London Philharmonic would be a manifestation containing three
> > independent expressions of these works, but the album wouldn't be a
> > work by itself. You can have dependent forward relationships, i.e.
> > "Work is an Expression contained in a Manifestation" but, as far as
> > I know, there's no way to specify that a manifestation containing
> > independent works as a separate work unto itself, and still stay
> > within the FRBR model. (please, correct me if I'm wrong...)
> As I said, the discussion on the RDA-L list came to a different
> conclusion, with folks involved directly in RDA and FRBR coming down
> (one rather harshly to me offline) that a compilation is an expression
> in itself. We didn't get so far as a compilation expression being one
> to one with a work, but I would like to move this discussion to that
> list, since the RDA experts are probably not following this list. I
> guess what I'll do is post the link to Jenn's site on RDA-L, since I
> haven't seen her mail there.