[mb-style] RFC: Works lists (and other related changes then implied)

[Brian Schweitzer]

On Wed, Mar 5, 2008 at 4:41 PM, Aaron Cooper <cooperaa at gmail.com> wrote:
>
>  On 5-Mar-08, at 4:35 PM, David K. Gasaway wrote:
>
>  > Brian Schweitzer wrote:
>  >
>  >> Thus, actually, why I'd opposed the implementation of the parody AR
>  >> as
>  >> an attribute to the cover AR.  The cover AR is distinctly a "song
>  >> instance"-"work listing" AR, linking from one artist's recording back
>  >> to another artist's work listing.
>  >
>  > I didn't get a clear picture of how you would enter a cover.  Would
>  > the covering artist have an entry in the works list for the cover
>  > will an AR to the composition?  If not, I don't think it's fair to
>  > say that a cover is necessarily a cover of a composition.  It could
>  > be a cover of a specific performance (i.e., specific variation in
>  > lyrics, instrumentation, rhythm, or whatever else you could imagine).
>
>  The picture I see is both artists having a work in their worklists
>  with one linked to the other with a Cover AR.  The original
>  composition will also have Composition/Lyrics/Arranged/etc ARs but not
>  the cover version.

I guess it depends on how we define "work".  Is a "work" a unique
composition?  If so, then there ought to not be, imho, a listing also
in the covering artist's workslist.  Or is a "work" simply an
indication that an artist has played a song?  Then it would appear in
that covering artist's workslist.

Personally, I think the first is the cleaner - and better - solution.

Now, so I'm not misunderstood, if we're dealing with a song like, say,
In the Pines / Where Did You Sleep Last Night, where the song has
evolved with every other artist who has recorded it, then yes, that
quite likely ought to appear in many works lists, each of the two
branches of development of that song linking together back through
time.

But if we're making an entry in a works list for each and every cover,
in some cases, it gets rather silly.  Just to consider two cases I
know pretty well, one rather common, one more to the extreme:

Nirvana: If we're using the former definition, then the count is at
about 99 songs.  Under the latter definition, however, that list
swells to around 243 - and that's not even counting untitled jams,
encompassing everything from Smells Like Teen Spirit to Kids Of
America, My Sharona, the Speed Racer Theme, and 867-5309 (Jenny).

They Might Be Giants: Under the former definition, they have 1,117
(!!!) songs (give or take a few dozen, depending on if any one track
is a new version, or just a different recording).  Under the latter, I
don't even want to imagine it - for a decade, while touring, they
played a "stump the band" game, taking random suggestions from the
crowd and covering the songs on the fly.  The count would be at least
3000.

In any case, I think there's actually four levels, not three, that are
involved.  We have "work" - "session" - "track master" - "instance".

Right now, we only have instances.

What luks has been working on is the track masters - the mix+master of
a particular recording is the same down to any number of instances.

What we're talking about now is several steps higher.

Session: The parent of track masters, this encompasses all the masters
and mixes from any particular recording.  Here, I think, is where
covers would actually link from, if we had the entity.  Here is where
the one work links to each performance of the track, live or studio,
by any one band.

Work: The single conceptual work.

As I see it, if we actually had all these elements in place (and note,
I'm not trying to rush the middle two here, only considering why I
think covers ought to not be in works lists) works would be only under
the group that actually created, or sufficiently modified the work
such that it was a "new" work.  They would link then to times anyone
recorded them (or performed them and that performance was recorded,
legit or bootleg).  The session would then link to each of the various
master+mixes of the session, which would then link to the recordings
sourced from any particular master+mix.

This would mean covers linking from sessions back to the originating
artist - but not the inclusion of every cover within the works list of
the covering band.  As we can't actually make that link yet - lacking
session entities - I would think it makes more sense to still link
from the instance (and eventually the track masters) back to the
originating band's work list.  That way, when we do eventually have
sessions, we're only modifying the AR to add in the session, rather
than having to actually relocate all the ARs for the cover work.

Brian