Mailing list mb-style, from: Jim DeLaHunt, Sat Dec 18 11:06:52 UTC 2010 
NGS 'Works' should help cut CSG Gordian Knot
I'm glad to see the recent discussion of style guideline to apply to Works (in the Next Generation Schema, NGS, sense). This is a very important topic. I'd like to offer a high-level motivation that can help us set goals for the Works discussion.
My contributions to MusicBrainz are driven by my desire to get good metadata tags for the FLAC files from my ripped CD's. All the work I do to understand the style guidelines, participate in mb-style (back when I used to do so), enter metadata, vote on edits, correct errors, was and is all with the hope that when I run Picard or some future tagger on my FLAC files, I'll get good metadata tags. And that mostly means satisfying, useful Artist, Track Title, and Album Title strings for each file (along with cover artwork).
My music is almost all classical and opera. So I'm plunged into the big, tangled knot of the Classical Style Guide (CSG). The CSG conventions guide editors to writing good MB Artist, Track Title, and Album Title strings. The Advanced Relationships provide a way to describe all the Artists what participate in a particular music file, far beyond the capacity of an Artist field, or a "feat." string in a Track Title.
But the CSG is a mess. It's big, it's complicated, it's difficult for experienced editors, and overwhelming for novice editors. And it's unavoidable, because in the old scheme, you only get good Artists, Track Title, and Album Title strings by following the CSG. If editors type in what makes sense to them, or what's printed on the physical release, the result is an inconsistent, unsatisfying mess. If I were to run Picard on classical music files drawing from MB records not following the CSG, I'd likely be unhappy with the result.
Why does classical music and opera require different rules than most music on MB? That's a deep topic. My simplified answer is that, for most music, the performer is the primary artist, and the Artist Intent of the performer, as expressed by what's printed on the the physical release, is the most satisfying text to put in Artist, Track Title, and Album Title tag strings. The composition and the composer are secondary. And the primary performer is likely unitary or a named collection of people (MB only tolerates, if not resists, multi-artist collaborations). The performance is typically contained within one Track.
For classical music, however, the composer is (usually) the primary artist, the composition comes second, and the performers are (usually) tertiary. There are many performers: orchestra, soloists, conductor, singers, chorus, etc. The Artist Intent as expressed by the *composer*, composition by composition, and understood by a shared cultural tradition of classical music, is the most satisfying text to put in Artist, Track Title, and Album Title tag strings. It forms a structure which frequently spans multiple Tracks. What's on the physical release is not always a reliable indicator of the composer's and the cultural tradition's Intent; good labels channel the composer's and tradition's Intent well, shoddy labels can put almost any junk there. ARs are valuable for classical music because they give a way to describe accurately the large number of performers that can contribute to a recording and to the underlying composition.
I believe the value of the Classical Style Guide to this point has been to put the Artist Intent of the composer and cultural tradition into concrete enough terms, that two skilled and hard-working MB contributors, steeped in that tradition, have a fighting chance of independently entering pretty similar texts into the MB Artist, Track Title, and Release Title fields. It also lets them agree on the same ARs.
For me the hugely important value of the Works notion in NGS is to provide a better place to attach the Artist Intent as expressed by the composer, composition by composition, and understood by that shared cultural tradition. Most of this information is independent of specific performances, and so it need not be encoded in the MB Artist, Track Title, and Release Title fields.
Give me a future version of Picard, which can draw a link from my music files, to MB Release and Track entries, to the related Work entries, and ARs; and from those Work entries and ARs draw satisfying text to put in Artist, Track Title, and Album Title tag strings, and I will feel rewarded for my work on MB. At that point, I almost won't care what gets encoded in the MB Artist, Track Title, and Release Title fields anymore. And yes, I'm willing to help develop that future Picard. And a future data-entry tool, that does ARs across multiple Tracks more efficiently.
In this world, the arcana of the CSG now applies to Works, not Tracks and Releases. Expert editors steeped in the CSG lore and the cultural tradition can spend their effort getting the metadata for Works right. Look at what has been accomplished by the CSG Standard (e.g. http://wiki.musicbrainz.org/CSGStandard/Beethoven) for a hint of what's possible. Novice editors need only enter simple Release and Track entries for a new Release; they or a more experienced editor makes the connection from Release and Track to Work (via whatever intermediate structure NGS provides); and voilà! the Release and Track entries benefit from the care put into the Works.
Also, I think some of the arcana of the CSG will wither away. Instead of having to write detailed rules for how to name a track containing movements of symphonies in general, the editors improving on Works metadata can agree on specific adaptations of high-level CSG principles for the Work in question. And if a better adaptation emerges, one edit to the right Work entry will benefit all Tracks recording of that work. Instead of having to write detailed rules for which performers to mention in a "feat." parenthetical, editors can simply add ARs to Tracks and Releases, and let the tagger generate the "feat." parenthetical if desired.
In the legend of the Gordian Knot, Alexander "unties" it by slicing it asunder with his sword. In the same way, I hope that the Works notion of NGS will slice the CSG asunder. I hope it will free most novice editors, and many expert classical music editors, from the CSG's burden.
I submit that we should make this a goal of the Works style guidelines. If the style guidelines for Works don't lead us to this world, then we should reject those guidelines and keep working to improve them.
There are many voices in MB, and on mb-style. The above is merely my view. I look forward to hearing yours.
--Jim DeLaHunt, Vancouver, Canada