[mb-users] distinct artist names

[Lauri Watts]

On 12/8/06, Will Holcomb <wholcomb at gmail.com> wrote:
> I've been poking around at the musicbrainz database for a couple days
> and I have a couple questions concerning artist collaborations. I've
> searched through the wiki some and the list archives, and haven't
> really seen an answer to my question.
>
> >From the "OfficialStyleGuideline" I read: "When two artists
> collaborate, file the track under the PrimaryArtist, and append the
> name of the SecondaryArtist to the TrackTitle like this '(feat.
> Everlast)'."
>
> I was wondering if this practice had been abandoned and the
> documentation not updated. When I run: "select count(*) from artist
> where name like '% & %' or name like '% and %' " it returns over
> 26,000 entries. Some of them are the names or bands or groups that
> have "and" in them like "Buddy DeFranco & His Orchestra," but the
> majority of them aren't. Is this erroneous information that has been
> entered, or is it intentional?

It's entirely intentional.  You probably need to read this:
http://wiki.musicbrainz.org/SG5DisasterRelief

You probably also don't want to hear that "and" comes in other
languages too: to really find them all you would also need to find
'og' and 'och' and 'y' and  'con' and  'und' and about 120 others :)

The practise has not been (entirely) abandoned.  If a track is very
clearly by one artist, for instance when it's only every been released
by that artist's albums, then this is still the right thing to do, but
it only applies if one of the artists really is 'primary' and yet both
have the cover credit.

That case isn't actually all that common, it's usually simply not that
clear cut.

More importantly, many, in fact, most of the feat. artists are in fact
credited that way on the covers, and completely correctly entered as
(feat...).  Elevating them to primary artist by creating a new
collaboration would not be correct.

> http://musicbrainz.org/artist/b77bb955-4b08-4a2c-917e-87085328a3ce.html

This is the correct and complete attribution on the release in
question, and none of these artists can be considered 'primary' at
all.

> Never Give All the Heart (feat. Brenda Fricker & Anúna):
>         was performed by Anúna and Brenda Fricker
> http://musicbrainz.org/track/c7ffb015-4e41-49d1-94bb-393bfcc5c7d1.html

This track is on a "The Chieftains" album, and right on the cover,
says (featuring Brenda Fricker & Anúna).  The fact the Chieftains
performed it too is implicit in their being the primary artist, and
doesn't need another relationship entered.

> You Nearly Lose Your Mind (feat. Willie Nelson & Waylon Jennings)
>         was performed by Waylon Jennings & Willie Nelson
> http://musicbrainz.org/track/423653ed-5515-4b0d-98ae-0e2a6fc21d41.html

This one's wrong, and should be broken up to have links to Waylon
Jennings and Willie Nelson. Some of these connections were automated,
and aren't always right (you can see in the edit history who put the
connection there), but luckily, and we can fix these things.  That
said, Waylon Jennings and Willie Nelson did release some collaborative
albums with both receiving primary credit on the cover, and *those*
albums are correctly listed under this collaborative artist.  (they
actually say just "Waylon and Willie" on the cover, but well, we know
who Waylon & Willie).

Here's the edit to correct that:
http://musicbrainz.org/show/edit/?editid=6092050

> http://musicbrainz.org/show/artist/?artistid=199813
>
> I am going to assume those are always an error since you don't gain
> anything with these composite artist names other than the tracks not
> being listed under the individual artists.

Accuracy is more important than perceived duplication.

That said, the AR's on this entry are incorrect.  They should be split
up the way I just did for the Waylon Jennings and Willie Nelson one
above.

But this artist entry also has two real entries.  It's fair to assume
these were credited this way; both are compilations and I'm known to
have a thing or two to say about compilation issuing label's accuracy
and QA, so the attribution on those covers may be wrong too.  But they
may well be correct, someone familiar with Nas and/or Nature will have
to look into that.  I'm assuming here this Nature isn't the Swedish
80's rock band Nature, or I'd look into it myself (but that'd be a
collaboration from out of left field)

In general, relationships should point to specific individuals, where
possible.   But that's not _always_ correct, often you simply have to
know the artists in question.  If it was always automatable, we could
just dump the soundscan data into the database and skip all this
tedious hand editing.  It'd probably save our exalted leader some
headaches (well, and it wouldn't be free, soundscan info costs a
bomb.)

> The reason I ask about the correct style for collaborations is I want
> to know when to submit corrections and what data is correct so far as
> musicbrainz is concerned.

If they really released things with equal credit, they are correct as
is.  If they really didn't ever do that, they probably aren't correct.
 If you really don't know, don't edit it.  And one-off collaborations
for a movie soundtrack count as 'really released' (those are in fact
pretty common, and the source of several of the ones you point out.)

> http://musicbrainz.org/track/1c20a5da-d780-4e66-8285-14bb9abb37ba.html
> The title record lists the artist as "Bill Haley & His Comets" there
> is a link relationship to "Marshall Lytle." What does this mean? I'm
> assuming it means that both artists performed on the track, correct?
> http://musicbrainz.org/track/01c7d02d-71cd-42fe-99bc-6dfe7aec5301.html

Correct, the artist it is listed under is obviously also performing.

Worth noting here, "Performed" is the catch all default, and they can
(eventually will, and should) be much more specific than just
"Performed".  And the relationships can be album wide:
http://musicbrainz.org/show/release/?releaseid=180709 is a pretty
detailed example.  Note that all the relationships at the top, that
apply to the entire album, therefore also implicitly apply to each
track on it.

In the Marshall Lytle case, someone who has (or can show evidence on)
the track, is highly encouraged to fill out the relationship with
what, precisely, Marshall Lytle was doing on that track.

> Artist "Brian Eno & David Byrne," link to "David Byrne." So both Brian
> Eno & David Byrne performed? Are there any issues with having a link
> to Brian Eno as well?

This one likely shouldn't have the link to Brian Eno at all.  Eno &
Byrne have released multiple works with equal credit, and this artist
entry is correct.  The most likely scenario: This was originally first
listed as a David Byrne track with, had the relationship to Brian Eno
added, and then moved to the collaboration "Brian Eno & David Byrne".
(Upon checking, that isn't quite how it happened, it started out in an
even worse state than that:
http://musicbrainz.org/mod/search/results.html?artist_type=3&object_type=track&artist_id=0&orderby=desc&object_id=158657)

> currently by an artist with the name "Christina Aguilera, Lil' Kim,
> Mýa & Pink," instead of making the track by Christina Aguilera and
> modifying the track name with a (feat.) addition, simply adding links

Simply put: Christina Aguilera did not perform this track, the others
are not guests or featured artists.  They all performed it, and are
all equally credited on the original release.  Moving it to Aguilera's
discography would not be correct, and she doesn't become the primary
artist simply by virtue of being the first one on the list.

-- 
Regards,
Lauri "wow, that one was really long" Watts