• Registered: 2012-02-29

Topic: "recording" ≠ "performance"

I currently have an open edit (http://musicbrainz.org/edit/16672845) that has gotten me thinking about the relationships between performance and recording. 

My edit was originally prompted by the MB page for the work "Marquee Moon" (http://musicbrainz.org/work/87c57a43-43 … 82bc0cb92d), which relates it to two "performances."  Clearly, I thought, that's wrong.  As I noted in my edit, both of those "performances" are the same "performance" in the generally understood sense of the word -- in this case, four people singing and playing instruments in a recording studio.

But PhantomOTO's point, in his note on my edit, is a good one.  If the recordings were merged, the work page would look better (at least to me) in that it wouldn't show two "performances" when really, historically, there was only one.  But there would be no indication that two different mixes/masters were made from that performance, plus the "Length" column on the new, merged recording page would show two different lengths, which seems confusing.

Most likely I'm going to cancel my edit.  If I had to vote right now, after thinking it over, I'd say that "recording," in MB terms, should correspond to the master recording from which copies are made, since those copies -- the tracks on our LPs and CDs, the MP3s on our computers and phones -- are what MB was created to catalog.  There would be no way to indicate precisely how those copies relate to the mixes and studio recordings from which they were made, but I don't have a problem with that, since those relationships are not well or widely documented anyway.

But I'd also say that "performance" should NOT be used as a synonym for "recording" on MB.  Of course, for many decades, starting with Edison cylinders, the relationship between them was basically 1:1, and in many cases it still is, but in many other cases it's more complicated.  A famous example is the Beach Boys' "Good Vibrations," which is a three-minute distillation of 94 hours of material recorded in 22 separate sessions over the course of seven months.  Or just think about sampling.  Compared to, say, the Avalanches' Since I Left You (~3,500 samples), my problem with "Marquee Moon" seems so straightforward it's quaint.

Basically, I'm coming to the conclusion that the term "performance" should not be used on MB, at least not in the structural way it is now.  In too many instances, performances are either not properly documented or impossible to separate from the recording process.  And performances are not really what MB was created to catalog anyway -- right?  If the work page for "Marquee Moon" would just show two "recordings" instead of two "performances," I'd be much happier.

Thoughts?

And forgive me if this has all been hashed over already.  I did some searching on the forums but couldn't find previous discussions of this.

2 Reply by mfmeulenbelt (edited by mfmeulenbelt 2012-02-29 23:44:45)

  • Registered: 2008-04-25

Re: "recording" ≠ "performance"

The problem is that despite our longing for factual data and correct terminology, you'll always run into edge-cases where the terminology is sub-optimal. No matter what word you choose, we'll always have to define what we mean with that word exactly.

mfmeulenbelt's Website

  • Registered: 2012-02-29

Re: "recording" ≠ "performance"

That's exactly what I'm trying to do -- that, and also define what we don't want to have to define :-)  In other words, I'm suggesting that the purpose of MB is to catalog "works" and "recordings" while omitting all the stuff in between, including "performances."

That seems to be where the general consensus already is -- I had noticed, for example, that the Wiki includes entries for "work" and "recording" but not for "performance."  I'm just thinking that by not using "performance" as a synonym for "recording," we can make the terminology a bit more consistent, to reflect that consensus and prevent confusion.  People can argue about whether "All Tomorrow's Parties" and "All Tomorrow's Parties" are really two different "performances" of "All Tomorrow's Parties" – or we can simply bypass that argument, and others, by replacing the word "performance" with the word "recording" in the code from which those pages are generated.

But I haven't been active on MB long enough to have a good sense of how hard it would be to make that change, what its ramifications would be, etc., which is why I put it up here for consideration.

  • From: Czech Republic
  • Registered: 2008-03-17

Re: "recording" ≠ "performance"

I agree with practik that we should stop using the term "performance" in meaning of recording.

IMHO it is just too wrong and confusing.

pabouk's Website

  • From: Singapore
  • Registered: 2007-02-13

Re: "recording" ≠ "performance"

FWIW, here is probably the wrong place to put such things up for consideration except in "testing the waters". The Style mailing list is the only place a decision/proposal is made.

While you and others wax lyrical about definitions on the style mailing list or elsewhere, stop for a minute to think about how endlessly confused and frustrated most users are as a result of with introducing both "work" or "recording" as concepts. Most MB contributors (important people, given they built the DB to what it is) have no idea what either term means, and nor do they care. They just want to add some tracklists for their artists, help out a bit, and tag their collection. Balance is hard. :/

voiceinsideyou's Website

  • From: here
  • Registered: 2006-12-19

Re: "recording" ≠ "performance"

I don’t agree that this is the wrong place.
The forum is handy for discussing and there is even a forum called Style Discussion (discussion about MusicBrainz style guidelines).

On topic, we indeed shouldn’t call these entities performances because we would then merge all the remixes (because there was no new performance recorded, only post-processing).

☞ Musicbrainz forums enhancer (LA TÉLÉ FAIT GROSSIR ET NUIT À L’ÉVEIL DU CERVEAU)
☞ other cross-browser OMG MB scripts for editors and voters (♪ jesus2099)

jesus2099's Website

  • From: Singapore
  • Registered: 2007-02-13

Re: "recording" ≠ "performance"

You don't agree it's the wrong place because you have a clear bias against mailing lists, you've made abundantly clear. But what's the point of fragmenting discussion and giving people false hope that discussion here changes the policies? It does not. The style mailing list does.

voiceinsideyou's Website

  • Registered: 2012-02-29

Re: "recording" ≠ "performance"

voiceinsideyou wrote:

here is probably the wrong place to put such things up for consideration

Thanks, voice, for pointing me in the right direction.

I take your point, too, about confused users who just want to add some tracklists.  In case I gave the impression that I want to add "performance" as a new MB term on an equal footing with "work" and "recording," I don't -- on the contrary, I don't think "performance" really has any place here.  And it seems to me that taking it out and consistently using "recording" could only cut down on user confusion.

Anyway, I guess I'm off to the mailing list.  Thanks, mfmeulenbelt and pabouk and jesus2099, for weighing in as well.

  • From: here
  • Registered: 2006-12-19

Re: "recording" ≠ "performance"

voiceinsideyou wrote:

You don't agree it's the wrong place because you have a clear bias against mailing lists, you've made abundantly clear. But what's the point of fragmenting discussion and giving people false hope that discussion here changes the policies? It does not. The style mailing list does.

So let’s shut down the forums, and its Style forum.
No need to keep it if it’s neat but useless.

☞ Musicbrainz forums enhancer (LA TÉLÉ FAIT GROSSIR ET NUIT À L’ÉVEIL DU CERVEAU)
☞ other cross-browser OMG MB scripts for editors and voters (♪ jesus2099)

jesus2099's Website

  • Registered: 2011-06-06

Re: "recording" ≠ "performance"

"Performance of" makes a lot of sense when you use it to denote a cover by another band, but it's not completely necessary. 'Cover recording' sounds weird, but logistically wouldn't make too much difference, as long as you can still define cover/ live/ partial etc

  • Registered: 2012-02-29

Re: "recording" ≠ "performance"

aerozol wrote:

'Cover recording' sounds weird

Yeah, I thought about that, but I didn't feel like anyone would get too tripped up by it.  Glad you seem to agree.  And yes, the attributes cover / live / partial / instrumental should definitely be retained.

  • From: Singapore
  • Registered: 2007-02-13

Re: "recording" ≠ "performance"

jesus2099 wrote:
voiceinsideyou wrote:

You don't agree it's the wrong place because you have a clear bias against mailing lists, you've made abundantly clear. But what's the point of fragmenting discussion and giving people false hope that discussion here changes the policies? It does not. The style mailing list does.

So let’s shut down the forums, and its Style forum.
No need to keep it if it’s neat but useless.

It's useful for people to get adhoc style advice "how would you handle this case?" - but when discussion extends to suggestions in changing the guidelines, ARs etc, and people who want to effect this change - the mailing list is more appropriate.

voiceinsideyou's Website

13 Reply by ym

  • ym
  • Member
  • Offline
  • From: Istanbul, Turkey
  • Registered: 2008-10-08

Re: "recording" ≠ "performance"

covers are performances true but when they are recorded they become recordings...

14 Reply by sbontrager (edited by sbontrager 2012-04-30 02:18:14)

  • Registered: 2012-04-30

Re: "recording" ≠ "performance"

Keep in mind that there are industry standards at play here
ISWC = work
ISRC = recording

The ISRC is actually issued when the song is mastered (usually), not when the artist is in the studio. If a new version is issued which only changes by fading either of the ends (within 10 seconds) then the same ISRC applies. A remix or remaster is typically issued a new ISRC.

This lends to counting different masters as different recordings here. But I don't like it.

I would prefer to have another layer of abstraction and count ISRC's tied to "masters" and "recordings" as what most people think of as a recording--when the recording was made (a very nebulous concept now that the drums can be recorded on one day, the vocals a week later and the horns added a month after when the producer changes her mind... again...).

work -> (performance of work) -> recording -> master (ISRC) -> track
work -> (performance) -> recording -> remix (ISRC) -> track
work -> (performance) -> recording -> re-master (either from earlier master or original sources == new ISRC) -> track

That's 1 work, one performance/recording, three masters for one "song" (orig. & remaster) and one "remix"

work -> cover performance -> recording .... and a remix and a later remaster of the cover.... suddenly we all go crazy because of too many layers of abstraction.

-----

I've had conversations around if different media warranted a different recording--even if from the same master. e.g. a vinyl release and a CD release cut from the same master (not counting SRC/Dither to get it onto the CD or applying the RIAA curve to get it on vinyl as remastering) would be two different recordings. I think this is insane unless it was clearly mastered differently for the various media, which almost never happens (outside of slight EQ tweaks). Same ISRC == same recording/master.

15 Reply by ym

  • ym
  • Member
  • Offline
  • From: Istanbul, Turkey
  • Registered: 2008-10-08

Re: "recording" ≠ "performance"

We don't need another layer for performance. All we need is a strict definition like:
1) A work is a (re)(partial)performed composition and/ lyrics, but not recorded yet --> A recording is recorded audio of work.
2) A work is a composition and/or lyrics, but not performed yet. --> A recording is the recorded audio of performed work.
 
Each would be valid IMO and I think currently NGS is modelling the second one, but the practical use is not yet so clear. For example, there was a work-recording relationship "recording is a performance of work", now it is changing to "recording is a recording of work" relationship. If this change will happen, I feel it will be more confusing if NGS is modelling the first or second (ie: which one shows performance?).
Also, please note that, having ISWC or ISRC in MB doesn't mean the definitions in schema are in one-to-one relationship with those databases.

Still, both models need some questions to be answered, like,
*do the altered compositions (like remix, djmix, etc),  altered lyrics (like hip-hop covers, etc),  products show a new composition/lyrics or just a new performance?
Looking at the current schema, it seems they all need a new Work, because you can only enter additional composer or lyricist to Works, but since these examples also denotes performances, it seems we are now modelling the first version (ie. performance is in the work). This is where the schema fails. It does not follow the same logic. It gives definitions different meanings in different practices. And noone can solve this with style guidelines.

  • From: Tartu, EE
  • Registered: 2010-11-24

Re: "recording" ≠ "performance"

It's completely impossible to limit what is a different work, because the limit just doesn't exist. It's subjective and it *does* have different meanings in different practices. In a classical context, Ravel's orchestration of Mussorgsky's Pictures *is* seen as a different thing (and often credited as Mussorgsky/Ravel). In a popular concept, Nothing Else Matters by Metallica performed with a symphony orchestra is seen as basically the same thing. That's why it's impossible to make a "correct" fit-for-all schema. Also note that most remixes (and definitely DJ-mixes) are credited as "additionally *produced*" not "composed", and as such they don't require new works right now. Hip hop covers are (thankfully, because they tend to suck) not too common, and hip hop "remixes" with new rappers on them are IMO clearly a derivative work since the lyrics tend to be almost completely different.

  • From: here
  • Registered: 2006-12-19

Re: "recording" ≠ "performance"

According to the fact that many posts were made, this seems to be the right place, voice.

☞ Musicbrainz forums enhancer (LA TÉLÉ FAIT GROSSIR ET NUIT À L’ÉVEIL DU CERVEAU)
☞ other cross-browser OMG MB scripts for editors and voters (♪ jesus2099)

jesus2099's Website

18 Reply by ym

  • ym
  • Member
  • Offline
  • From: Istanbul, Turkey
  • Registered: 2008-10-08

Re: "recording" ≠ "performance"

reosarevok wrote:

It's completely impossible to limit what is a different work, because the limit just doesn't exist. It's subjective and it *does* have different meanings in different practices.

Noone can model non-exist things. If it is a limitless thing, then modeller put boundaries and make it limited.

reosarevok wrote:

In a classical context, Ravel's orchestration of Mussorgsky's Pictures *is* seen as a different thing (and often credited as Mussorgsky/Ravel).

At work (and also recording) level we don't and shouldn't care about how it is credited. I thought this was whole point of the NGS schema. To identify, different recordings, releases of same works... But as I said, to answer this, you have to decide what work (and recording) includes and excludes.  Is orchestration (I suppose something like arrangement?) a creates a new performance or new composition? Did Mussorgsky composed or just performed Pictures?

reosarevok wrote:

In a popular concept, Nothing Else Matters by Metallica performed with a symphony orchestra is seen as basically the same thing.

You can't decide because you don't know if a work includes a performance or not...

reosarevok wrote:

That's why it's impossible to make a "correct" fit-for-all schema. Also note that most remixes (and definitely DJ-mixes) are credited as "additionally *produced*" not "composed", and as such they don't require new works right now.

I've never noticed this before. I mostly see a remixer credit generally. Anyways, it doesn't matter actually, because I don't still know what a work is...

reosarevok wrote:

Hip hop covers are (thankfully, because they tend to suck) not too common, and hip hop "remixes" with new rappers on them are IMO clearly a derivative work since the lyrics tend to be almost completely different.

I think here my super-duper English sucked again sorry, what I meant was the added hip-hop/rap lyrics and/or tunes to a normal song, something like, Snoop Dogg's Riders on the Storm for example. Some people might see it as a cover, but I don't think so, because it also includes performance (or samples) from original version. I'm sure there is a good term for this in English which I am not aware of. I just call it a Snoop Dogg version. So IMO, it must have a new work in current schema because I cannot enter Snoop Dogg as lyricist otherwise. See? This is a limitation. So what schema tells to me is, if there is added lyrics to the original work, you need a new work in database... Now we can argue, if this is true or not in this case, but if the schema does not limit you, you can't argue about anything...
This example also is same for the compositions, if there is an added musical work, you need a new work.
An example, metallica plays Nothing else matters, with x band (orchestra), artist a does the orchestration, but Metallica wrote the original composition. So, did the artist a altered the original composition? no he just arranged it. sure he probably wrote some scores, but the original written composition is the same. So, I just see this as a new performance (even cover) of Metallica's original version. So, I don't think there will be need for a new Work. But this is how I see it. This doesn't mean others will see it like this and if anyone says we need a new work I won't argue, because the schema doesn't limit us... This is where the things gets ugly, because now it means we are not speaking the same language. When someone looks at the data they all need to see the same meaning. Not maybes...

  • Registered: 2011-09-15

Re: "recording" ≠ "performance"

I tend to agree with practik and sbontrager. I think the term cover recording is a more accurate term than cover performance. MB does not deal with performances but recordings.

For me the distinction between work and recording is clear. Recordings are the data we have that represents a particular sound. So remasters, remixes and covers are new recordings, which is in line with them having a new ISRC. We shouldn't be concerned with performance as that is already implied by the recording. I don't think vinyl mastering should constitute a new recording. This is part-and-parcel of the vinyl process.

Works are the intangible, intellectual things that are performed and can be attributed to an individual or group. If a change to that thing is made by sampling it, changing lyrics or orchestration, whether that becomes a new work can only be decided on a case by case basis. There's no consistent way to decide as the effect of a change is entirely subjective, so it is a subjective question to ask what kind of change makes a new rendition of a work substantially different to the original. I suppose it has to be put to a vote.

I don't see how recordings an works are confusing to new users. Works will only crop up if you go looking for them, they are not part of the add release process. Recordings appear on one page of the process and are by default set to 'add new'. If a user doesn't understand, they just press continue. Simple.

These are useful features, if a user is interested in them they should read up on them before making changes. To be honest, I don't know why anyone would touch them if they didn't have a clear idea what they were.

20 Reply by ym

  • ym
  • Member
  • Offline
  • From: Istanbul, Turkey
  • Registered: 2008-10-08

Re: "recording" ≠ "performance"

Lixobix wrote:

I tend to agree with practik and sbontrager. I think the term cover recording is a more accurate term than cover performance. MB does not deal with performances but recordings.

So, what you say is; any "live" performance cannot be a cover? That is odd...

Lixobix wrote:

We shouldn't be concerned with performance as that is already implied by the recording.

Is it really? If anyone says some remix is not only a recording but also needs a work, because, it has an artist (remixer) which creates a complete new work, how will you argue with that?

Lixobix wrote:

Works are the intangible, intellectual things that are performed

So what you say is; a work implies a performance? Wait, you just said the opposite above!...

Lixobix wrote:

There's no consistent way to decide as the effect of a change is entirely subjective, so it is a subjective question to ask what kind of change makes a new rendition of a work substantially different to the original.

Well, ISWC is doing it. Why not study how they do it and apply similar (probably not same) logic here? Sometimes, there is no need to reinvent the wheel.

Lixobix wrote:

I don't see how recordings an works are confusing to new users. Works will only crop up if you go looking for them, they are not part of the add release process. Recordings appear on one page of the process and are by default set to 'add new'. If a user doesn't understand, they just press continue. Simple. These are useful features, if a user is interested in them they should read up on them before making changes.

First of all if you say a new user not to use some part of your database before studying, you will certainly loose them... That's really a bad example from a marketing point of view. Also, you can't even decide if a user needs a recording, work or both, so how do you expect someone to not to use it until they learn it? What will come next? don't listen the cd before you decide what it is?

Lixobix wrote:

To be honest, I don't know why anyone would touch them if they didn't have a clear idea what they were.

I don't understand how you can't see a new user desire to use Work objects, as in "where the hell I will enter the composer or lyricist?"... "damn I just wanted to enter a damn cd, what is this gumbo jumbo? work, recording? relation? no, I don't want a relation!! I thought this was a music site!! bah!!"...
So, IMO, instead of expecting that your userbase is super intellectual, reading every guide before using things, type people, you have to think how the user interface and database can be more user (yeah what they call dummy-user indeed) friendly.

ps: Please, don't think that I am calling any user of Musicbrainz, "dummy". Not sure about it, but I think dummy-user was a marketing term, which was used to describe user friendly systems, like; "even a dummy can use it". It is also used in computer programming, like; "always think that your users as dummies and design for them (aka, never estimate anything about your users)".

  • Registered: 2011-09-15

Re: "recording" ≠ "performance"

ym wrote:
Lixobix wrote:

I tend to agree with practik and sbontrager. I think the term cover recording is a more accurate term than cover performance. MB does not deal with performances but recordings.

So, what you say is; any "live" performance cannot be a cover? That is odd...

No. A live performance can be a cover. My point is that that MB is not concerned with whether an artist performs a cover version at a concert, but whether there is a recording of that event. If the concert was never recorded, it is irrelevant to MB. So in that sense we deal with recordings rather than performances, whether covers or originals, live or studio.

ym wrote:
Lixobix wrote:

We shouldn't be concerned with performance as that is already implied by the recording.

Is it really? If anyone says some remix is not only a recording but also needs a work, because, it has an artist (remixer) which creates a complete new work, how will you argue with that?

If there is a cover recording, then there has obviously been a cover performance. Otherwise, what would have changed between the original and cover recordings?

I will discuss remixes below.

ym wrote:
Lixobix wrote:

Works are the intangible, intellectual things that are performed

So what you say is; a work implies a performance? Wait, you just said the opposite above!...

No. A recording implies performance. Works are intangible, intellectual things. They are performed, and those performances are recorded. However, the performance is not part of the work. Every performance of a work will be slightly different. But the work itself stays substantially the same. A work exists without being performed. Thus the compositions written on paper by classical composers are works, before they are ever performed. Or the lyrics Bob Dylan writes on his typewriter, before they are ever sung.

ym wrote:
Lixobix wrote:

There's no consistent way to decide as the effect of a change is entirely subjective, so it is a subjective question to ask what kind of change makes a new rendition of a work substantially different to the original.

Well, ISWC is doing it. Why not study how they do it and apply similar (probably not same) logic here? Sometimes, there is no need to reinvent the wheel.

I don't know much about ISWC or the principals they use in deciding whether something is a unique work. Yes, perhaps they could be used here. If so, they should be put in the style guidelines. However, I think the point still stands that it will often be hard to quantify the degree of change required to make a unique work.

Therefore, whether a remix becomes a unique work is subjective. They can sound very similar to the original, or dramatically different, with only a few of the original parts involved. But there will always be something distinctive that links it to the original.

A remix is never a 'complete new work' because it always uses some distinctive feature of the original, e.g., vocal, synth line etc. So it always has some link to the original. In that sense, I lean towards never considering a remix to be a unique work, but I admit it is a grey area.

ym wrote:
Lixobix wrote:

I don't see how recordings an works are confusing to new users. Works will only crop up if you go looking for them, they are not part of the add release process. Recordings appear on one page of the process and are by default set to 'add new'. If a user doesn't understand, they just press continue. Simple. These are useful features, if a user is interested in them they should read up on them before making changes.

First of all if you say a new user not to use some part of your database before studying, you will certainly loose them... That's really a bad example from a marketing point of view. Also, you can't even decide if a user needs a recording, work or both, so how do you expect someone to not to use it until they learn it? What will come next? don't listen the cd before you decide what it is?

Lixobix wrote:

To be honest, I don't know why anyone would touch them if they didn't have a clear idea what they were.

I don't understand how you can't see a new user desire to use Work objects, as in "where the hell I will enter the composer or lyricist?"... "damn I just wanted to enter a damn cd, what is this gumbo jumbo? work, recording? relation? no, I don't want a relation!! I thought this was a music site!! bah!!"...
So, IMO, instead of expecting that your userbase is super intellectual, reading every guide before using things, type people, you have to think how the user interface and database can be more user (yeah what they call dummy-user indeed) friendly.

ps: Please, don't think that I am calling any user of Musicbrainz, "dummy". Not sure about it, but I think dummy-user was a marketing term, which was used to describe user friendly systems, like; "even a dummy can use it". It is also used in computer programming, like; "always think that your users as dummies and design for them (aka, never estimate anything about your users)".

What I mean is that I think most new users want to contribute releases with the correct album, artist, tracks and dates, so they can correctly tag their files. Obviously, if we can make other aspects more clear that is a good thing, but we should not get rid of useful features because they are confusing at first. The work and recording concepts make a lot of sense once you understand them, although admittedly require some effort to understand at first. However, if a user is keen to enter composer, lyricist etc. I think they have to learn about works/recordings. If we didn't have them, users would have to add the same information for every release containing that track.

What I'm getting at is you can get started on MB without worrying about recordings/works. If you're interested you can learn about those features later. So you have advanced features that experienced users can appreciate but aren't needed in order to make your first additions. I agree that there should be better implementation of guidelines/definitions within the editing process itself, but I don't think there is anything that can be done to recordings/works to improve them other than explaining them better.

  • From: Tartu, EE
  • Registered: 2010-11-24

Re: "recording" ≠ "performance"

Meowth, that's right! Thanks Lixobix for saving me a lot of typing by basically stating all my opinions!

  • Registered: 2011-09-15

Re: "recording" ≠ "performance"

reosarevok wrote:

Meowth, that's right! Thanks Lixobix for saving me a lot of typing by basically stating all my opinions!

:D

24 Reply by ym

  • ym
  • Member
  • Offline
  • From: Istanbul, Turkey
  • Registered: 2008-10-08

Re: "recording" ≠ "performance"

http://wiki.musicbrainz.org/Work wrote:

In MusicBrainz terminology, a work is a distinct intellectual or artistic creation, which can be expressed in the form of one or more audio recordings.

The problem here is the meaning of this sentence can be applied to almost anything. It doesn't say anything about work actually, it doesn't limit you. It doesn't define anything. It creates a lot of gray areas.

http://wiki.musicbrainz.org/Work wrote:

While a work in MusicBrainz is usually musical in nature, it is not necessarily so. .

Now we have more broader meaning... A work can even be anything!!
Although, later in that wiki page, some limitations and examples are given, they are mostly to define classical works and how to link derivative (or such) works through ARs, it still doesn't say what is a "derivative work".

http://wiki.musicbrainz.org/Work wrote:

A work can be associated with one or more other works. There are two types of work-work relationships:
Part-of-work relationship
A work can be expressed as a part of another work.
Derivative work relationship
A work can be expressed as being derived from one or more other works. Examples: instrumental work with lyrics added later, translation of a work into a different language, mashup.

All I see here is you can enter derived works, which is a nice feature, but I still don't see what a derived work is. Again a huge gray area.

http://wiki.musicbrainz.org/Work wrote:

Discrete works
An individual song, musical number or movement. This includes recitatives, arias, choruses, duos, trios, etc. In many cases, discrete works are a part of larger, aggregate works.

It starts with a good sentence but ending gives again "question marks".
Ok Here is a possible example scenario that I feel anyone will be facing a lot: So, I enter a pop song as work (ok this is not one of that "many cases"), then I enter remix of that song as a new work which is called a derivation work (yes I think a remix is a derivation). Oh wait someone voted against me telling me that remix cannot be a work. Uhm, maybe he is right. Let me check the definition of "derivative work", oh wait he can't be right there is no definition. Hmm, he is talking about, something called "common use". Ye I should apply a common use I will cancel it.
Problem solved? If you think so there is nothing to talk about...

  • Registered: 2011-09-15

Re: "recording" ≠ "performance"

ym, I think you are seeking a clear definition of work that cannot be given. I'll divide my answer into two parts, one for classical and on for popular music.

For classical music, the tradition generally involves a composer writing the composition on paper. This is a work. It may then be recorded by a number of different orchestras, conductors, soloists etc. but none of these recordings is a unique work; the notes on the pages have not been changed.

Taking the example given earlier, where Ravel made a new arrangement of an older piece, it constitutes a new work as the notes on the page have been changed. I think the concept of work is much simpler in relation to classical music, as the tradition is clearly and solidly structured.

With popular music, the distinction is not always so clear; often due to the way that style and technology rapidly change it. I think the majority of popular music works can be clearly distinguished: Much of the time the distinguishing features are the lyrics and melody, although other musical parts may sometimes be distinguishing, e.g. the guitar line in Eric Clapton's Layla. These features are never substantially changed when the work is covered. For example, any cover of a Bob Dylan song will contain substantially the same lyrics and melody.

Remixes are a difficult area because they are a comparatively new idea. I think the general consensus, here on MB and in the music industry generally, is that remixes are not considered works. This makes sense in relation to the fact that they are given ISRCs rather than ISWC. Also, you can draw a comparison with the description I gave above in that the remix (at least a good one) will always contain the distinguishing features of the work. For example, the remix of Brimful of Asha by Fatboy Slim contains the original vocal line (lyrics and melody) of the original Cornershop work and is therefore not a work.

On the other hand, where an artist samples a distinctive feature of a work but uses that one sample in the context of their own composition, that is a work. For example, the Afrika Bambaataa track Planet Rock uses a sample from Kraftwerk's track Trans-Europe Express; but it is one synth line, not the vocal line and lyrics as well. Therefore it is a work.

The real problem is the grey area between these two extremes. A line must be drawn between something being sampled and something being remixed. There are no clear definitions or rules, because the effect the remixer/sampler has on the work is subjective; for some it may make the work completely different, whilst for others it is a minor change. For what it's worth, my opinion is that generally where all the distinctive parts of the original are retained, it is a remix (or cover) and not a new work, whilst if only some are used in a wider context it is a new work. The later would cover hip-hop/rap tracks that add new lyrics to a work, amongst other examples (these would incidentally be derivative works as they take some of the original distinctive features but add new ones too).

The problem is that users will always disagree as to where to draw the line. So sometimes it will come down to a vote, but much of the time I think it is quite clear.

On a side note, the definition of work as being possible to be a book, poem etc. makes sense to me. Because the work is the arrangement of ideas, how the sounds and words should go together. Until the work is performed it only exists in the mind or on paper. So a song can exist in the writer's mind or on a page, just as an author's poem or book would. I think this point was made as readings of books and poetry are recorded and added to MB, so it is important to separate the original poem or book from the recording by a specific reader.

Aggregate works refer to a larger work made up of other works. So each movement of a symphony is a work but also part of the aggregate work that is the symphony itself.

The guidelines should give better examples, but like I said a line must always be drawn between categorisations, so you can never give a completely specific definition of work.

Hope that helps :)