Attached is the IRC chat log from this evening's (PST) chat. Thanks to one who participated. ;-)
Paul
<pbryan> Welcome to part 2, the hey-I-work-for-a-living!-edition of our IRC chat... So, hopefully either you attended part 1, or you read today's transcript/. Probably the best way to do this is to go through the four points of the agenda from earlier. #1: What should and should not be a work. In the last chat, we decided it's something that is written and is backed by recorded audio, not as a hard and fast rule, but generally. Any issues with the scope of works? <brianfreud> "something that is written" is really broad; imho kepstin was expressing the same kind of thing I think of when he said "represents new creative input" that was what I'd intended when I suggested the "12 AR test" <pbryan> How is "represents new creative input" less broad? <brianfreud> "written" doesn't mean anything was changed As example, there were some works listed in the original K list for Mozart which later turned out to simply be copies of other peoples' compositions which Mozart had copied out verbatim. Those are "written", yet imho, not works <pbryan> Are you meaning that, for example, an unwritten improvisation should qualify as a work? <brianfreud> I've never thought so, though I recall ruaok trying to convince me otherwise :P <pbryan> Okay, so you're talking about who wrote what then? <brianfreud> As for "backed by recorded audio", I'd reference the email I sent to the style list an hour or so ago yes <pbryan> So, in such works, it sounds like the composer is actually not Mozart. <brianfreud> yes <pbryan> So, that sounds like a straightforward composer attribution to someone else. <brianfreud> I think "written", but in the sense of "the guy who did the writing was changing/adding something of substance", not just the actual physical act of "writing" <pbryan> I used the term author, actually. IIRC... * pbryan checks the log. ;-) <brianfreud> yes, looks like you did Doesn't "authored" become too narrow though? <pbryan> How so? <brianfreud> Does an arranger "author" anything? Same Q for most of the other 12 ARs <pbryan> An arranger is the author of the arrangement, is (s)he not? <brianfreud> in a sense <pbryan> Librettist is author of the libretto, etc. I think authorship (actually a word!) is more meaningful than "represents a new creative input", but if there's a better word, I'm happy to consider it. <brianfreud> The "12 AR test" definition I'd suggest was that a new work was something where, even if based on another work, the specific set of ARs for the new work would be different, looking at only the ARs from Composition Relationship Class, Compiled Material Relationship Class, and Remixed Material Relationship Class <pbryan> That seems like a good general test to determine whether a new would should be created. s/a new would/a new work would/ Doh, s/a new would/a new work/ <brianfreud> (eg: Arranger, Composer, Instrumentator, Librettist, Lyricist, Orchestrator / Compilation, DJ Mix, Medley / Mashup, Remix, Samples) <pbryan> Right. * brianfreud wonders if anyone else is here <pbryan> It doesn't seem like it. <brianfreud> :P well, if noone else came, I don't want to make you have to do a whole new mtg, just for me... <pbryan> Well, I blocked-off the time, so I don't mind getting the input. <brianfreud> lol, ok <pbryan> So, in principal, I think we're talking about the same thing. <brianfreud> I've gotten used to never being able to make the dev mtgs, but mid-day on Mon is hard for a style mtg :P <pbryan> :-) <brianfreud> yeah, I think it boils down to the wording of the definition, not the concept itself <pbryan> Right, and I think it's going to undergo revisions. <brianfreud> Now, where I do disagree with that definition is in the second clause <pbryan> The part about it being recorded? <brianfreud> yes why is that considered at all? <pbryan> It's not a hard-and-fast rule, but the consensus seemed that the objectives of MB is around recorded works. <brianfreud> Recording = performance. Work = composition. So why involve any kind of performance-related test when dealing with something that has nothing to do with performance? <pbryan> Well, works are performed, and MB to date has been focused on recorded performances. I don't sense that the objectives of MB were meant to be expanded into being a definitive catalogue of written works. <brianfreud> Theoretically, I should be able to pull up artist Foo in MB, artist Foo in ASCAP, and create new works for each any every ASCAP-tracked composition for Foo, without consideration as to it being recorded or not. See, I disagree. <pbryan> Theoretically, yes, you could. <brianfreud> Yes, MB originated with a focus on recorded performances. But I think it long ago became something else. <pbryan> How so? <brianfreud> I'd put that point either when we added ARs, or when we added Labels. But either way, "MusicBrainz is a community music metadatabase that attempts to create a comprehensive music information site." I see "being a definitive catalogue of written works" as falling well within that mission of being a "comprehensive music information site". <pbryan> Is that definition codified in some way? <brianfreud> which one? <pbryan> Of what MusicBrainz is. <brianfreud> That's a direct c&p from the first line of http://musicbrainz.org/ :P <pbryan> Ah, there it is, voila. <brianfreud> The only other clarification I see is on http://musicbrainz.org/doc/AboutMusicBrainz "MusicBrainz is a user-maintained open community that collects, and makes available to the public, music metadata in the form of a relational database." <pbryan> So, I think this point should probably be discussed with the style council in more detail. <brianfreud> 'metadata' is then defined by the database schema itself, on that page. Further, same page, "the project has expanded its scope from being a CDDB replacement to a true "Wikipedia for music"." So when we expand the schema, we expand what the db itself tracks, by definition. <pbryan> Indeed, and it's a matter of what the accepted scope of the DB should be that drives its use. <brianfreud> I don't see that a specific AR not being possible - ie, a recording-work AR - as being itself an overriding reason why only works with recordings should be allowed. <pbryan> So, bottom line, your position is that we should be using MB works to catalogue the (complete) works of authors. <brianfreud> I'm not even saying complete. However, I do think that works should be permitted, even if unrecorded. <pbryan> And I think the current view allows for that, just that it doesn't seem to be the driving force. Nikki's position is that it shouldn't be a hard-and-fast rule. I tend to agree. <brianfreud> Four reasons: 1) basic interest, 2) in rare cases, it make make an otherwise "unmusical" person "musical" (re: Personal Association Relationship Class and it's limits on adding people for ARs), 3) permitting complete work lists for classical composers (where a lost work cannot possibly be recorded), and 4) It allows works to be added for works on *upcoming* albums, even though the album is not yet released (and thus, there is not *yet* a recording) <pbryan> Alright, this makes sense. Anything else on point #1? <brianfreud> nope <pbryan> Okay, #2: Granularity. The consensus seems to be that there should be super-works and sub-works. High level: Concept album, opus, opera, musical. Low level: song, movement, number, recitative. Any further input? <brianfreud> did you see http://wiki.musicbrainz.org/User:BrianSchweitzer/sandbox/Works ? <pbryan> No, I haven't. Looking now. <brianfreud> I generally agree with that kind of granularity. However, I found, while working on CSGv2, that people kept messaging and emailing me with all sorts of cat-corner cases where that basic granularity breaks. These were just the ones I can recall offhand, with a few quick examples. I kept meaning to write that up for you in the past week, but didn't get a chance until 90 min or so ago :P <pbryan> So, I'm struggling to see what the page conveys, except possibly the edge cases. Can we use one as an example? <brianfreud> basically, that's all it is <pbryan> Okay. <brianfreud> however, the problem is, a few of those edge cases are rare, but most aren't. Taken together, those represent maybe 20% of classical, plus 50-75% of opera (and soundtracks/scores, which noone mentioned in the last mtg) <pbryan> Well, I see a lot of these being cases where some work is excerpted in a track. <brianfreud> well, maybe then quickly we hit on each <pbryan> Alright. <brianfreud> Movement as Opus: there is no granularity. "Work" == "Opus" <pbryan> So, it's just the high-level case, it's an opus. No sub-works. <brianfreud> Well, I guess it all also depends on what you intend to do with granularity. <pbryan> We have not covered how one might order or organize a list of works, nor do I know that there will be any facilities within NGS to do so. <brianfreud> in general, that's all these boil down to. how to organize works, and how many super-levels one is part of/sub-levels each is broken into <pbryan> I doubt this will be within anyone's capacity in NGS 1. Post-NGS I certainly think the complexity of managing works will inform the next steps. <brianfreud> I think my real answer is, until we know what is being asked by "granularity", it's not really answerable. at least w/r/t NGS, not post-NGS <pbryan> Well, I think we've demonstrated the upper and lower bounds; the question you seem to be raising is one of what's in the middle, and how do we organize/manage/visualize it all... <brianfreud> I wish I could find the schema luks had done for NGS back after that summit; he had an idea there re: RGs. They weren't like today's 1-level of granularity RGs. Instead, they were infinitly encompassing. At one level, perhaps you had all things like today's RGs. Then you might have a level above that, still an "RG", but with all volumes of a box. Then maybe one RG even higher for all boxes in a series. Etc... <pbryan> Wow. <brianfreud> I see "Opus", "Work", "Movement" - whatever you call it - as having the same role to play. <pbryan> Yes, indeed. And there is still a "work type" table, which the works table FKs. <brianfreud> yes, but there we'd still be defining specific types of works. this didn't. It really was just a concept of infinite numbers of containers, each with its own local best use. <pbryan> RIght. Any other points on granularity? <brianfreud> So "Opus" == "Work", 1 level. But take Beethoven's 9th, movement 4, you might have 5 layers before you even get out of that movement. :P nope <pbryan> Okay. I think clearly there needs to be a hierarchy of works. <brianfreud> isn't hierarchy the same as ^^? or at least, where we ended up re:granularity? <pbryan> Well, it's implied. We did discuss whether derivative works should be represented in a works hierarchy. I think the answer right now is no. Duplicate ARs where needed, but one work will be distinct from another. The idea of saying one work's composition AR is identical to another work's creates too much complexity, IMO. <brianfreud> I don't think this really is hierarchy, though. Isn't what you're talking about simply one potential work-work AR? <pbryan> It is of a sense, just not the super-part, sub-part; more a taxonomic hieararchy. <brianfreud> {{Work}} is a {{some attribs}} version of {{Work}} <pbryan> Anyway, unless you disagree, we can move on to opera/classical implications. <brianfreud> And I definitely think we'll need some variations on ^^ AR Oh, wait you mean inheriting the ARs from one work to any derived works? I don <pbryan> Right. Too hard for now... <brianfreud> I actually think that's a good idea, though perhaps (early) post-NGS <pbryan> Okay. Definitely grist for the mill. ;-) <brianfreud> lol, nor the first it's been discussed... it was kind of assumed during clean up CSG :P <pbryan> So, opera/classical... I saw your comments about maybe having the entire libretto one day and linking to the portion where it starts. An interesting vision of the future, to be sure. ;-) <brianfreud> I think that's the penultimate solution, though perhaps too complex for joe user. Copyright might be an issue, but I doubt it; just about anything written in "modern" (ie, copyright is in force) times, there are defined split points. So the issue really is only pre-copyright-in-effect <pbryan> And, this won't be works as we know it. So, for now, for works... <brianfreud> for now, though, I think the concern is valid, but rarely the actual case. <pbryan> I presume that we're not going to put every possible phrase a track may begin with so we can link a track to it. <brianfreud> I hope not. <pbryan> Same here. <brianfreud> yes, split points in opera can be whereever, but in reality, that's not the case. <pbryan> The solution seems to be that for tracks in an opera that are pure dialog for example, they may not link to a fine-grained work. <brianfreud> When I worked on all of Mozart, I think he was either #2 or #3 for total largest # of tracks in the db... and it wasn't a real issue there. <pbryan> Mozart is a special case, from my experience. Almost all Mozart operas seem to split along the same lines. Very rarely do they digress. <brianfreud> Well, I've not run into it in Wagner much either Mozart tends to always be split along the NME split points <pbryan> I've seen it all over the map with Donazetti, Rossini, Bach... <brianfreud> I think Bach is pretty standardized as well. However, I think we may be missing the larger picture. ie, what is it we're actually trying to represent with the work that is being linked? <pbryan> Right. <brianfreud> Are we really trying to link to the specific libretto that happens to be on track X? <pbryan> The answer seems to be by consensus: no. <brianfreud> I think maybe if we instead think of it as akin to a sample. <pbryan> A sample... <brianfreud> When Foo samples work Bar, we say "Foo sampled Bar", not "Foo sampled Bar from 2:10 to 2:14" <pbryan> Okay... <brianfreud> So perhaps we should simply have something like... here, take a quick look at that sandbox wiki page again down in the opera example ok, we have No. 21 Aria "Il mio tesoro intanto" (Don Ottavio) <pbryan> Yes. <brianfreud> then No. 22 Duetto "O statua gentilissima" (Leporello, Don Giovanni) <pbryan> Yes. <brianfreud> in this case, it's all the same act and scene, between those 2. <pbryan> These are clearly distinct works to me. <brianfreud> But maybe what we're really trying to point to, between those, is the "scenelets" between them. <pbryan> Right, often a track will be the brief dialog between numbers, or even be the brief dialog, followed by the actual number. If it's just the dialog, I'm inclined for it not to link to a fine-grained work at all. <brianfreud> So Act II, Scene XI (Don Giovanni, Leporello), then Act II, Scene XI (Don Ottavio), then Act II, Scene XI (Don Giovanni, Leporello, Il Commendatore), then Act II, Scene XI (Don Giovanni, Leporello) I don't mean the dialog. <pbryan> In the case of dialog and number, link to the number, which is a distinct work. <brianfreud> However, however opera is split, I don't think it *ever* splits *within* a "scenelet". <pbryan> (the dialog being ignored) <brianfreud> I don't think that really works for opera There's still music there. <pbryan> Indeed, and it's a part of the coarse-grained work for sure. But I don't think there's a meaningful fine-grained work to attach it to. <brianfreud> Musical theater, dramatic theater, sure, I don't see the need for a work entry for non-musical dialog. Hell, I typically use [dialog] and not the composer when setting the TA for musical theater bootlegs with dialogue present. But Opera, with the music there, is I think different. <pbryan> Some operas don't have any numbers. <brianfreud> Ok, let me reverse the question. <pbryan> There are no distinct movements at all. I want to say Wagner does it this way... <brianfreud> Go back to point #1 a moment. How is the "thingie" between No. 21 Aria "Il mio tesoro intanto" and No. 22 Duetto "O statua gentilissima" not at least one work? <pbryan> What is this thingie of which you speak? <brianfreud> all of the libretto+non-numbered music which begins at the end of No 21, and ends with the beginning of No 22 <pbryan> You consider this a work? <brianfreud> I'm asking you to apply the working definition of a work to it and tell me where it fails. <pbryan> Well, my first question is, how would you entitle such a thing? <brianfreud> Let' <pbryan> BTW, getting close to the top of the hour. <brianfreud> Let's ignore the questions of how it would be titled, and what split points it has, for the moment Even if you don't answer right now, I think that's the question to consider. <pbryan> It's certainly written, and it's certainly a part of the larger work. <brianfreud> Also consider, there may be a different composer for that "thingie", one which is left unrepresented, if it is not a work. <pbryan> How we can (and whether we should) distinguish it as a fine-grained work is debatable. Alright. <brianfreud> (example of such: http://wiki.musicbrainz.org/CSG_Standard/Mozart/Der_Stein_der_Weisen ) yeah <pbryan> Cool, well, I apologize I can't continue for now. <brianfreud> imho, I think it is a work, though I'm not certain how it should be split or titled. np <pbryan> It seems like we'll need to have another one of these IRC chats, after I draft the first version. <brianfreud> Anyhow, thanks for the 1 on 1 :D <pbryan> :-) Thanks for the input.
_______________________________________________ MusicBrainz-style mailing list MusicBrainz-style@lists.musicbrainz.org http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-style