#musicbrainz

8)

and it's simple, true, to merely interrelate them. Much more complex, however, to interrelate them in the way we don't, such that edit don't merely have local scope, but 1 edit can fix multiple errors.

Huh?

Ok. Right now, each and every time "Take Five" gets entered, it has to be titled correctly, composition ARs set correctly, performance ARs set correctly, etc.

If, on the other hand, you had "Take Five" linked to a Master, which held the mastering ARs. That Master is linked to a session, which holds the production ARs. That Session is linked to a Performance, which holds the performance ARs. That Performance is linked to a Work, which holds the work title and the composition ARs.

So you want to set the composition ARs for all Take Fives in the database? Instead of 1 AR for each track, 99% of which likely won't be noticed, and which won't be voted on,

you set it for the work. Now it's 1 edit, not a bunch of them, and at a higher level, where hopefully, more people will see it to vote on it, and even if it expires in, it's now centralized, so 1 edit can fix it, not dozens to fix borked composition ARs on each track.

the model for the basic artist-track AR is simple. The model for that system, on the other hand, is much more complex.

...and much more powerful, yet less actual editing/voting going on.

So you're talking about the complex MB that doesn't exist. Ok, I didn't make that leap quite yet.

well, I'm including in the model all the elements we take for granted, all the parts we currently compress down to the existing elements, rather than compressing the model as well

that's why something like the parody AR's implementation bugged me; not that the implementation didn't work now, but that that implementation would be totally borked the moment anyone tried to uncompress it

And yes, that would be more complex. On the other hand, even though I'm fond of removing redundancy from the data etc. I'd also like to see the guy who creates a usable UI for that, and the guy who does it so that we can fill that data using what we get from the cover leaflet.

does it have to be that though?

Usable? Yes, please.

no, the part about from the liner

That's what most people have to go with.

If you don't have session info, you link it from the track right to the work - no session info, no inherited session ARs. Someone else later has all the session info, they could insert the session info, now the session ARs inherit.

Just like we do now with dates

And even though we can leave refining to people who aren't "most people", you still have to have some structure in it.

sure. I'm just suggesting that, considering 50% or so of all open edits are almost always ARs, that's an example where interface doesn't really matter, but the model very much does

Hi

Adjusting where people add ARs, putting them at the highest level possible, rather than always at the track level, would greatly reduce the number of AR edits anyone has to make.

Interface doesn't matter?

the interface doesn't matter if the elements themselves aren't even implemented... :P

BrianFreud: let's keep it current. isn't the backlog the major issue here? along with editor retention?

NGS? i've stopped holding mybreath

yllona: :P

yes, but that is what I'm talking about. Why is there a backlog? Why is the open queue so huge that noone, no matter how dedicated, can now get through the entire queue?

A large part of it is edits that enter data we already have, just somewhere else.

and i don't think the existing NGS schema will be viable, by the time it's implemented either

well, track merging is coming within 2 months, and works maybe within 2 months

those 2 alone will go a long way towards removing a lot of redundancy in ARs

BrianFreud: that's a simple answer. its too easy to enter a bad edit. there's no mentor/vet process for adds, too much ad hoc automation

I usually add ARs to "the earliest version".

And then use cover/release/version ARs from there.

yes, but moving those to "the work itself" is cleaner, no?

"earliest version" is so often so difficult to correctly identify, and should someone later identify a new earlier release, there's no auto-migration capability for those ARs. The work as the target avoids those issues entirely.

I don't actually know what's in the edit queue. I go through my subscribed artists, but that's probably not the best way of getting the overall picture. 8)

lol, 1 sec...

http://wiki.musicbrainz.org/ChangeDefaultDataQu... (and http://wiki.musicbrainz.org/ChangeDefaultDataQu... )

earliest release is not that hard to identify, if you know your music

btw, 34 weeks later, we've averaged 2,553 edits per day since then

BrianFreud: and how much junk?

yllona: I just added a release where most of the tracks are composed by "trad."

yllona: Good luck searchign for the earliest recording.

srotta: and that's what institutions like the library of congress (and it's equivalent) are for

between just the releases I added today, there's at least 25 covers, everything from hair metal to japanese classical - finding the "earliest" for each would be so time-intensive for minimal benefit, I just don't bother.

yllona: That already makes it hard to identify.

srotta: forgive me, perhaps it's my art history background, but establishing provenance is part of the game ;)

meh, I'm all about linking it to the work in theory, which to me makes more sense anyhow.

yllona: And we're talking worldwide here, I don't have an idea where to start looking for Georgian folk songs.

srotta: the vatican

(No, AFAIK that release didn't have those).

yllona: I think you perhaps miss the point

nope

given a single track, sure, you could likely find the earliest release, given enough time to do it.

pick something and clean it up. i'm about quality not quantity

But that's time not doing other ARs, like performance, production, composition, etc, or doing other types of editing, or voting.

yllona: Oh, well then, that solved the problem.

and it assumes there is an "earliest" to link to in the first place :)

srotta: i'd be happy to help you with any sort of chant (gregorian or otherwise)

yllona: Georgian, not gregorian.

8)

i said otherwise

yllona: My favorite so far is "Smile" by Chaplin.

Easy, everyone knows it, but I still don't know where it was first released.

smile? one second

BrianFreud: And yeah, I agree with works. Tracking down earlier versions is interesting but sometimes also frustrating.

And not really intuitive.