Editor "upvoting" on releases and data


#1

After replying to a fellow editor on this release:
https://musicbrainz.org/edit/52663480
I felt something I said should be proposed.

In this release shown above, this is a release that is no longer available from iTunes. It is there still, but under a different ID and a different cover. Basically, the prior is before her new label stepped in and made a new release of it.

So, the issue I have is I have no reference to provide, well, not a valid one anyway. I have metadata to tell me what the reference was, but it cannot be verified by a non release owner. This is especially bad in the case here… where the cover art looks like a generic screenshot of a video or something. But as an owner of both versions of this release, I can say that the data and artwork I provided is what was provided originally by iTunes under each of the specified IDs.

What would be great, especially in cases like this where there is no good reference, is to have the ability for other users that may own the same release to upvote, or vote, or what ever you may wish to call it… but in some way indicate that they have the release shown here and are confirming the accuracy of said data. This would be nice because without it, you trust my word as the editor. That is weak. But if other owners confirm it, that adds to the credibility of the edit. For those who might understand this part, sort of like how strength of public keys are done in things like PGP, or other personal identification digital signatures, fingerprints, etc… where the more users that sign your key with theirs, the more support as being valid your key has. Also… like Accurate Rip database and what I hope AcoustID database becomes… most should be able to relate to that.

I am posting this here in order to get thoughts from others on how such a thing might be done, if others think it might be good and how it might work to make a proper proposal.


#2

can you do an archive search to find old versions of sites?


#3

I can only speak to my experience with the releases I have added with this issue. Yes, you can use archive.org to do the historical search. However, none of the ones I have added have been there.


#4

I would like to add to and clarify my query here. Instead of discussing reasons not to do it, I am looking for pros and cons of doing it.

I see pros as a new level of authenticity, verification, proof, etc of releases and their data. I see a con as someone needs to do some work to implement it int he system. But I see no con as it may impact MB as a system. It could only benefit, and that benefit is and will be limited by its use by editors. That is the reason of this query. How can it be, or could it be, put in place to achieve a benefit worth the cost of labour to implement.


#5

The following looks negative. But I think maybe you can present significant benefits.

In the situation where the “vote to verify” system is in place, after time and resources have been put into coding, and documenting it, and an additional complexity added to maintaining, learning, understanding and editing the system, how will things be better? What will work better, easier, more reliably? What problems will be reduced?


#6

The disappearing internet references is an issue in its own right, but what is to be done when it used used to prove the existence of a release, in this case a digital download.

I ran into this issue 2 weeks ago when I was adding the first digital downloads I have ever bought. I bought and added https://musicbrainz.org/release/e624a585-908c-4d50-be0f-edbcc98e8f10 which is the digital version of the CD https://musicbrainz.org/release/2fabfa54-41f1-4db6-bf5e-7c806c589ae9 but missing what was track 3 “We Can Work It Out” which is a Beatles cover. The CD release had a link to the older digital release of the full album which was no longer valid and I removed with the documented edit https://musicbrainz.org/edit/52412528

I did not like removing the link since it did point to what had been a valid download, but the release page should not contain links that do not exist. So what happens when someone who bought that download wants to create a release for it? Maybe a historical link page with documentation on the disappearance, I do not know.

URL’s for downloads may come and go, maybe some of the web sites will have historical information that can be pointed to or will keep them around but say something like “not available anymore”. The industry does not care and cannot be counted on as a reference point. The original post asks a very good question.


#7

I can try :slight_smile:
In other examples of where this is in use, like Accurate Rip, this plays a key role in the system and its validity. As one marks/confirms a result, that mark is added to the tally. So when a user looks at it, they see that, lets say 4 users confirm it. This is nice, but if that 4 users was 40 users, there is a greater level of confidence that what is there is actually correct.

An example here… I could add a release and simply say it is real because I have “CD in hand”, which is way too common here. That is meaningless to me. As an editor, I may look up this release and see it is already there. I could then check it and say I confirm it., thereby telling other editors that I place my “stamp” on it in addition to the adding editor.

I really see no need to down vote, if you disagree, it very well may be because you do not have that specific release. The benefit comes into play with editors who agree/support the release as it was entered. With a CD release, this is not so important as you can upload images that basically tell all, wherher the CD is still for sale or discontinued, the image shows it to you. But on digital releases, this is just not possible.

EDIT: I guess I did not put this in this post. There is an issue with places like iTunes. Using iTunes as example, when an ID is no longer valid, there is nothing to reference. At this time, there is no good source of data for expired references, as it applies to digital releases. Physical always has the picture/image of what is in your hands, digital does not. As much as I may complain on MB and digital releases, one thing MB can really step up and do is provide a nice historical picture of digital releases where the reference is no longer valid. People do buy these releases, and once that link is gone, so is the source data. MB could be capturing this, and using this “upvoting” as a means to support the correctness of the data.

Why should that even be needed? Well, I can edit my metadata, or download some pirate copy and use that metadata as real. So I think if we can find a way to have users commit to data, it is a benefit. I guess I then need to go back on a statement… maybe a user should be able to say I disagree. If I have a iTunes release with ID 123456 and I say what is here is not what I have, that should be a downvote. Because the editor who added it might have had bad data. I need to remember not to assume that all people are in possession of real releases, authentic releases. Most all statements I make use that assumption.


#8

You can set the link as no longer valid and leave it there, vs removing it.

EDIT: Sorry, it is setting it as ended.


#9

I should have asked, it is gone now, but good to know for the future, I dislike removing history.


#10

Totally agree. I am happy someone told me this, I forget who, but someone deserves that credit.

@dashv - after a failed first attempt, I added your reference link back in as ended.


#11

These factors go to the reliability of MB data.
A “vote to verify” system would make problematic Release data more obvious.
(There is a related idea of “machine voting for MB DiscID” which would count,record and display the number of times a specific MB DiscID had been found for a specific Release.)


#12

I love this idea. In fact, I presented a similar idea to the tally of acoustIDs. Currently, Picard only allows an ID submitted once. I suggested that a tally be kept, and duplicate acoustIDs be allowed.

Is this bad, or is this good? I am unsure on your context.


#13

I am trying to sort out that edit which may be wrong.
I will contact you directly so that we do not mess up this important discussion.


#14

If you could confirm at the release level, what does that mean if additional data is later added? Wouldn’t the confirmation have to apply to a sort of snapshot of the release at that point in time?


#15

This is a very good point. Let me check on some other databases for how this is handled, so I have some examples of implementation to refer to.

One example I know off hand is versioning, where I “stamp” an entity at a specific version. Another is locking, where once an entity or set of data is verified, it is locked requiring a moderator/admin to make an edit (or unlock it). Neither of those two really apply to MB though.

I could see the locking work in a way like this…say that a release ends up locked. The reason is not to matter for this, it is just locked. So in order to make an edit, the editor would propose an edit. The difference here is that the edit requires, lets say, 3 yes votes to pass. If it has not yet received the required yes votes, it remains pending, or could be killed by no votes. Only issue here is the web UI for this entity would need to alert viewers of this pending edit, if for no other reason, to avoid conflicts.

I am just thinking out loud here, none of this is to be taken other than some thinking to potentially make an improvement.


#16

I think it would be good.
A RG with multiple digital versions of “Beyonce’s Greatest Hits” could be looked at and have the commonest ones be relied on as having existed. A version that has no verification and track 1 as “Home on the Range by Will Rogers” etc, would not be so reliable