Suggestion: Stricter rules for adding cover art

cover-art
Tags: #<Tag:0x00007f23c655f320>

#1

People adding cover art that may or may not be correct, with no source or often no comment at all, is one of the most frequent annoyances while going through subscriptions. Too often, it appears that the user has just made a web search or such and uploaded the first result to some version of the album, which is often wrong.

If I may suggest freely, the help text when adding cover art could posit the following rules:

  1. The comment field must not be empty.
  2. The comment field must contain the source for the image.
  3. A brief description of what constitutes a good source and not (e.g. “own scan” is OK. A link to image source is good. “Web search” is not enough.)

Any thoughts?


#2

There are some edit types already for which leaving an edit note is mandatory. As far as I can tell, the results are not too promising. Those people that are too lazy will, instead of leaving no edit note, simply leave useless edit notes (like “…”) or even deliberately wrong information. For release additions, there was originally this message:

You must provide an edit note when adding a release. Even just a URL or something like “CD in hand” helps!

We had to remove the second sentence because people were just writing “CD in hand” everywhere (even for digital releases etc.).


#3

Maybe a better alternative would be a longer voting period for certain kinds of edits, or even requiring some kind of community response before applying certain edits? (I realize the latter is probably not very practical, but we can dream…)


#4

Yeah, but if there were such rules, we could at least immediately vote no on such obviously non-compliant edits like “…” and point the user in question to those rules. In extreme cases, there would also be the option for admins of inhibiting their further edits. But if there is nothing but a soft recommendation to enter a source, it’s more difficult to motivate shooting down the edit. But maybe you’re right and the sad truth is that everyone would start writing “own scan” for everything and the situation would be no better. :frowning:


#5

I’m not a fan of the latter idea though, since it should still be possible to get cover art added for artists with few subscribers. If I have a pet band that nobody else cares about, it’d be annoying if I had to chase for supporters for my edits.


#6

I agree it wouldn’t be the ideal solution. I just can’t see a way to improve album art submission without some kind of actual human intervention.

Maybe rather than vetting every artwork, it should be by editor - some kind of stricter filter until a person demonstrates they’re consistently adding correct artwork.


#7

maybe require a vote from a certain percent of subscribers to the item in question? Even auto apply if there are no subscribers?


#8

I’ve added quite a bit of album art—all of it scanned myself—mostly to classical releases. I have 901 add cover art with 0 votes, 2 with only no votes (both on the same release, and canceled), and 248 with at least one yes vote. IOW, about a fifth of my add cover art edits attract a vote.

So requiring that at least one review be obtained would be problematic.

OTOH, since cover art is effectively applied immediately (even though the vote is still open), I think it’d be fine to change the 7-day approval period to something longer. That’d give more time for anyone to object—and one person voting no is all it takes to make the edit be rejected, correct? (Provided he/she is not outvoted, of course).

And if even after two weeks, or however long, not a single person has stopped by to check—then I think our actual problem is a lack of people who care about that artist/label/etc.

(Another thing: it’d be nice if it were easier to move cover art to a related release).

[I’ve certainly run in to some annoying cover art uploads, mainly of annoyingly tiny variety].


#9

I wonder how many things have subscribers who don’t actually pay attention to them—I I have a bunch (entering a lot of details on classical releases = a lot of new artists created + default of subscribing to any artist you create). I’d clear them all out, but there is no select all button :frowning:

[Found a solution: v = document.getElementsByTagName(“input”); for (i = 0; i < v.length; ++i) { v[i].checked = ‘checked’ } ]

…, anyway, that could throw off a X% requirement. Not to mention, if there are subscribers paying attention—why didn’t the bad edit get a “no” vote?


#10

You can use POWER VOTE (obsolete version but where you can see doc and screenshot), that you can install from one of:

It is the same principle than your code but with vote buttons, and it will not erase your existing votes, except with the shift button override.


#11

Making begining to edit easier and reducing the learning curve gradient would give us more editors who edit for the fun of editing rather than having their tags “good enough” for themselves.

This might be achieved by doing UI work, getting greater server reliability and developing a “chunked” introduction pathway for learning to edit.

With more people editting for fun then checking other’s work and voting becomes more likely.


#12

@IndridCold

  1. Would be useless because everyone can enter whatever he/she likes. What about “Comment for IndridCold because he/she wants this field must not be emtpy”?

  2. The source for the image is useless, because even a link to a reliable source like Discogs can be wrong. I prefer a web search if the image matches the release then a Discogs url pointing to the wrong release

  3. A description doesn’t help to get a more correct image.
    A description is only useful, if it helps for voters to decide why such an image should be accepted even if something is not perfect (“bad picture quality”, “missing sticker from wrapping”)

IMHO this steps can help improve the uploading of correct images:
a) Write clear visible rules about the minimal size, quality and format for images on the uploading page
b) Write all the point that should be checked BEFORE uploading an image, like identical “number of tracks, sort order of tracks, release year, release country, format, label, stickers etc.”


#13

I have on two occasions entered an image of a cover that did not match the tracks - but I had the CD in hand; recently unwrapped after having purchased it. In that case I did write a note indicating that cover and the track data didn’t match; but that I had double checked and it was in fact the correct image.


#14

@dukeja:
Why do you not correct the existing tracks according to your image?
Or add another new release matching your image?


#15

Because the error was in the cover art itself. There was printed track information on the CD itself that was correct. And I had other sources to indicate the situation. The track data was correct and verifiable. You’re not suggesting that I should modify the cover art are you? In the case that I had - there was no existing data at all about the release. It was entirely new.


#16

Yes, it might have to be “active” subscribers rather than all subscribers.

As for why the bad edits might not be downvoated, maybe no one noticed if the difference is subtle, or maybe people feel that a representative image is better than no image.


#17

I confess I’ve never paid much attention to my subscriptions. When I started getting notifications about them, i was a very new editor and didn’t know how I’d gotten subscribed to anything in the first place. Much later I figured it out, but since they come to my email I rarely think about them when I have the opportunity to do something about them. I would be much more likely to pay attention if there were a clear notification when I visit musicbrainz to do editing.

As for subtle differences, there are some cases where something about the way the image was produced has produced distinctly different images, both of which are correct. The example I can think of is https://musicbrainz.org/release/41faec46-867e-4b3b-9a50-19ea6dc225e7, where I argued for keeping both images. I own this CD and the gold part which is very dark in the second image is foil embossed, so the first image seems more “true” to what I see, but the second is accurate as well.


#18

I disagree with this. If someone uploads the cover art from Discogs and it’s later revealed that the Release was linked to the wrong entity on Discogs, you would know that the cover art upload was also sourced from a wrong source. If there’s no source given at all, you have no idea where the image was originally taken from.

A generic web search is also not a good link, since the search results will vary over time - it is much better to link directly to the image you eventually end up using (and possibly describe how you found it).


#19

It is currently not possible, there is an improvement request to search in edit notes.

It is already possible to search for edits opened by untrusted editors. (Edit: This was an unintended bug rather than a public feature and it has been made unavailable since.)


#20

… then if we actually got people to vote, is there any reason to think they wouldn’t just vote yes? Seems like maybe putting some instructions on the review edit page would help.

Or maybe just give in and let people uploading images check a box that an image is representative only—then you could just ignore those.