Keeping or dropping Amazon as a CAA fallback on MusicBrainz.org


#1

Are us users allowed to add some feedback? Please move this to any relevant new thread if I am posting in the wrong place.

PLEASE don’t remove the Amazon Artwork Link. That is a very useful and clever backup for when the artwork isn’t in CAA. Maybe make more emphasis the artwork is being leeched, but assuming MB isn’t charged for this Amazon artwork is better than nothing.

There are also a lot of releases I have seen where only very limited artwork has been added. We all use this database in different ways. And it is very handy the way the Amazon link is used as a backup. A very clever trick.

In the last weeks I have been trawling around releases that are clearly very old entries in MB. Many of these only have that Amazon link for their art.

(This post needs to be split to a new thread… I bet there are many people out here using artwork in different ways… a deeper discussion is need :))


Notes from #MetaBrainz Meeting 2018-06-11
#2

When using the Encyclopedia I don’t care where the CA comes from.
Just that there is accurate CA.

Even now with Amazon CA supplied, on an aesthetic level, I find MB overly text-based.
Steps that result in a reduction in visuals would be a step away from appealing to many potential encyclopedia users who have an aesthetic sense that goes beyond display of text.

If a new and better source/s of CA for Releases where CAA is lacking can be found and displayed then I’m all for it.


#3

Keep in mind Amazon cover art can be uploaded to the CAA, which means there shouldn’t necessarily be any problem of having less coverage.


#4

I like it to keep Amazon image as a fall back for the many many releases that we don’t have CAA for.
And uploading the Amazon image in CAA is bad idea most of the time as the cover is not good.
I think hiding the fall back will result in more people uploading bad Amazon images to CAA.

I have by CSS the display of CAA looks better than the fall back (flat image), we could use that in MBS to show that it’s only some utilitary display, not how the actual cover may look like, exactly.

Shadowed display for a CAA release:

image

Flat display for an Amazon fall back release:

image


Notes from #MetaBrainz Meeting 2018-07-30
#5

Yeah… it can be uploaded. But if someone turned off the clever bit of code that automatically goes and gets this artwork, then suddenly there will be thousands of releases with No Art.

To me that seems a backward step.

No one is going to go through the whole database replacing all this Amazon art. So we’d suddenly have lots of missing information. (I assume that would be many thousands of releases)

If it is disliked, then make the comment underneath it bigger to make it clearer that it needs replacing.

Nudge people into uploading better quality, usable artwork. There is another thread near here were someone was uploading scanned images of the rear of a cover only. It did show how some people really don’t use the artwork at all. Yet to me it is a VERY valuable part of this database.

In the last weeks I have been trawling through some items which have barely one or two edits and that was many many years back. No one has updated those records since. If the Amazon artwork link was killed off then these records would loose out.


#6

Could I add my vote to keeping Amazon. In addition to the points already made, I find that there are times when CAA seems to fail entirely.


#7

I don’t like it because it makes many releases have the wrong cover art. This is due to the overwhelming number of Amazon links to the incorrect releases. Is there a way to keep it available in Picard’s options without it displaying on the release pages of MB? Also, if the link on the cover art wasn’t showing on the release page the correct cover art would likely find its way to be uploaded much faster.


#8

(I find CA and images of Artists and Label imprints to be a necessity for any mature, popular encyclopedia of recorded music.)

I value the presentation of images on MB pages highly.
Rather than reducing the (in my user experience) too small number of images that are currently presented even further, I suggest that Amazon images (or some other source) be made selectable as RG images and sub-titled as such on Release pages.

Or encyclopedia users would just go to another site which looks “better” (more images).

In real life (RL) there would be a mix of;
uploaders who can get access to the correct images (there are many many Releases where extensive searches produce no CA at all and even more Releases where the only CA that I can find is Amazon),
people who make no change in their loyalty to MB due to the absence of graphics,
those who go elsewhere due to the absence of graphics, and even those who,
find a text based presentation of meta-data desirable.


#9

I do to, but only as long as it’s the correct image which seems not to be the case due to all the inaccurate Amazon links. If the correct cover art is uploaded to CAA, than there is no need for the Amazon link. I’m way more concerned with info being correct than visuals that are incorrect. Just my opinion.


#10

Slightly off topic, can anyone tell me why some releases with Amazon ASINs get cover art and others don’t? For instance this Bill Monroe boxed set that I was just working on shows as “No cover art available” even though it links to an Amazon product page that has cover art.


#11

When the release gets either a bar code or an ASIN, you have to wait quite some time before a batch crawler will make a cover available in this release’s side bar, it is normal.

※ Maybe the Amazon cover from bar code no longer exists in MBS.


#12

So that is not a fault of the image, it is an error in the ASIN which needs correcting. If anything, by showing the wrong image in those cases it is a nudge to the user to correct the ASIN. If there was no image then the user would have less chance to know the ASIN needed updating.


#13

Either that or Amazon’s data is wrong.

It could be, or on the other hand it could cause further confusion. For example, the user may see that the image is different from their copy and so create a new release, rather than check the Amazon link. Or, worse, they might change the other data to match what’s on the Amazon page, making the situation worse. Showing the image from the Amazon link by default may give a false sense of definitiveness to a single piece of data that is often entered fairly casually.


#14

This is why I think we should keep it but we should, in its caption, emphasise that « This Amazon image is just an illustration and is not representative in any ways of how the actual release looks like. »


#15

Picard’s usage of Amazon CA and MusicBrainz.org’s usage of the same is entirely separate. Removing MusicBrainz.org’s usage of Amazon CA would have no effect on Picard.

I think the people that would do that are/have already doing/done that. 40% of our releases have CAA images attached, 29% have Amazon links (of all releases, the 31% listed at the link are of all Release-URL relationships), and there is certainly some overlap so Amazon CA is used for less than 29% for sure—not to mention the instances where our Amazon CA URL algorithm fails, meaning that actual Amazon CA URL coverage will be even less.

Because we can’t use Amazon’s API (I think; I don’t remember the exact reason) and because their cover art URLs don’t follow a universal, consistent pattern that we can refer to, which means it’s a bit hit and miss whether it’ll work or not (the current solution mostly “hits” though).

Just as often, Amazon just plain has the wrong image, even if the MB Release→ASIN link is correct.


I feel like the ones here (strongly?) defending keeping the current Amazon CA resolver uses arguments that would just as well apply to automatically getting/using cover art from pretty much all other stores (iTunes, Bandcamp, SoundCloud(?), Google Play, …) that we’re not currently supporting, but no one has actually suggested that we do(?). Supporting these features means developer (employed or volunteer) time gets spent supporting them (e.g., MBS-9735), that could have gone to do other things, instead of constantly having to play catch-up with services outside of our control.

Obviously, I’m just one person (and MB user/editor) though, so if most people (strongly) disagree with removing this, then it won’t be removed. :slight_smile:


#16

I am fairly confident that less than 5% of the world’s population have visited the MB encyclopedia.
(Can you give me a more accurate number?)
Which leaves some 95% of the world’s population up for grabs by music metadata sites.

This is a good argument.
I’m concerned that the unattractiveness of a overly text-based music metadata website is not being given adequate weight in terms of attracting new users though.
It seems that for the forseeable future only Artists with WP pages with an image that meets WP’s stringent image requirements will get an image on their MB page. And Label imprints seem in the same bucket.

I could live with an even uglier MB in the short and medium term, if that was a trade-off for a longer term MB with many Artist images and Label imprints.

But I’m not hearing that non-WP Artist images and Label images are on the roadmap.
Rather I’m hearing that even less pages will have images on them.

If I’m wrong about the roadmap then I’d be wrong about the weighting of the benefits of Amazon images.

This issue seems to be one of those where tensions between coders and bon vivant s emerge.
Whilst very much a bon vivant type, given to dancing with butterflies, and making very basic errors in data entry, I am impressed by the system and stability that the server-mole types have produced. My warning is that a stable music meta-data site that delivers high quality metadata is unlikely to succeed in the medium or long term if it looks too much like a spreadsheet.


#17

Exactly my thoughts, 100% ACK!


#18

https://tickets.metabrainz.org/browse/MBS-4628
https://tickets.metabrainz.org/browse/MBS-4683

This is one of the reasons that one of MetaBrainz’s current focuses is “fixing the MusicBrainz site design for an improved user experience”:

I’d suggest to read that blog post and get involved as described there.


#19

These would be good evidence if they were assigned and alongside a statement by Rob.
As those tickets stand I hope they happen but I don’t count them as “on the roadmap”.

But convince me that the current resources devoted to the temporary fix of Amazon CAwill be diverted into UX and beautiful pictures and I’ll change.


#20

I am certainly not suggesting expanding to those other shops. Just not removing what we have now.

I understand the “not wanting to waste developer time”. I am not suggesting that. Just asking not to remove something that is “better than nothing”.

(It is a little funny that the [MBS-9735] you link to is actually a bug that is creating the feature you want… if the ASIN lookup is failing on Japanese Amazon for this guy you want to see them to scan artwork and upload it instead. As can be seen in the notes in that bug report users are often too lazy to bother. And that user is actually complaining about having to upload artwork. Not everyone has a scanner to hand)

I also agree that Amazon is not a reliable source for artwork. They don’t have the selection of geeks like here at MB and over at Discogs keeping the quality up. Personally I think artwork should come from scans of CDs that are in hand. Or a real source noted when the images are uploaded.

Some of us really care about our artwork. And want to see that overlapped into this excellent MB database. You quote that 29% of releases here on MB are using Amazon artwork. To me that is a LOT to throw away.