Page MenuHomePhabricator

Share A Fact images should include license or attribution info about the image
Open, LowPublic

Description

Share a Fact cards should attribute the author of the image and state the full licence of the image used in the card.

Related tasks:
T95379: Make Share a Fact cards only use free images
T164957: Show complete license info for images shown in full screen

Event Timeline

Ragesoss raised the priority of this task from to Needs Triage.
Ragesoss updated the task description. (Show Details)
Ragesoss added a subscriber: Ragesoss.
Krenair set Security to None.
Krenair added subscribers: Krenair, WMF-Legal.
Deskana claimed this task.
Deskana added a subscriber: Deskana.

In terms of satisfying our legal obligations for attribution, I've been in close consultation with the Legal Team about this and we have complete sign off from them on the current feature.

The default text that's shared with the image does link to the page that the image is from, so attribution is possible in that manner. As a Wikimedian, I agree that our lack of consistent attribution leaves a bad taste in the mouth. But I'm not sure what we can do in a manner that's appropriate for the medium, though. Putting a big wall of text with a URL in the image might make us feel slightly better about it, but really we're not providing any more value to the user, since they would still have to hand-type the URL into their browser, and it may actually detract from the experience rather significantly.

Could we embed attribution data as EXIF?

It seems like if we're going to bother mentioning a license at all for this little snippet of text, then we should do the same for the image. Conversely, if we're treating the image as fair use, why not do the same for the text and avoid the CC icons altogether?

This seems like a not-quite-coherent approach in between.

Could we embed attribution data as EXIF?

Nice idea. I've heard that Twitter strips image metadata, so practically speaking that might not get us very far on some platforms. But it's worth investigating. I've filed T91821 to track that.

It seems like if we're going to bother mentioning a license at all for this little snippet of text, then we should do the same for the image. Conversely, if we're treating the image as fair use, why not do the same for the text and avoid the CC icons altogether?

Adding those logos was an explicit ask from the Legal Team (see T86843), so I'd ask for their thoughts on that. @Slaporte may be able to provide the rationale.

My view on having these icons is that it's important for awareness as much as it is for attribution, but I am not a lawyer. :-)

Legoktm changed the task status from Resolved to Declined.Mar 17 2015, 9:35 PM
Legoktm added a subscriber: Legoktm.

Changing status to reflect intention.

Could we embed attribution data as EXIF?

Nice idea. I've heard that Twitter strips image metadata, so practically speaking that might not get us very far on some platforms. But it's worth investigating. I've filed T91821 to track that.

See T20871

Why does it mean we'd have to include a URL? Isn't the normal case more like adding "User:Kangaroo13" or "Tiglath J. Pileser" after the CC icons?

Image attribution is harder than it sounds.

This parallels a lot of the issues surrounding Commons metadata. Where would we draw the attribution data? Image uploader is not satisfactory if the uploader merely carried out the task of uploading, and the image came from somewhere else. Specific author data is not consistently presented on Commons, either, since it cannot be easily extracted from the database or an API.

This parallels a lot of the issues surrounding Commons metadata. Where would we draw the attribution data? Image uploader is not satisfactory if the uploader merely carried out the task of uploading, and the image came from somewhere else. Specific author data is not consistently presented on Commons, either, since it cannot be easily extracted from the database or an API.

Well, these questions have been hashed over and over with MediaViewer. There *is* an API to query author data (CommonsMetadata), and the File metadata cleanup drive is there to get more files compliant. So from my perspective, the infrastructure is there and the data (mostly) as well − the Share A Fact feature does not appear to make use of it.

Well, these questions have been hashed over and over with MediaViewer. There *is* an API to query author data (CommonsMetadata), and the File metadata cleanup drive is there to get more files compliant. So from my perspective, the infrastructure is there and the data (mostly) as well − the Share A Fact feature does not appear to make use of it.

That is an oversimplification of the problem. Take the case where there is an HTML table as the author's name. This is possible, thanks to Commons. How do we put that in the card in a way that doesn't detract from the experience? Also, for technical reasons, native apps cannot render HTML in any meaningful fashion inside a card, so actually using the HTML is also not technically feasible.

This class of problems are related to MediaViewer, but they are different because we're also dealing with highly limited space in the card.

The fact is that this problem is nontrivial, both from a product/design and a technical perspective, and we weren't willing to block releasing a feature that is totally legally sound while we figure it out.

Also I've said this before, but it bears repeating. It might make people wince when I say it, but 99.9% of people using the Wikipedia app do not know what image attribution even is. As Wikimedians it's pretty core to us, but for the vast majority of users it means nothing, and we weren't willing to detract from the experience for the majority just to make ourselves feel a little better.

What JeanFred said. The API does exist, and does work for a very large portion of the collection. So it should be used. Having edge cases, even many edge cases, should not be an excuse not to do even what is eminently possible.

As for the "99.9% of people" argument, I find it fallacious. A similar percentage also routinely infringes copyright elsewhere on the Web and in life. Should we then stop caring about licenses, because the overwhemling majority of our readers don't care? The community has always held firm about the importance of hewing to free licenses and, indeed, has made some inroads in educating (small but growing parts of) the public about free licenses and the importance of attribution. It behooves us to continue to do so with this feature to, despite, and only to the extent allowed given, the technical difficulties.

Finally, there is the issue of setting an example. Not providing reasonable attribution when we _can_ provide it (again, no one is asking for *unreasonable* attribution) is not a good example.

What exactly are you asking for then? Please be specific.

I've stated that there are serious technical blockers and limitations to providing author information on these cards. I'm not sure why that point isn't getting across.

What exactly are you asking for then? Please be specific.

The reverse of the bug title, that is: Share A Fact images should include license or attribution info about the image

I've stated that there are serious technical blockers and limitations to providing author information on these cards. I'm not sure why that point isn't getting across.

Not really. MMV is able to provide this information, so the metadata *is* available and apparently usable. You've made a claim that it is possible to have an HTML table as an authors name, but no one is even asking to support that use case.

Anyways, let's re-open this as an enhancement request rather than a bug.

Legoktm renamed this task from Share A Fact images do not include license or attribution info about the image to Share A Fact images should include license or attribution info about the image.Apr 18 2015, 1:04 AM
Legoktm reopened this task as Open.
Legoktm updated the task description. (Show Details)

I've edited the description to remove incorrect information. WMF-Legal has confirmed that the logos on the cards are accurate because the legal facts of the situation are that the cards themselves are under CC BY-SA irrespective of whether there is fair use material used in them. They also confirmed that there are no legal issues with the current implementation.

This task is purely about attribution to conform to some other, non-legal standards for attribution.

I've edited the description to remove incorrect information. WMF-Legal has confirmed that the logos on the cards are accurate because the legal facts of the situation are that the cards themselves are under CC BY-SA irrespective of whether there is fair use material used in them. They also confirmed that there are no legal issues with the current implementation.

This task is purely about attribution to conform to some other, non-legal standards for attribution.

Just to confirm-- the cards are generated from a CC BY-SA article, and the icons reflect that the card is free to be shared and remixed. It's acceptable to include fair use and public domain content within a CC BY-SA work. I'm still preparing a more detailed response on the attribution questions, and I will follow up here with a link when it's on wiki.

Thanks!

That is an oversimplification of the problem. Take the case where there is an HTML table as the author's name. This is possible, thanks to Commons.

You've made a claim that it is possible to have an HTML table as an authors name, but no one is even asking to support that use case.

Regardless of whether the field contains a (seemingly complex) HTML table or something as simple as a link to the author's user page. Neither is plain text. Attribution/author very rarely is plain text. But this doesn't matter. The way to display this is to simply extract text (could be as simple as strip_tags()). This is trivially done with one or two lines of code.

While Mediaviewer does run in a browser and could (technically) render HTML, it generally does not. For aesthetic reasons it, too, strips html and just displays the text. Depending on the view mode, Mediaviewer does go the extra mile and allows plain <a> to pass through (for clickability) but we don't need to do that here.

Even if Fair Use is the excuse to ignore the the CC attribution requirements, this feature has to be disabled in all language versions, where Fair Use does not exist (e.g. most of continental Europe).

And even in the english Wikipedia Fair Use policy is limited to "for critical commentary on # the work in question, # the artistic genre or technique of the work of art or # the school to which the artist belongs":

One may discuss, whether the non-attributed use of freely licencened images as article teasers in social media falls within that policy or not...

To use images as an article teaser, that were already on Fair Use in Wikipedia, is imho a clear copyright violation, because there is no commentary context whatsoever. Considering all the artist and work related articles in Wikipedia this is a serious thread.