Jump to content

Wikipedia talk:Wikidata/2017 State of affairs/Archive 8

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 5Archive 6Archive 7Archive 8Archive 9Archive 10Archive 14

Response by Jon Katz, Wikimedia Foundation, to recent Wikidata concerns

Hello All,

Sorry it took so long for a real response. There is a lot to unpack from the conversations taking place on-wiki and multiple stakeholders within the organization to consult. I really want to make sure we are speaking carefully and thoroughly enough to avoid a misunderstanding like the one that arose out of the recent withdrawn Wikidata description RFC. As evidence of that intent I will try to clarify a few things from the perspective of the team and ask for your insights in coming to a better shared understanding.

Looking things over there appears to be a few major threads of discussion.

  1. Why weren’t Wikidata descriptions turned off everywhere in connection with English Wikipedia in the context of the recent RfC?
  2. What are the Wikimedia Foundation’s plans for Wikidata here on English Wikipedia?
  3. What are the unaddressed concerns about Wikidata content appearing next to English Wikipedia content?
  4. Now that we know community members did not want Wikidata descriptions to appear in any reader interface for English Wikipedia articles, what should we do about it?

1. Why weren’t Wikidata descriptions turned off everywhere for English Wikipedia in the context of the recent RfC?

For this concern, see this reply. It was a misunderstanding.

We realize now that we've gone down a path that some may find confusing. Wikidata descriptions first appeared on visual editor, then mobile apps, and then the web. Along the way the use has always been seen as a navigational aid to readers to help them understand where they are quicker and with more confidence, not as integral part of Wikipedia articles.. It was not our intent to encroach into governance of the content. That being said, it is clear others see this differently. I don’t think the answer is definitive one way or the other.

2. What are the Wikimedia Foundation’s plans for Wikidata here on English Wikipedia?

Currently, experiments and plans for further Wikidata integration into Wikipedia projects exist, mostly driven by the Wikidata community and Wikimedia Deutschland, not the Foundation. I honestly don’t know much of the specifics of these, but one example mentioned was a Wikimania presentation by Wikimedia Deutschland about the historical success of Wikidata and where it could potentially go in the future. For example, in the linked presentation there’s a rather interesting prototype of what editing of Wikidata’s data directly from Wikipedia could look like - as requested by many editors on several Wikimedia projects using Wikidata’s data. The whole presentation provides a substantial insight into the work of the Wikidata community.

In addition to Wikidata descriptions, there are currently several instances where information from Wikidata appears alongside English Wikipedia content. For instance:

As many of you may be aware, the Foundation also supports efforts (like Structured Data on Commons) that leverages Wikidata to benefit the entire Wikimedia movement - beyond individual projects and their language variants.

While the Wikimedia Foundation has no concrete plans for integrating other data from Wikidata with Wikipedia, teams at the Wikimedia Foundation have been looking into ways to meaningfully integrate Wikidata for quite some time. This is because we believe structured data provided by Wikidata is strategically important and complementary to the content provided on Wikipedia. As other folks have mentioned, at this point, the Wikimedia Foundation has no specific plans in place for further integration of Wikidata into Wikipedia. If interested, you can read the Readers teams’ annual plan programs and goals to enable those plans for the coming quarter (the three-month period starting in October). That being said, the Wikimedia Foundation is supportive of the general idea of greater integration, and are eager to address potential concerns.

I believe that the benefits of using Wikidata as a separate repository, in the way the wikis currently use Commons, will eventually outweigh the drawbacks, but acknowledge we’re not there in many places. Any effort to further integrate Wikidata would have to address some of the serious concerns mentioned around control and quality.

3. What are the unaddressed concerns about Wikidata content appearing on English Wikipedia?

I will describe our understanding based on the above and other conversations and we are interested to learn and understand more.

We understand there are concerns over control, vandalism and the quality of content. We know that control by way of visibility and easier means to moderate, in particular, are legitimate issues that need to be addressed.

Significant additional work has already happened to address issues of quality. If you are interested in learning about the work that has happened and is taking place, please let us know and we can take the time to accurately document some of the most important items

Additionally, there are some relevant projects underway to improve the control and quality of Wikidata-related contributions. Some of these are organized by Wikimedia Deutschland, some assisted by Wikimedia Foundation efforts, and all developed via some form of community involvement.

We want to talk with you about which of these would help to address the concerns that you have. These projects are in varying stages of progress and represent varying degrees of effort.

Here are some steps particularly around the area of vandalism:

  • ORES (which was first made available on Wikidata in 2015, and since saw several Wikidata-specific improvements to its vandalism detection algorithm and better integration into the RC interface) helps patrollers to do their work much more efficiently.
  • Introduction of anti-vandalism bots making automated reverts of edits that are very likely to be vandalism
  • The Wikidata team recently supported a scientific competition to improve for anti-vandalism detection (similar to those that have been held for Wikipedia before)
  • Allowing institutions to cryptographically sign data that comes from them so it is easy to verify and spot when that data has been changed
  • Checks of Wikidata’s data against other databases to highlight differing data and spot mistakes

Are these the right sort of things for us to be working on? Admittedly progress has been slow, but knowing specifically what is most important will help us prioritize it.

Regardless of our progress, independent research is also beginning to provide us with a more thorough understanding of vandalism on Wikidata. A new study found that vandalism rates on Wikidata have decreased significantly in recent years (from 2013 to 2016) and are quite low: ”... -lately- only between 0.1% and 0.2% of [human edits on Wikidata] are malicious”.

There are also concerns around understanding the policies on Wikidata and English Wikipedia and how they work together. Policies for important topics like notability and verifiability exist on both projects. Regarding BLP policy, Wikidata currently operates under the same resolution from the Wikimedia board as all Wikimedia projects. Wikidata also has a draft policy on living people. Respecting the governance of the projects, we’d like to ask the projects to continue working together to further clarify how these policies interact.

Lastly, I heard the concern that by showing Wikidata content on Wikipedia and calling the site Wikipedia, we were misrepresenting the content. Presumably the same concern applies to Commons images and I don't have any constructive ideas for how to address this concern, if others share it. Open to suggestions.

4. Now that we know community members did not want Wikidata descriptions to appear in any reader interface for English Wikipedia articles, what should we do about it?

As you may know Wikidata appears in other places on Wikipedia. Wikidata descriptions appear not only below the title on mobile web in all languages except English and Russian, but in the following places:

  • Apps under the title
  • Visual editor (desktop and mobile)
  • Portal search (www.wikipedia.org)
  • Mobile web and mobile app search
  • Related pages (bottom of mobile web and app pages)
  • iOS widgets
  • App maps feature
  • Apps feed feature

In almost all cases, these have been integrated for more than a year, sometimes two years. I personally think that it would be very hard to justify removing all elements of a reader experience that have been in place for this long on the basis of the concerns raised. Pulling the descriptions would be difficult for a number of reasons, including user interfaces that rely on Wikidata descriptions (redesign needed), removing expected and relevant information for our hundreds of millions of readers, and the technical cost of removal. In the apps in particular, it is unfeasible to consider the impact of pulling Wikidata descriptions from so many places. It would force us to reconsider the information architecture of the applications from the ground up. I think it would be much more productive to figure out how to give Wikipedia editors more control and otherwise take steps to reduce vandalism than to remove pieces of our readers' experience that are so deeply integrated.

Conclusion and Next steps

Hopefully, the above gives some insight into how we at the Wikimedia Foundation are thinking about this (at least, as relayed via a single employee). We want to figure out next steps together.

I feel that an RfC is not the best way forward in this discussion, as it creates a black/white dichotomy and decreases our ability to have an in-depth, nuanced conversation about a complex matter. The foundation has a responsibility to not only the folks who are able to be present here, but the contributors across projects currently uninvolved and those that use our knowledge across projects and platforms.

The Readers teams at the Wikimedia Foundation are working under the premise that if Wikidata integration is to be successful, the following are of highest importance to the English Wikipedia community members interested in this aspect of the project.

  • Be able to patrol/review Wikidata changes that appear in English Wikipedia articles.
  • Be able to edit these changes on-Wikipedia, without having to go to Wikidata.
  • Evidence the Wikidata community takes quality as seriously as English Wikipedia

These same items are listed above with links to the relevant work we have planned for them. I hope the explanation provided shows that we are working toward addressing these concerns. But are those, in fact, the primary needs?

What do you think? Jkatz (WMF) (talk) 02:48, 14 September 2017 (UTC)

responses

Thanks for the detailed reply. I think a key point in your comments is the integration of thr wikidata description into multiple locations, from which I think it is clear that those features have been designed with the assumption that Wikidata should be treated like Commons. That might have been a reasonable assumption to start with but as you can see from the conversations above can't be assumed now. En-wiki editors treat Commons as a reliable partner. I think ongoing design work should incorporate an understanding that this is not yet the case for Wikidata. Are there any Wikidata elements other than the short description that show up in any view of WP content? To your point about the difficulty of removing existing Wikidata description integration, I think it's going to depend on whether there's a better answer - if editors here can keep that data reliable, via watchlist integration, or better participation at Wikidata, that's fine, but we'd need consensus that that was the case. Mike Christie (talk - contribs - library) 04:24, 14 September 2017 (UTC)


I read what you wrote twice, and I find myself asking. Why in the world are you, speaking as the WMF person in charge of the reading team, here advocating for Wikidata?
There is more I would have to say, but i am stuck right there. Baffling. Jytdog (talk) 04:51, 14 September 2017 (UTC)
Hello,@Jytdog:. So I'm not Jon, but here's a thought I had reading your confusion. It's not that baffling that someone from the WMF would be advocating for Wikimedia projects. It's kind of a "see the forest for the trees" thing. Folks at the foundation, including myself, see our work through the lens of not a single project, but of many diverse projects with numerous interdependencies. From MediaWiki (the wiki software itself!), to Commons, to server infrastructure, to Wikidata, to all the non-Wikipedia projects - all these things are considered when working for the movement. So yeah, I'd advocate Wikidata, and Wikivoyage, and new projects in the incubator, and more. English Wikipedia is large, it's raucous :), but it's not alone in our thoughts. CKoerner (WMF) (talk) 12:24, 14 September 2017 (UTC)
When developing Mediawiki, you need to take the needs of all projects that use the software into account, obviously. That's why we asked e.g. to disable Flow here, but don't push for you to disable it everywhere. But what you are doing here is different, here you are pushing for the use of one project on another, whether they want it or not. Just like we saw with e.g. the search results from "sister" projects. Wikivoyage is not a sister project I want to see pushed here in any way, it is fundamentally different to our principles (while e.g. Wiktionary is a logical sister project). Wikidata is something that on the one hand may use our data as much as they like, and on the other hand could be a "here we are, how can we help you" thing which we may use (for interwikilinks) or not use (for most other things), not something that the WMF should insert into our content without some serious communication and agreement upfront (just like the idea to generate articles directly from Wikidata would need a very good consensus on enwiki before it should be attempted here). And I guess that Jytdog is also baffled that someone from the Reading team would promote something that is so utterly ill-suited to reading as Wikidata. Fram (talk) 12:49, 14 September 2017 (UTC)
Thanks for the thoughts Fram. It seems to me there's some confusion over where navigation ends and where content begins. CKoerner (WMF) (talk) 13:18, 14 September 2017 (UTC)
Chris, ARGH. I was too terse and you definitely took advantage of that. Of course it is fine to advocate for projects. Fatuously obvious. (You have responded glibly to me like this before, and i have pushed back on that before)
What is baffling is that somebody from the WMF is here in Wikipedia with a very strong stance on a matter that involves en-WP content and governance, advocating that Wikipedia should allow Wikidata integration.. With no self-awareness of how overstepping that is. This is how far gone things are. It is much worse than I thought.
I just quoted this at Jimbo's talk page yesterday - I will do it again here. Terms of Use:
These Terms of Use tell you about our public services at the Wikimedia Foundation, our relationship to you as a user, and the rights and responsibilities that guide us both. We want you to know that we host an incredible quantity of educational and informational content, all of which is contributed and made possible by users like yourself. Generally we do not contribute, monitor, or delete content (with the rare exception of policies like these Terms of Use or legal compliance for DMCA notices). This means that editorial control is in the hands of you and your fellow users who create and manage the content. We merely host this content.
And later: The Wikimedia community and its members may also take action when so allowed by the community or Foundation policies applicable to the specific Project edition, including but not limited to warning, investigating, blocking, or banning users who violate those policies. You agree to comply with the final decisions of dispute resolution bodies that are established by the community for the specific Project editions (such as arbitration committees); these decisions may include sanctions as set out by the policy of the specific Project edition.
That is the fundamental deal here. The various projects each generate content, have policies that govern content and behavior, and have governance that enforces those policies. Within each project. Wikidata is different from en-WP. And it is not the WMF's role to get involved in intra=project content and governance. It is "We merely host this content". It is not - "we make editorial decisions about content"
There are differences in policy and governance between Wikidata and en-WP. These are serious when you get down to actually working.
It is really easy to see. Imagine the world where there is integration of live Wikidata in Wikipedia.
a) Somebody adds content to Wikidata that violates en-WP BLP's policy.
b) That change appears in en-WP
c) I change that here, as soon as I see it
d) that change appears in Wikidata
e) person who added that in Wikidata, changes it right back.
f) the BLP violation re-appears in en-WP.
g) Welcome to hell. Separate projects, with separate policies and governance. Not resolveable. There are very practical content and governance issues, of which you and Jon appear oblivious.
h) On top of that:
1) I don't care about Wikidata content. I don't want it to die, and I wish that project well, but I don't care, and I don't want to volunteer there.
2) I have no desire to try to change Wikidata policies and governance
3) I have no desire to try to enforce en-WP policy inside Wikidata. To me that would be as stupid as me buying a house in the UK, and expecting US law to govern there. I try not to want stupid things. Seeing them as stupid helps me not want them. (And I can only imagine that for somebody who edits mostly Wikidata, having some fuckwit from Wikipedia showing up and demanding that Wikipedia policy should govern there, would be insulting and a huge waste of their time.)
4) And if there is somebody who persistently adds content to Wikidata that consistently violates en-WP policy, what I am supposed to do about that? really - it takes an understanding of policy and "what works" to have a problematic editor's behavior effectively addressed by the community. And it takes time and effort to present a "case" of a problematic editor to the community, so that others can see it, and decide what to do about it. I know how to do that here, to help with en-WP governance. I have no idea "what works" in the Wikidata community or even what kind of behavior would be seen as worthy of that community's time. And I don't really want to take the time to learn. I don't believe that either you or Jon even understand what I am talking about in this point #4, but it is how community governance actually works.
5) to underline this, I don't volunteer my time for Wikidata. I volunteer my time at Wikipedia. Because this is what I want. There is nothing stupid about that. This integration kind of forces me to edit Wikidata - if I see a BLP violation, I really should fix that, right? And if some Wikidata editor is persistently adding content that violates en-WP BLP policy, I really should do something about that, that works in the Wikidata community, right? So this integration is really exploitation. (that is sometimes explicitly said but in a way that is fake and nice ("Oh, it would be great to have the more active and bigger Wikipedia community work on Wikidata". It is actually distraction, exploitation, and setting up arguments that cannot be resolved or can be only be resolved with a serious time commitment)
Summarizing, there are issues of the WMF
A) overstepping into community content and governance (the whole advocating "response" post, is overstepping)
B) ignorance of the differences in each project's governance that have very practical consequences
C) trying to create a situation that is harmful to both communities of editors (out of ignorance of B, surely. but "the road to hell is paved with good intentions")
Again, the whole post is baffling to me. Things are very far gone, in a bad direction.
And I ask myself, why is this even happening to a level where I have spent all this time (my volunteer time) addressing this? How did things get so fucked up at the WMF?
Bottom line - The WMF has power as the publisher. It is arrogantly overstepping its role and this is seen as entirely normal and OK within the WMF to the point where a product manager comes here and writes a long post from that stance. The WMF doesn't understand the problems it would create. Jytdog (talk) 14:29, 14 September 2017 (UTC)
Sorry about the confusion there. As you said it seemed rather obvious to me as well, so I was baffled as to why you'd be making that argument. I misunderstood. CKoerner (WMF) (talk) 14:47, 14 September 2017 (UTC)
I will also say "hell no" to what you say about Wikidata in Wikpedia. It is deeply wrong headed. But I will talk more about that later. I am just too baffled on the fundamental stance you have taken to write more. Jytdog (talk) 04:54, 14 September 2017 (UTC)


I hesitate to say what the community wants without a broader community discussion, but my impression is that experience with the wikidata experiment has been shifting consensus against it. Wikidata isn't like Commons. I think most people who are unhappy with wikidata won't much care about the improvements you've suggested. Wikidata worked fine for its original purpose, interlanguage links. Having articles pull content from wikidata has been controversial, and we've got consensus against the wikidata-descriptions push content. Alsee (talk) 06:07, 14 September 2017 (UTC)
Hey @Alsee:, what do you mean by "original purpose"? I'm under the impression that Wikidata was presented back in 2012 as being exactly like Commons. "It will provide data in all the languages of the Wikimedia projects, and allow for the central access to data in a similar vein as Wikimedia Commons does for multimedia files." I don't think I'm the only one operating under this assumption and would appreciate any clarifying sources if you have them. CKoerner (WMF) (talk) 12:07, 14 September 2017 (UTC)
Yes and no. That was the intent, however the differences have cropped up subsequently to that. Commons has fairly restrictive policies regarding its content which are compatible with ENWP/Wikipedia. ENWP chooses what we want from Commons to display on a relevant Wikipedia article. Because of Commons policies we can trust the content we *choose* to use is valid and not problematic. Wikidata has no compatible content policies and is effectively (in the case of descriptions) forcing its unverified and potentially legally problematic content onto Wikipedia articles. If Wikidata had compatible policies (as commons does) this would not be a problem. However it currently does not, and is unlikely to do so anytime soon. Which is why ENWP *requires* no wikidata pushed-content (descriptions, infoboxs are a different issue) is visible when people visit a Wikipedia article on *any* platform - or at least the option to manually turn it off from the ENWP end. Only in death does duty end (talk) 12:16, 14 September 2017 (UTC)
I didn't pay a lot of attention at the time Wikidata was launched but interwiki links were widely discussed as an obvious use case, so that might be what Alsee is referring to. Yes, I think the original idea was that it would be treated just like Commons. Fram's point below is the key issue in the current discussion -- the description is not visible to or controllable by a Wikipedia editor. That is quite unlike Commons. Other issues that have been raised with Wikidata here relate to whether we should consciously choose to use it for citations, or filling infoboxes; generally those effects result from a choice by an editor here, and we still haven't come to consensus on which of those uses should be allowed. For the descriptions, it's quite different: it's invisible and uncontrollable from en-wp. That's what needs to be fixed. Mike Christie (talk - contribs - library) 12:15, 14 September 2017 (UTC)
CKoerner (WMF), I am not deeply familiar with the founding wikidata so I apologize for any imprecision in my comment. As I understand it, linking equivilent articles across languages was the initial use of wikidata on wikipedia. As I understand it, the interinterlanguage link mess alone was enough to justify that a solution be built. Whatever concepts were pulled together at wikidata's founding, I have the impression that interlanguage links could reasonably be credited as a primary factor moving the project from concept to reality. Alsee (talk) 17:30, 14 September 2017 (UTC)
My understanding of what the original intentions for Wikidata were, matches the text at Wikidata:Notability: "Wikidata in its first phases has two main goals: to centralize interlanguage links across Wikimedia projects and to serve as a general knowledge base for the world at large." What is not said there is "to serve as a data source for Wikipedia" or something similar. Fram (talk) 09:51, 15 September 2017 (UTC)
Then that's an incomplete understanding. See e.g. [1]: "Wikidata will support the more than 280 language editions of Wikipedia with one common source of structured data that can be used in all articles ". Regards, Tbayer (WMF) (talk) 13:42, 15 September 2017 (UTC)
Mine is from a current policy page, yours comes from a blog. I never read that blog, I am not a fan of the WMF good news show. Posts from the same year include things like this on AFT and all kind of glowing articles on VisualEditor (the default editor in 2013, as it was so good by then of course). I've made a comment about a typical WMF announcement earlier today at WP:VPT[2]. It's impossible to get a complete and accurate understanding about most WMF projects from WMF communications, and reading that blog will hardly be a help in that regard. Fram (talk) 14:27, 15 September 2017 (UTC)
Can be used in all articles is not the same as will be inserted into articles in other projects by outsiders without consultation and without editorial control from the afflicted projects. Can be used in all articles is fine, as it implies the option not to do so if the editors of any particular article choose not to do so. Wikidata is welcome to use material from Wikipedia, provided they comply with the conditions of use. Wikipedia is welcome to use material from Wikidata, under similar conditions. Neither project is welcome to insert their content into the other in a manner which is not editable from within the recipient project. If WMF as the host find it necessary to arbitrarily insert content from one project into the displayed page for another article they should clearly label it as content from another project, and be aware that some of us will consider it to be spamming. · · · Peter (Southwood) (talk): 21:35, 17 September 2017 (UTC)
Pbsouthwood, I was making a factual correction to mistaken claims above about what the original intentions for Wikidata were, not describing an interpretation of it by WMF or others.
Regarding your remarks on labeling, FWIW: The app's new description editing function (which, as mentioned in the recent announcement that triggered the current debates, has been live on all Wikipedias except English already, but can be tested for English in the not yet released alpha version of the app) also adds the following explainer text, reachable by tapping on an info link: "Descriptions are stored and maintained on Wikidata, a project of the Wikimedia Foundation that provides a free, collaborative, multilingual, secondary database supporting Wikipedia and other projects".
Regards, Tbayer (WMF) (talk) 08:07, 18 September 2017 (UTC)
I did not presume to judge your intention by quoting the blog. I pointed out that the purpose stated in the quote differs substantially from what appears to be the current reality, and not in a good way. It goes further than loose interpretation. I will not presume to judge whether it was intentional misinterpretation, or accidental, or even some other possibility that does not immediately come to mind. The resulting implementation puts the cart before the horse. Instead of supporting the Wikipedias with a source of data that all can use, it is becoming a source of data of dubious provenence that is being imposed upon the Wikipedias in a currently unacceptable way. It will be interesting to see who gets the blame if/when WMF is sued for defamation or libel or whatever when some litigious person or organisation sees this as an opportunity to curtail a competitor or even just sues out of spite against the principle of free information. I would like to see an unambiguous statement from legal that there is no risk to the Wikipedias from this. Can you assure us that this has been fully cleared by legal? · · · Peter (Southwood) (talk): 12:17, 18 September 2017 (UTC)
@Pbsouthwood: So long as the site is simply relaying user-provided content (whether provided here or at Wikidata), WMF is covered by Section 230. That's not an argument not to try our very hardest to get things right and to eradicate error -- above all, there's our reputation to protect. But it does mean that overwhelming legal peril for WMF is not such a concern, so long as they act quickly to fix any individual cases they are made aware of. Jheald (talk) 12:40, 18 September 2017 (UTC)
I will take your word about the app, as I do not use it, and it is likely that someone else will do the checking anyway, However I will ask how obvious that link is to the casual user, and how obvious it is that the link will explain where the information is sourced, and that it does not come from Wikipedia, which may be the obvious expectation. · · · Peter (Southwood) (talk): 12:17, 18 September 2017 (UTC)
@Jkatz (WMF):, theer are many things I could say, but I'll stick for now with one major thing which you seem to have overlooked in your response. When you are comparing the descriptions with things like Commons or the use of Wikidata elsewhere on enwiki, there is one truly fundamental difference: when we display e.g. a Commons image, we explicitly link to it from the article. When we display other Wikidata content, we explicitly link to it from the article or from an enwiki template. With the descriptions, in every example you mention, this is pushed by the WMF, not requested by enwiki, and not something we can disable even on an article-by-article basis.
For that reason (plus some other reasons which also apply), please disable Wikidata descriptions everywhere on enwiki, except where it is explicitly asked by an article or enwiki template in some way. Fram (talk) 11:49, 14 September 2017 (UTC)
@Fram:, So if Wikidata descriptions could be 1) disabled on an article-by-article basis and 2) were hyperlinks to the Wikidata entry (like Commons images), then things would be better. Is that what I understand? That's a new twist on the subject. One I have not heard proposed thus far. I also understand your desire to remove completely, I'm not trying to be obtuse. I guess what I'm trying to say (haven't had coffee yet this morning) is that if we could snap our fingers and have these things you'd be a smidgen happier? CKoerner (WMF) (talk) 12:14, 14 September 2017 (UTC)
Those things would help; I think if we could also see the changes in watchlists in a comprehensible way, and if it were easy to fix problems or vandalism spotted via the watchlist, that would go a long way to resolving the issue. Mike Christie (talk - contribs - library) 12:17, 14 September 2017 (UTC)
@CKoerner (WMF):. No, if Wikidata descriptions were something we (enwiki) requested, in an article or by using a template, then things would be better. Changing the text (description) to a hyperlink to Wikidata wouldn't be better, it would be even worse. "Things that get disabled on an article by article basis" are things which get, by us, pulled from Wikidata through a template, but which we, at enwiki, can even then decide not to use in one article or another. The main concern is that we, the enwiki community, can implement where we want to display what Wikidata info, and can again disable it if we don't want it, on an article or in general. Fram (talk) 12:42, 14 September 2017 (UTC)
@Fram: I hate to ignore a direct ping, but the cold I had the last couple days took over last night, my brain is mush, and I realize I need to take a break. I'm taking a day or two to recuperate and then will get back on here. Jkatz (WMF) (talk) 15:37, 14 September 2017 (UTC)
I agree with Fram here - there seems to be a fundamental error in the idea that Wikidata is like Commons. No, its not when there is an effort to push content onto en-Wikipedia from Wikidata without en-Wikipedia's control. Commons doesn't push information/images onto en-Wikipedia, so en-Wikipedia has control over what information/images is used and what isn't. Currently, Wikidata isn't in that situation and given our sourcing and BLP policies here, its a huge concern that Wikidata information is being pushed onto en-Wikipedia without the ability of en-Wikipedia to control that. Until that fundamental misunderstanding is cleared up, much of what was written above is really just hot air, to be blunt. Ealdgyth - Talk 11:58, 14 September 2017 (UTC)
Mike says it would help and you say No. What am I suppose to do with that!? :p Seriously though, and I'm just talking shop here, this is one of the difficulties I have in understanding the needs of our communities, but I appreciate the considerations. CKoerner (WMF) (talk) 13:21, 14 September 2017 (UTC)
There's certainly some disagreement here about the best answer, but I'd add that I didn't say it would resolve it. My answer was in the spirit of "more tools would help". I don't think it would be enough. Even with the extra tools I suggested, it's debatable if it would resolve the problems we're discussing. Mike Christie (talk - contribs - library) 13:35, 14 September 2017 (UTC)
User:CKoerner (WMF), what you and the WMF should do, is stay the hell out of it. It is not the WMF's role to make these decisions. Unilaterally changing Wikipedia content is completely outside of the WMF's role. Creating tools that allow integration of Wikidata content in WP is one thing, but whether the en-WP community decides to use them is entirely different. That is an internal en-WP editing community decision. Baking integration tools into the Wikimedia software and driving Wikidata content into en-WP would be... extremely controversial. We are talking Flow-level controversial, if not worse, because this would be a content thing. More than anything, WMF needs to be neutral player, and not advocating for X or Y when it comes to content and governance. Jytdog (talk) 14:52, 14 September 2017 (UTC)
Sure Commons pushes information in. Imagine I add a portrait image of a person to an article and then someone on Commons replaces it with an image of a penis, and suddenty we have an image of a penis like it is a portait and nothing shows up in the watchlist. It just fortunately does not happen very often.--Ymblanter (talk) 12:17, 14 September 2017 (UTC)
The key difference as has been pointed out multiple times is that a)commons has policies against this, b)commons has enough editors that people will SEE this, c)Wikipedia editors will see this looking at the article and report it promptly. Wikidata has no equivalent to WP:V or WP:BLP to restrict what content is hosted on wikidata. Wikidata does not have enough editors to spot and handle deliberate vandalism. Wikidata changes do not necessarily show up when even looking at the article on Wikipedia as in the case of descriptions, its being pushed to mobile/app devices where users are unlikely to report the issue, or even attempt to fix it. The comparison's to commons need to stop because Wikidata is currently nothing like commons, either in its function, its management, or at a basic policy level. Only in death does duty end (talk) 12:24, 14 September 2017 (UTC)
What's more, if Commons ceased to reliably patrol for this sort of vandalism, the result would almost certainly be that en-wiki would be less and less willing to use it. En-wiki editors don't get exhorted to go and edit on Commons and help keep things clean there; they have an active editing community that we rely on; if they didn't we wouldn't rely on them. Mike Christie (talk - contribs - library) 13:38, 14 September 2017 (UTC)
Of course Wikidata has content policies and guidelines that disallow the equivalent of penis picture vandalism; it's quite absurd to imply otherwise. At least it seems you did not look at the one concerning descriptions before posting these far-reaching claims. Regards, Tbayer (WMF) (talk) 14:16, 14 September 2017 (UTC)
Please provide a link to the relevant policy page on Wikidata that complies with ENWP's WP:V and WP:BLP policies. Only in death does duty end (talk) 15:27, 14 September 2017 (UTC)
Note btw that whereas reupload vandalism is relatively uncommon and typically get detected and reverted on Commons quickly, changing descriptions which goes contrary to the English Wikipedia NPOV policies is relatively common, and the descriptions stay on Commons sometimes for years - and are actually shown on the English Wikipedia - this is what one gets if one clicks here on a file.--Ymblanter (talk) 07:42, 22 September 2017 (UTC)
I have always thought of Wikidata being exactly like Commons -- a common storage for data that does not have a natural home in a single Wikimedia project. I don't understand where the idea that Wikidata is something alien really comes from. It might require improvement, but avoiding its use is making sure it will never be improved (remember Wikinews and other projects that died?) —Kusma (t·c) 12:25, 14 September 2017 (UTC)
Can we have an option to show the Wikidata "short description" on every article page, so we can easily check whether it is ok or not? —Kusma (t·c) 12:25, 14 September 2017 (UTC)
Imagine that the WMF would start every article with an image, taken from Commons, which they think is best suited for the article. This is basically what they do now with the descriptions. With commons, we can remove all images from an article, we can host the image locally, ... E.g. for our featured article on the main page, all images get fully protected, just like all templates and so on. However, the short descriptions, shown to many people, is not fully protected. This is, as far as I am aware, the only part of the TFA that doesn't get protected (and probably doesn't get checked by TFA promotors either). It's just an example, but it's an indication of how this use of Wikidata is not like Commons. Fram (talk) 12:42, 14 September 2017 (UTC)
The FA itself usually isn't protected, only the bits that go on the Main Page are. Any Wikidata items used on the Main Page need to be checked for their vandalism potential and technical solutions for protection should be found. Why do you make this fundamentally different than the Commons? —Kusma (t·c) 12:55, 14 September 2017 (UTC)
Because the short descriptions are not hosted on ENWP, nor visible when checking through a standard browser. They are not part of the article in any way, nor part of the review. Any wikidata items transcluded through a template would be checked, anything inserted after the fact by an outside source would not be. Only in death does duty end (talk) 12:59, 14 September 2017 (UTC)
Thanks, I sould have checked that before posting my reply. My point remains (pushing unrequested data which is not visible to most editors and which we can't avoid by removing the link or (as with commons) adding a local version instead, is totally different to how we use Commons: the comparison with Commons is fair in that ragard for many other uses of Wikidata on enwiki, but not for this one), but my example was wrong. Fram (talk) 13:11, 14 September 2017 (UTC)
No, your analogy is deeply flawed in other ways too: The description is not selected by WMF, as you imply ("WMF would start every article with an image, taken from Commons, which they think is best suited for the article"), but by editors - there is only one for that article, selected by editors, and if an editor thinks a different wording is better suited, they can edit it. And yes, editors will see a different Wikimedia domain appearing in their browser's address bar when they do that, instead of en.wikipedia.org, but that's also been the case with Commons since 2004. Regards, Tbayer (WMF) (talk) 14:16, 14 September 2017 (UTC)
The WMF has decided that the Wikidata description is the right thing to show on top of enwiki content. The WMF thinks this info, inserted outside enwiki, is best suited for enwiki articles and should be the first thing mobile readers (and so on) see. The description is written by editors, but is selected by the WMF as the thing to display here. Nothing "deeply flawed" there. "[...] if an editor thinks a different wording is better suited, they can edit it. And yes, editors will see a different Wikimedia domain appearing in their browser's address bar when they do that, instead of en.wikipedia.org, but that's also been the case with Commons since 2004". First of all, until very recently (and still for most cases), a reader wouldn't know where the info came from, and choosing "edit" wouldn't bring him to the right place. So no, in reality, a normal reader or editor can not edit it. Furthermore, don't like an image? You can edit it in the article, there is no need at all to go to Commons and you will not, contrary to what you claim, "see a different Wikimedia domain appearing in their browser's address bar". Fram (talk) 14:52, 14 September 2017 (UTC)
The WMF team and only the WMF, made the editorial decision to take the description field from Wikidata and insert it into en-WP articles on mobile and the apps. That is a sole WMF decision. And it is about content. Do not misrepresent what was done. You are making things worse, not better, by obfuscating what was done and who did it. Jytdog (talk) 14:55, 14 September 2017 (UTC)
(I'm probably not helping here) There's a bit of code you can add to your common.js to do just this. CKoerner (WMF) (talk) 13:04, 14 September 2017 (UTC)
Thank you. I guess you mean Special:Mypage/common.js ? —Kusma (t·c) 13:06, 14 September 2017 (UTC)
I have installed it, it seems to do exactly what I was looking for. —Kusma (t·c) 13:08, 14 September 2017 (UTC)
Also, there is always the "Wikidata item" link in the left sidebar, which leads one to the description (if it exists) with one click. Regards, Tbayer (WMF) (talk) 14:16, 14 September 2017 (UTC)
I tried this to see what it does; for others reading this, if you install the script, the Wikidata description will appear at the top of the page, which at least makes vandalism easy to spot on the page you're reading. A good next step (assuming these descriptions don't disappear as a result of this or other conversations) would be to make it equally visible in the watchlist. Mike Christie (talk - contribs - library) 02:05, 15 September 2017 (UTC)
Wikidata description: "8th President of the Republic of Singapore"
  • No big deal It's interesting that Mr Katz's constituency is the people who read Wikipedia. I'm not sure that he's come to the right place to get the views of the general readership but, as it happens, I dip into the iOS app on most days. The one line summaries in question are quite inconspicuous in my experience and so there isn't a significant issue or problem. I reckon that the images we take from Commons have much more potential to cause trouble. For example, consider an image which is currently on the main page (right). This purports to be Halimah Yacob and I have no reason to doubt this. But the provenance of this image seems weak as it just seems to have been scraped and cropped from Flickr. In Wikipedia terms, this image fails both WP:OR and WP:RS because it's original work and self-published. And it's much harder for people to challenge and change such an image because most readers won't have an alternative readily available. But we're used to such images and are comfortable with the risks as, in practice, they seem to be low. The risks of using WikiData descriptions seem quite small too and so we shouldn't worry too much about them. Andrew D. (talk) 21:45, 14 September 2017 (UTC)
    Absolutely. And of course Commons does not have anything remotely close to WP:V. Most of the Wikidata info cones from databases, which we call reliable; most of the Commons info has doubtful provenance.--Ymblanter (talk) 06:48, 15 September 2017 (UTC)
    This shows the breakdown of how many statements on Wikidata are unsourced, sourced to Wikipedias, or sourced to other references. Note also that not all "other" references are reliable. Nikkimaria (talk) 13:17, 15 September 2017 (UTC)
Sure. None of the files on Commons are sources, most are junk, but the English Wikipedia continues to accept them no problem.--Ymblanter (talk) 13:53, 15 September 2017 (UTC)
"None of the files on Commons are sources": yes, files are not allowed to be used as sources. "Most are junk"? And we can remove any from enwiki we consider to be junk. We accept the images we want to, and reject the ones we want to. How is any of this relevant to the use of Wikidata as a source of information we can't refuse and which isn't sourced. "Most of the Wikidata info cones from databases": the identifiers come from databases and vastly distort the percentage of "sourced information". The actual information on Wikidata is usually either unsourced or Wikipedia-sourced. Only for a few select types of items has there been an effort to actually source it. Fram (talk) 09:12, 18 September 2017 (UTC)
Unfortunately I have to repeat my point for the third time bacause somehow people do not want to listen. If the file from Commons is chosen to illustrate a statement X, there typically is no source in terms of WP:RS which states that the file illustrates X. As it was noted in this discussion, many Commons files have unclear provenance (e.g. taken over from Flickr) so that we have to rely on unsourced statements of people who are not even Wikipedia users and are not notable in any way. This is potentially problematic and can lead to libel court cases. However, it is not perceived by the local community as problematic. In contrast, Wikidata has sources to most of the statements, these sources are reliable by Wikipedia standards, and the share of these sourced statements increases.--Ymblanter (talk) 13:04, 18 September 2017 (UTC)
Well, you said "none of the files on Commons are sources" which is true but irrelevant, as images are not allowed to be used as sources on enwiki. Now you go back to another issue, that many files on Commons are poorly sourced, which is a completely different and unrelated point. "In contrast, Wikidata has sources to most of the statements, these sources are reliable by Wikipedia standards, and the share of these sourced statements increases." That's what they want us to believe at least. Have you used the "random item" on Wikidata and actually checked the sources? I'll start a new section below about this. Fram (talk) 13:12, 18 September 2017 (UTC)
If you read the whole thread starting from the message of Andrew Davidson this is exactly what he was stating, and I agreed with him. I am not sure why other people continued throwing around irrelevant arguments, but when I see a wrong or irrelevant statement formulated as an attempt to refute what I am saying I usually try to respond, because people nowadays usually do not read the whole argument and only stick to the last statement.--Ymblanter (talk) 14:20, 18 September 2017 (UTC)
[ec}I am surprised by your claim that most Wikidata statements are reliably sourced by Wikipedia standards. This is not the impression I get from most of the discussion here. Can you direct me to the statistics on what number/percentage of Wikidata statements are sourced, and the analysis of the reliability of the sources? It would also be useful to know how one can tell the difference between a reliably sourced item and one without a reliable source.
I agree that the provenance of some images at Commons is dubious, but at least if we don't like them they are really easy for us to to delete from an article. Some editorial judgement is required, and can be applied. Cheers, · · · Peter (Southwood) (talk): 13:37, 18 September 2017 (UTC)
Answered in the topic below.--Ymblanter (talk) 14:20, 18 September 2017 (UTC)
  • User:Jkatz (WMF) how far out is the "only show relevant item" ability? I see that as the biggest blocker to using WD more on EN WP. We have had significant vandalism to a WD description appear at the top of one of our most read medical articles for nearly 6 months / one million page views nearly two month and half a million page veiws. This is a serious problem. Doc James (talk · contribs · email) 01:38, 15 September 2017 (UTC)
The undisputed usefulness of this proposed feature notwithstanding, Doc James has since clarified on Facebook that this example (pneumonia, Q12192) did actually not involve a vandalized Wikidata description. Instead, I understand these nearly 6 months / one million page views concerned the use of Wikidata information in the article's infobox (which was an editor decision and entirely separate from the app use of descriptions) and a dispute about data added by ProteinBoxBot. Of course there are incidents of vandalism too, as discussed below, but the open question is how likely they are to affect the reader overall (the quoted result from the paper by Crescenzi et al. indicates that the likelihood may not be very high). Regards, Tbayer (WMF) (talk) 10:51, 15 September 2017 (UTC)
Thks User:Tbayer (WMF) yes am mistaken on the specifics of that one which was a different WD issue. But between Sept 24th 2016 until Nov 14th 2016 this article had vandalism in the description.[3] during which time it received about 400,000 views.[4]. This is one of our most viewed medical articles. Doc James (talk · contribs · email) 12:15, 15 September 2017 (UTC)
But the vandalized description did not affect the vast majority of these 400,000 views (the descriptions were not being shown on either desktop or mobile web back then, and the app share is a rather tiny portion). And regarding app users, I would think it may have gotten corrected much sooner if the description had been editable for them - that's exactly the shortcoming which is now being fixed, per the recent announcement that triggered all the current debates. Regards, Tbayer (WMF) (talk) 13:42, 15 September 2017 (UTC)
Thanks User:Tbayer (WMF). I was under the impression that this was used on all mobile. Eran tells me that the improved watchlists for WD within Wikipedia is almost done. So hopefully all will be better soon. Doc James (talk · contribs · email) 01:19, 16 September 2017 (UTC)
The article "Q5506618"? Yep, undisputed usefulness
The article "Q5506618"? Yep, undisputed usefulness
"::The undisputed usefulness of this proposed feature notwithstanding," I certainly dispute its usefulness, as it gives a false sense of security and doesn't tackle the actual problem. I also hope it is in reality a lot better than what the WMF shows in their help pages. "And regarding app users, I would think it may have gotten corrected much sooner if the description had been editable for them - that's exactly the shortcoming which is now being fixed, per the recent announcement that triggered all the current debates." It wouldn't need to be corrected if the WMF did what was asked in the first place, and did what was normal (i.e. not pushing outside content to the top of enwiki pages) in the first place. You are defending a poor band-aid solution instead of implementing the real solution, and act hurt when people don't fall over backwards in gratitude. Fram (talk) 09:12, 18 September 2017 (UTC)

Arbitrary break

I'm a long-time editor who has spent some time introducing students to editing wikipedia, before and after the advent of the VisualEditor. They and I now regularly use the VE for most of our editing. This one small application of Wikidata material (showing a one-line description below the article title) is incredibly useful, and substitutes for clicking through to a disambiguation page or link for every single wikilink we enter using the VE. So, the functionality is useful.

But it's impossible to see or edit from enwiki. (Maybe the former is desired, but the latter is a problem.) Moreover, the alias is literally part of the interwiki system from the Wikidata side. It should match up with the article, as understood by people on enwiki. So, I think it's crucial for it to be durably changed and maintained from within the corresponding language Wikipedia. Surely, there's also some way for it to show up as an editable object in the VisualEditor. This problem should be fixable without addressing the bigger problems at Wikidata. And the solution should involve Wikipedia-generating content overriding/correcting/appearing on Wikidata.

(Also, Wikidata has a huge lag in data verifiability and maintenance that makes errors and vandalism quite persistent. I just removed the unsourced "ealiaesean" as an unsourced alias for Björk that has persisted since 4 April 2013, but appears literally nowhere else on the searchable Internet. If Wikidata is to become our indexing engine, it has to shape up.)

Zooming out for a minute: As someone who just started seeing these debates a few months ago, I find the lengthy tirades, raised hackles, and persistent frustrations attached to enwiki/WMF arguments to be a diversion from these (overlapping) groups of people understanding one another. Those Wikipedians who have chosen to spend time on governance issues tend to share a focus on defeating vandalism, COI editors, and other on-wiki threats. Conversely, WMF seems to be kept up at night by the threat of declining relevance to mobile users (fast becoming the majority of the Internet) and of having the lifeline of visitors shut out by Google's structure AI development, which could provide (free) Wikipedia content, without sending users to Wikipedia in return. Both of these are important fights to wage: without high quality safeguards, Wikipedia will lose credibility, and without an ongoing stream of visitors, Wikipedia would lose its relevance and the banner-ad visitors who keep the lights on and the servers running. In this case, as in many, I see the concerns of the "Wikipedia community" as highly legitimate, but the rule-based solutions proposed by community members as thoughtlessly aimed and undercutting ongoing initiatives to make Wikipedia's information more structure, more searchable, more accessible by mobile users, and available in formats that are more familiar to the current generation of digital device users. But unless and until we speak from a position of interdependence, rather than boundary-setting, we won't be able to collaborate in keeping Wikipedia the kind of smart, universally accessible knowledge tool it could be.--Carwil (talk) 18:20, 14 September 2017 (UTC)

Differences between Wikidata and en-WP are not imaginary or bureaucratic; they are very real.
WMF has overstepped and made a mistake. It happens.
The solution is not to pretend it didn't happen.
The solution is to undo the mistake, and find a different solution to the perceived content problem. Jytdog (talk) 19:20, 14 September 2017 (UTC)

Jkatz (WMF), I appreciate you and the other staff members coming here to discuss this with us. You also produced a very thorough list of upgrades for wikidata descriptions on wikipedia. I believe the people asking for improvements will be very satisfied by it. I also appreciate the reasons you gave for not wanting to remove wikidata descriptions. However I would appreciate it if we could clarify the scope of discussion here.

If you are offering to upgrade wikidata descriptions as your preferred option, and we're also discussing local solutions (using lead sentence and/or {{short_description|}}), great.

If not, then it's not really an offer. That would just be a proposal to increase how much of wikidata is forced on us. People who do not want wikidata descriptions also do not want increased integration of wikidata descriptions.

I think most of the discussions on this page are going in pointless circles, debating the pros and cons of wikidata in general. I think everyone pretty much knows the various positions. I'm not seeing much real movement. The noise here needs to be refined into a consensus. If you're offering to upgrade wikidata descriptions, I'd be more than happy to collaborate on drafting a Village Pump RFC with your proposed upgrades. Although to be honest, I give "upgrade" less than 50% chance of acceptance unless something unexpectedly shifts consensus. I'm an experienced RFC closer, I carefully studied the previous RFC[5] in terms of the proposed improvements. A minority thought wikidata descriptions were fine as-is. Another minority asked for improvements. However among those who were concerned with the descriptions, the concerns were primarily for descriptions to be local. You're basically trying to build a coalition-consensus between people who want improvements, and people don't care about improvements because they think current descriptions are fine. That might get you to consensus, but I think you fall short. I think consensus will be for local descriptions. Alsee (talk) 02:11, 17 September 2017 (UTC)

No, absolutely not. What JKatz wrote are things that I and many other people already knew, and they do not speak to any of the issues. The WMF has still overstepped its role, and of course nothing that they wrote addressed the policy and governance difference between the projects. And all that the proposed tool to edit Wikidata descriptions here in en-EP would do, is set up unresolveable cross-project edit wars. Someone changes X there in Wikidata, it propagates here, I change X here, that propagates to Wikidata, and who ever did it there, reverts it there, and that propagates back here.... It is just a bad idea. Jytdog (talk) 02:35, 17 September 2017 (UTC)
Jytdog, maybe you read it too hastily? Are you saying "absolutely not" to an RFC on which way to resolve the issue? As I indicated, I think consensus will be for local descriptions. Alsee (talk) 02:59, 17 September 2017 (UTC)
yes i did. struck. What you write about the descriptions thing if a bit off though. The consensus is likely to be just be "no" to any "description" at all, not "yes" to local descriptions. You seem to be begging the question somewhat and accepting the WMF's stance that "descriptions" are needed at all in en-WP articles. Jytdog (talk) 03:06, 17 September 2017 (UTC)
Jytdog, I think duplicating the lead sentence on mobile views of the article would be redundant. I had intended to say "except where it's redundant", but it that got lost during pre-save rewrites. The description is also used in a several other places such as the link-creator button in VE and in the search results at the www.wikipedia.org search portal. It's clearly useful to have descriptions in those places. Using the lead sentence would be a definite upgrade, because many pages have no wikidata description at all. VE and the search portal currently just show a blank description box for those articles. Alsee (talk) 05:08, 17 September 2017 (UTC)

Why WMF should immediately stop pushing content from Wikidata

First, I am not a lawyer, & my knowledge of this portion of how defamation works concerning the common carrier provisions of Communications Decency Act is abysmal, but the more I think about it, the more I wonder if the WMF has really thought thru the legal ramifications when they pushing content from one of the Wikimedia projects to another.

As I understand it, if I write something defamatory about some living person on a Wikimedia website, in the end I am legally responsible for it. The WMF is held blameless because they are a common carrier. This is the legal arrangement the US phone companies have had since forever: after all, why should a telecomm be held responsible because two of their customers happen to be evil people who used their phone to plan how to blow up a school or otherwise cause massive mayhem? It's not their job to regulate what their customers discuss on the phone. In the same way, the WMF is held blameless about content on any of the Wikipedias, or Commons, or Wikisource, or Wikidata: it's not their job to regulate what is in the content.

However, once the WMF begins to edit the content of articles by taking material from one project & adding it to another, then the Foundation is making them liable for that content. At least it seems to me, & unless I'm an expert in this portion of the law -- or a judge has made a ruling on this matter -- I would advise against the WMF touching content at all. (Yes, there are exceptions & precedents about the Foundation editing articles, but these govern specific & extreme cases.) Because by inadvertently copying defamatory material from, say, Wikidata to Wikipedia, that potentially makes the Foundation legally responsible for it. And, as I just said, the only way to determine if the Foundation really is responsible is for it to be sued. Which costs lots of money I (speaking for myself) would not want to see spent.

Or maybe I'm just another ignorant editor, too deep into trying to keep facts about Romans of the 2nd century AD straight, to know that the project team already had the WMF legal department look at this & sign off on it. Meaning that the lawyers are perfectly happy with the risk of the Foundation being sued over an oddball case of potential defamation that will likely take years to sort out through endless hearings into something that can actually be taken to trial. But I've had enough brushes with situations involving potential legal responsibility to think otherwise. -- llywrch (talk) 07:41, 21 September 2017 (UTC)

@Llywrch: WMF aren't manually reviewing data taken from Wikidata. It's fairly clear that Section 230 continues to apply. Jheald (talk) 09:09, 21 September 2017 (UTC)
I agree it's a bit muddy though. Likely only possible to figure out in court, little precedence for a composition like this. —TheDJ (talkcontribs) 09:27, 21 September 2017 (UTC)
Maybe, Jheald. But what have the lawyers said about the Foundation directly modifying content on a project in this way? Would one of the WMF employees like to answer?

Please note: it's not my intent to trash Wikidata. As I've said elsewhere, it has tremendous potential for Wikipedia & other Wikimedia projects. (An obvious example is demographics: putting the raw data in one location alone is a big benefit for all of the projects; having them where they can be manipulated independently by each project -- or each page in each project -- is an even bigger benefit.) But if some vandal added something that violates WP:BLP on Wikidata, & the Foundation adds it to a content page on en.wikipedia, what would be the Foundation's exposure? Unless a lawyer can assure us any such lawsuit against the Foundation would be dismissed as frivolous, I would be concerned. (And remember, since the servers have been moved from Florida to Virginia, the laws regarding this issue have possibly changed. So there is a lot of uncertainty & potential risk here.) -- llywrch (talk) 15:01, 21 September 2017 (UTC)

The Foundation aren't "directly modifying content". They're merely showing user-supplied content sourced from two different sites adjacent to each other, without manual intervention. That is squarely covered by Section 230, which is Federal law, applicable anywhere in the United States.
If the Foundation were to take ownership of these short descriptions, maintaining them in-house as their own creation, that might be more tricky. But I don't think anyone is talking about that. Jheald (talk) 15:11, 21 September 2017 (UTC)
I don't know about the legal argument but what you are writing about the short descriptions appearing say in the android app is very weak. Passively hosting usergenerated content is very different from actively playing DJ. The systematic presence of the description anywhere other than Wikidata is 100% the product of WMF decision-making and action. Jytdog (talk)
The systematic presence of any content anywhere on any WMF site is the result of the code WMF maintains. If you think s230 doesn't apply, let's see some hard citation or argument or precedent. Otherwise it's simply how WMF chooses to present user-generated content. Jheald (talk) 18:14, 21 September 2017 (UTC)
Nope. Passively displaying vs active mixing are entirely different things, especially here where the WMF has been very clear that the content change they introduced was meant very much to be impactful. They know that putting the description as the first thing will affect what people do - they want it to. The mashup is a derivative work authored by the WMF. But as I already said I am not commenting at all on issues of legal liability - I don't play lawyer on Wikipedia - the WMF has lawyers. Jytdog (talk) 18:24, 21 September 2017 (UTC)
I'm pretty certain that this is a legal non-issue. It's user generated content, and it doesn't matter how much software it's sent through or where it appears. The issue is that the staff aren't manually reviewing and moderating the content. They aren't legally responsible for stuff that they literally haven't seen. Alsee (talk) 03:19, 22 September 2017 (UTC)
I'm not talking about liability; I am talking about authorship, and creating this kind of montage is undeniably a step removed from what the original author wrote. The WMF takes that authored content and pastes it elsewhere, very intentionally and this is a form of authorship. See for example Wikipedia:Copying within Wikipedia -- the person who does the copying is the "editor" recorded in the history, and we are supposed to attribute in our edit notes and record where we took it from. This dual nature is present as well in the WMF's decision to paste the Wikidata content where ever it - the WMF - places it. Jytdog (talk) 03:44, 22 September 2017 (UTC)
I had de-indented, intending to respond to the original liability concerns by Llywrch. Alsee (talk) 04:07, 22 September 2017 (UTC)
Llywrch, I checked with the Foundation's Legal team, and they read through this discussion. They don't have an issue with the feature. -- DannyH (WMF) (talk) 00:45, 23 September 2017 (UTC)