The Signpost

News and notes

Media Viewer—Wikimedia's emotional roller-coaster

Contribute  —  
Share this
By Tony1
Coding in the Wikimedia Foundation offices

The Wikimedia Foundation has withdrawn the temporary "superprotect" it had created and invoked on 10 August to prevent DaB., an administrator on the German Wikipedia, from continuing to wheel-war to disable the new software feature, Media Viewer. In her announcement, the Foundation's executive director, Lila Tretikov, made it clear that the WMF "needs to be able to make an ultimate determination after receiving community feedback regarding production changes that impact all users." The superprotect action has brought to a head a steadily rising drama on three WMF content sites: the German and English Wikipedias, and Commons (the last significantly driven by German-speakers).

The wording in the English Wikipedia article on Media Viewer belies the storm that has engulfed parts of the Wikimedia movement. It states simply that "this multimedia browser displays images in larger size when you click on their thumbnails, as an overlay on the current page. To reduce visual clutter, all information is shown below the image, and can be expanded at a click of a button." The commotion from the launching of this product on the last three projects—perhaps those with editors who feel the greatest investment in their now well-established workflows—has ricocheted around a confusing array of issues. On the surface, there are two flashpoints: one is the superprotect itself, and how it might have impacted on the relationship between the Foundation and the editing communities—especially the German-language community, which has a long-standing and justifiable pride in its achievements and an aversion to US-based centralisation. The other is the timing of the opt-out launching of Media Viewer, and whether the software should have been opt-in by default until editors' concerns about certain features were resolved. However, the drama reaches into yet more complicated and sensitive matters, with implications for the technical and social aspects of the Foundation's software development and release, and the status of local community RFCs in relation to software development.

The wiki wall: A caricature of the superprotect dispute, now displayed in the category on Meta that covers pages related to the issue—11 of them alone devoted to "Media Viewer consultation". The cartoon is a play on the fall of the inner German wall in 1989.

Aside from the Foundation's history of less-than-ideal product launches, there are several important precursors to the current technical–social polemics. One occurred after the Foundation's launching of Visual Editor in mid-2013 (now generally conceded as premature, and still in development more than a year later). On 23 September, English Wikipedian Kww edited MediaWiki:Common.js to disable the newly rolled-out software. The action—which was at least consistent with the results of an RFC on the English Wikipedia—broke what had been an implicit taboo against community administrators injecting code into the MediaWiki namespace, unilaterally modifying the interface for all users of a project. This appeared to establish a precedent: in November 2013, DaB. edited the German Wikipedia's site-wide CSS to disable the link to the newly introduced Beta Features, a program that allows users to test new features on WMF sites. His intrusion, without RFC support, has only just been reversed.

DaB. is a longtime Wikimedia Germany functionary who gained wide respect for his almost single-handed technical stewardship of the chapter's Toolserver (2005–14); that suite of servers, physically located in Amsterdam, gave the whole WM movement access to a range of functionalities developed by the volunteer technical community. When the Foundation decided to displace Toolserver two years ago with the WMF-run WikiLabs, DaB. spearheaded an unsuccessful push to keep the existing infrastructure.

The putative justification for DaB.'s edit to MediaWiki code to disable Media Viewer was an RFC on the German Wikipedia that gained majority support for two actions: to switch off Media Viewer, and then to insert code that would make it opt-in rather than opt-out. There were entreaties to DaB. by several editors that he not execute just the first without the critical second of these RFC votes. After a Bugzilla request was closed as "wontfix" by the WMF, DaB. injected code into the JavaScript that not only disabled Media Viewer on the German Wikipedia, but according to one complaint broke the design of file description pages. While DaB contended that he was looking for a fix to make opt-in possible, the question arises why he had not prepared both actions in advance and performed them in quick succession. The change was reverted by a German Wikipedia administrator, Raymond, and further wheel-warring with the WMF led to the now hugely controversial superprotect.

Despite users' ability to disable Media Viewer in their preferences pane, the opt-in/opt-out issue has been part of a storm of protest on three WMF sites.
There was furious reaction on the German and English Wikipedias, and Commons—nowhere more vehement than on the German-language site, where the Foundation's blunt overriding of a local administrator action seems to have unleashed long-held feelings of interlinguistic frustration. The Signpost understands that many German Wikipedia editors agree with DaB.'s actions, although there is evidence that this is not a unanimously held opinion.

What has probably caught most Wikimedians off-balance has been the emotional ferocity that now swirls around both the temporary superprotection and the Media Viewer release—sometimes in ways that make the intercultural and the technical hard to disentangle. DaB., for example, has written: "my action has shown how mendacious and power-mad the WMF is". Co-founder Jimmy Wales was sufficiently concerned that he wrote on his talkpage, inter alia:

Among the first-day responses to Jimmy's call for "constructive dialogue" on addressing editorial concerns about Media Viewer were: "you are completly blind to the facts", "Your autocratic vision will destroy Wikipedia", "all the new garbage the foundation is introducing ... The foundation only gets more abusive as their software gets worse", "throwing dust in the eyes of the fools", and "ditch Möller [vice-president, Engineering and Products]". Despite the invective, Wales persisted in his attempts to persuade participants that dialogue between the engineering and editorial sides is the way to go—marred only by a less-than-diplomatic reference to "climbing the Reichstag".

Pete Forsyth—whom Erik Möller threatened to desysop when he tried to disable Media Viewer on the English Wikipedia—has authored and promoted an anti-superprotect petition (now the subject of an application for a banner advertisement). But in a clear separation of the superprotect and software-development issues, he wrote on Jimmy's page:

Notably, Jimmy doesn't support "superprotect as it is currently implemented (particularly not as a "staff-only" right—that kind of thing just introduces unnecessary divisiveness)". However, he endorses the principle: "that there is no reason to allow admins to disrupt Wikipedia to make a point by editing the sitewide javascript. I think there can be reasons for technically proficient admins to edit sitewide javascript but it is a major security risk and potential point of conflict and so in general I don't think there is any valid objection to shrinking the group who has that right."

He continued: "It makes no sense in the long run for us to have a situation where hundreds of wikis each have a completely different configuration based on local voting. That is not a viable process and we already have huge problems to the extent that it has happened. ... We need to be pushing hard for reunification of software features across the projects."

User:Pine told the Signpost that "the WMF is not the only big organisation that has products blow up on the launching pad: just look at Microsofts's Vista, Windows Me, and Office 2007 for Windows." Nevertheless, the Foundation and its new executive director, software expert Lila Tretikov, face a challenge to the cohesion of the international movement and the ongoing credibility of the WMF's software development program. There is now evidence of a flurry of activity among staff to consult with editors. Among this is the suddenly more obvious presence of the Community Engagement (Product) staff, who were assembled on a temporary basis to assist with the roll-out of Visual Editor in mid-2013. These staff now appear to have longer-term presence, and late last year were transferred to the "Product" part of Engineering and Products. Rachel diCerbo took up the senior position to run CE (P) in May, and if this group of staff succeeds in their mission, the Media Viewer dispute might be the last time matters get out of control. On 29 August, the staff set up a page specifically for Media Viewer consultation, although regrettably the talkpage seems to be dominated by broadsides; and a draft of a process ideas page for community–engineering discourse appeared to mark the arrival of the new senior staff member, "inviting users to brainstorm ideas to improve how software components get build [sic] and delivered to communities".

The Signpost will provide future coverage of how the WMF intends to revamp its software development process, including its liaison with editorial communities, and whether attempts at user requirement documentation such as the Media Viewer roundtable last year can be vastly improved. Among the issues that may become important are the WMF's unique situation as a technology organisation that has little control over "local innovation space" such as templates, which appear to be nothing short of an international mess; a proposal to create a WMF Board volunteer technical committee; and the ways in which engineering documentation and testing might serve to deliver products more smoothly to the movement, including an examination of the relationship of the WMF's software product engineering to models such as Agile and Waterfall.

Editorial note: The author interviewed Pine, a regular Signpost writer, along with three other Wikimedians, to gather background information for this article. Pine had no role in drafting the text.

+ Add a comment

Discuss this story

These comments are automatically transcluded from this article's talk page. To follow comments, add the page to your watchlist. If your comment has not appeared here, you can try purging the cache.

There's a malformed URL at the end of the first paragraph which is missing a question mark; the correct link is Diskussion mit DaB. -- Michael Bednarek (talk) 12:17, 31 August 2014 (UTC)[reply]

Fixed. NE Ent 13:35, 31 August 2014 (UTC)[reply]
Thanks Michael, and NE Ent, who beat me to it. Tony (talk) 13:40, 31 August 2014 (UTC)[reply]

I have read sentences like "The Signpost understands that many German Wikipedia editors agree with DaB.'s actions, although there is evidence that this is not a unanimously held opinion." now a few times in those discussions, and it leaves me always again flabbergasted: Sure there are different opions, after all the German Wikipedia is not North Korea. His act that i don't fully agree upon since it was not setting the Meinungsbild result properly, but approve since it was an act against the abuse of the power of the WMF, was and is debated. As it was the initial poll and everything further. And i have read of other editors who agree that much of the support for DaB. came as reaction of a feeling to be putted in a group together who gets surpressed by the Foundation, not by pure joy of what and how he did. But back to my point: One of Möller's favourite things to state on the Mailinglist is that there also would be people voting against the Meinungsbild proposel and against the disabling of Superprotect in the Umfrage and so on. That is naturally true, since those are not votes like in the German Democratic Republic or in other dictatorships, but democratic votes where also the opposition can take her position freely. What is striking to see is that there is no willingness to see that those polls had normal and then big numbers of participation and a pretty high majority of votes against the WMF position. Möller's mentioning of the small opposition while ignoring the democratic voiced position of a large majority is intolerable to me. And this denial of recognition for the democratic vote of the editor community is what makes me so angry. I find it in Möller's staments, i find it on Jimbo who doesn't see that the Foundation without votes in the communities is not able to produce good software so the democratic element of Wikipedia won't make that so much worse ... And the surprise about disagreement and opposing voters in a democratic project takes me by a surprise that feels not really good to me. But anyway, thanks for the article. The coverage of this topic is needed since the underlying conflict is crucial. --Julius1990 (talk) 14:02, 31 August 2014 (UTC)[reply]

Interesting for me is always, that the WMF really thinks they act for a mayority of users and the authors who vote against the MV are a minority that understands nothing. Maybe - but these users said what they think, said what they want. Clearly. With a clear mayority opinion (my personal opinion was not by the mayority! But in a democraic system I have to respect the mayority!). But where is the RfC of the WMF? They say, they act for others - but did the really do? Where I can read what they said? What they want? The WMF only states things without any proof. The authors at de:WP had a proof. A democratic proof! I know, that the WMF thinks, we authors act childish. But at the end - this time it is definetly the WMF, that acts very childish. "We know better, because we know, what others want and think, others, we never aked about it...". It would by fine, when the Board and the Office would understand one thing: We, the authors, photographers, volunteers, are Wikimedia!. The WMF is here for us. Not we for the WMF. Without us you all in Frisco would be nothing. And nobody of you would have this job. We are the people who work out with our unsalaried work your paycheck! We don't need to be ruled by the WMF. But you people in the office have to understand, that you are here to support us! Or we don't need you! Marcus Cyron (talk) 14:49, 31 August 2014 (UTC) PS: Why I should be interested in what "Jimmy doesn't support" - he is not the person who has to decide this! The en:WP should start a Jimmy emancipation.[reply]
It's more of a two-way street than that. Without the computing resources that the WMF provides, there would be no Wikipedia, and we editors would not be able to accomplish anything. Also, most of the money that supports the WMF comes from readers, not from editors. The MV was specifically designed to give a better experience for readers, who don't participate in votes. Looie496 (talk) 15:07, 31 August 2014 (UTC)[reply]
What generates the big amount of donations is the content of Wikipedia, not the software. Without the content, noone would take notice of the work of the WMF. So of course the power is not one-sided. It would be enough, to inform the donators about a disconnect between the organization, they are supporting, and the editors, that build all the content, and the donations will decrease. So it is in both parties interest, that there is not such a disconnect: The Foundation will earn enough money to produce good software for editors and readers, and the editors can work happily with tools, that are fulfilling their urgent needs (which are quite a lot). --Magiers (talk) 16:35, 31 August 2014 (UTC)[reply]

I would like to note that the Wikimedia Foundation's disrespect for community consensus together with the introduction of an exclusive right to overrule local elected administrators in the influential German blog has been compared to the authoritarian thought of Carl Schmitt who once said that only he is souvereign that decides on the state of emergency. I also hold that this was a reactionary move by the Foundation that does not fit into a world where people increasingly claim the right to rule themselves, not to be ruled by someone somewhere from "above" without democratic legitimation. I agree to most of my fellow editors that the WMF exists for the sole purpose of assisting the editors and providing a platform we can work on. We are Wikipedia. It has often been said by Wikimedia officials that this is a movement. It is now up to them to make this happen, as a movement is always ruled from its base, or else it moves on to another framework that better serves its purpose. This dispute is not about software, but about self-determination.--Aschmidt (talk) 16:14, 31 August 2014 (UTC)[reply]

"We are Wikipedia"? No. Wikipedia is a stool that stands on three legs: (1) the editors, who provide the content; (2) the WMF, who provide the software and servers; (3) the readers, who provide money, motivation, and new editors. All of those legs need to be solid in order for Wikipedia to stand, and all of them need to be balanced. Looie496 (talk) 19:41, 31 August 2014 (UTC)[reply]
I agree with you, Looie496. Referred to the first two legs, I'm confident that the CE(P) team is able to improve the situation. It's a challenge though, as mentioned in the news article. --Patkenel (talk) 20:50, 31 August 2014 (UTC)[reply]

I was struck by one bit of dialogue on Wales' talk page (my emphases):

Another journalist who is dead wrong (according to Jimbo/WMF) [1] I'm sure. And another: "The Foundation has a miserable cost/benefit ratio and for years now has spent millions on software development without producing anything that actually works. The feeling is that the whole operation is held together with the goodwill of its volunteers and the more stupid Foundation managers are seriously hacking them off." JMP EAX (talk) 09:26, 29 August 2014 (UTC)

Other than the extreme nature of the comment ("without producing ANYTHING" is too strong) why do you think I would disagree with that? This is precisely the point of the new CEO and new direction - to radically improve the software development process. That statement, while too strong, is indeed an accurate depiction of what has gone wrong. I've been frustrated as well about the endless controversies about the rollout of inadequate software not developed with sufficient community consultation and without proper incremental rollout to catch showstopping bugs.--Jimbo Wales (talk) 09:43, 29 August 2014 (UTC)

If we're looking for reasons for the current situation, this is where most of them are to be found. And there is as yet little indication that things will really be any different with the elephant in the room, Flow. Andreas JN466 21:07, 31 August 2014 (UTC)[reply]


I have made statistics of from where the signatures on the petition list come from.m:Talk:Letter_to_Wikimedia_Foundation:_Superprotect_and_Media_Viewer#Statistics and when you take the number of active editors into account you find that it is actaulla Italians that is second most acticve on the list even if the absolute number from Enlish is bigger.--Anders Wennersten (talk) 16:10, 31 August 2014 (UTC)[reply]

The statistics of both the original Media Viewer RfC and the Superprotect RfC held in the German-language Wikipedia are available in English translation here on Meta. What is remarkable about the Superprotect RfC is the high level of participation: the vote on the main proposal (requesting revocation of superprotect) had 664 people voting in support, with 103 opposed and 32 abstentions, for a total number of participants of 767. I believe this is a higher participation than any other RfC in Wikimedia history, with the exception of the SOPA protest vote held on the English Wikipedia in 2012, which had a slightly higher participation (but included more IPs and single-purpose accounts). It's also worthwhile viewing these numbers in relation to the size of the German Wikipedia community. In June 2014, the last month for which figures are available, the German Wikipedia had 843 editors making more than 100 edits a month (compared to about 3,000 in en:WP). Moreover, the yes-voters in the Superprotect RfC supporting the request to lift superprotection included 98 de:WP users with admin privileges or other advanced permissions vs. just 4 such users among the no-voters. (The total number of admin accounts on the German Wikipedia is 252, vs. 1,400 in en:WP.) Given the greater size of the en:WP community, the German Superprotect RfC result is equivalent to an en:WP RfC with 2,365 votes in support (including 544 administrators) and 366 (including 22 admins) opposed. Perhaps this may help communicate an idea of the strength of feeling. Andreas JN466 20:53, 31 August 2014 (UTC)[reply]

The high participation can be easily explained: this RfC and related discussions have been canvassed like nothing else I've seen in my Wikipedia history. I saw one announcement at VP, and got several talk page messages about it. SOPA was never advertised so widely (through user talk page mass messages). Nor was any other vote I am aware of. In other words, it's a fallacy to say that people care about this more than about other issues; it's simply that the (mostly) German editors who organized this campaign ignored WP:CANVASS and related policies to a degree much higher than is usually a custom here. To be frank, I find it problematic. The SOPA issue was IMHO much more vital, yet the community couldn't reach consensus on whether to advertise it, prevented WMF from doing so (by saying "no consensus"), and overall ensured that very few people know about it. Whereas here, an issue of lesser importance is being canvassed the hell out of, and the reason for it? SOPA was supported by WMF and WMF tried to play by the book; this anti-WMF MF hate campaign is organized by individuals who don't operate under the same restrictions. I am not fond of MF myself, but I find abuse of statistics to show fake support for this issue annoying. Most people don't care about MF much; they are just canvassed to express their view about it on a scale mostly unprecedented in the history of the WMF movement. --Piotr Konieczny aka Prokonsul Piotrus| reply here 04:37, 1 September 2014 (UTC)[reply]
The German Superprotect RfC was advertised through a site notice for a few hours, and was then downgraded to a watchlist notice for the remainder of the seven days it ran. I somehow doubt you could organise an RfC on anything on en:WP that would have similar percentage participation, and a similarly clear result. Andreas JN466 19:30, 1 September 2014 (UTC)[reply]
Beside that the critic on canvassing is pretty pointless since the WMF didn't want to accept the normal not canvassed vote on German Wikipedia because it wasn't canvassed out to as many people as possible. --Julius1990 (talk) 20:31, 1 September 2014 (UTC)[reply]
Yes, we know. The terrible germans. Nothing more need to say. We are all bad. Very bad... It is in our blood. Marcus Cyron (talk) 22:32, 1 September 2014 (UTC)[reply]
  • It's important that there be an article on this issue, but I do object to you writing twice that DaB "hacked" the software. That word has a negative image, due to the press presenting programming as something criminal. It would be sufficient to write "changed". I also object to the statement that WMF provides the software – this, too, is often contributed by the community or by other communities (MediaWiki). I agree with MarcusCyron, even if not quite as emotional, that it is important that the people who work for the foundation understand their position. They are here to serve the community. And the community is not the readers, but the editors and admins. That we happen to have millions of readers is very cool, and it is great that they pay so much money to keep the project afloat. But without the community, Wikipedia would be just another advertising platform filled with garbage. I feel that every person working for the Foundation needs to have a goal of producing 100 (or 50 or whatever) edits a year in their own "free" time, in order to understand how much WORK this is, but also how proud one is to be contributing. And that is what gets the emotions churning. Lila needs to understand that there are many issues boiling under the surface of the community, soon to break out like Bárðarbunga at the slightest tremor in some question. Superprotect was NOT a good idea. And all software rolled out must be opt-in by default, especially given the track record of the WMF. Learn to discuss, to grow slowly, and above all, to listen to the community – the community of editors and admins, not what some opinion poll or such has determined to be the will of the readers. WiseWoman (talk) 06:18, 2 September 2014 (UTC)[reply]
    • Wisewoman, ok, my US Encarta dictionary says: 2 [ no obj. ] gain unauthorized access to data in a system or computer: they hacked into the bank's computer. [ with obj. ]: someone hacked his computer from another location. (as noun, hacking): outlawing hacking has not stopped it. Also "program quickly and roughly."

      I do increasingly hear it used non-pejoratively, by hackers themselves. But I'll replace it with "changed". Tony (talk) 06:33, 2 September 2014 (UTC)[reply]

    • Indeed, Wisewoman. Terribly biased article in usual Signpost style. --Nemo 07:04, 3 September 2014 (UTC)[reply]

Really? I am astounded by all the drama around this minor thing. I could not care less whether new software features are opt-in or opt-out. --Hispalois (talk) 17:40, 2 September 2014 (UTC)[reply]

  • Yeah, that's really quite astounding. I can't fathom why the WMF invented and deployed this weapon af mass alianation superputsch in a hurried cloak and dagger manner against the german community. It was just about opt-in/opt-out, and some slightly botched js-programming to implement the MB(RfC), if they had just made the right programming, or waited a few hours until some other german admin would have done it, there would not have been this buzz about it. But no, MV, some superficial bling-thing, was treated by WMF like a matter of life or death, and it had to be deployed asap to all, even if it couldn't read a lot of attribution and licence data, didn't fit in the expected handling of WP pages etc. It was obviously someones pet project in the WMF that had to be implemented, come what may. And regarding "hack": Superprotect was definitely a malicious hack, DaB.s actions were mere a bit botched programming, like the MV itself. --♫ Sänger, superputsch must go (talk) 07:45, 3 September 2014 (UTC)[reply]
  • As Dean Inge said, "There are two kinds of fools: one says, 'This is old, therefore it is good'; the other says, 'This is new, therefore it is better.'"

    I was particularly struck by LilaTretikov's remark that "MV is such peanuts, we should really not be spending our joint mental cycles on."[2] I am not sure which is more astounding: the amount of anguish occasioned by the rollout, or the amount of resources committed to developing it. Nuts! ~ Ningauble (talk) 11:35, 3 September 2014 (UTC)[reply]


The Signpost · written by many · served by Sinepost V0.9 · 🄯 CC-BY-SA 4.0