The Signpost

Disinformation report

Croatian Wikipedia: capture and release

Contribute  —  
Share this
By Smallbones

A decade-old case of "project capture" of the Croatian Wikipedia (Hr.WP) by nationalist administrators may have been resolved with the help of a report published this month on Meta by the Wikimedia Foundation titled "The Case of the Croatian Wikipedia". The report was authored anonymously, presumably to avoid harassment, and is an independent view of an expert on the subject matter.

The admins, led by Kubura, inserted disinformation and used sockpuppeting and other abusive tactics, according to a separate RfC which globally banned him last November. Blablubbs, who participated in the RfC, said that Kubura had an "army of socks". Blablubbs decided to help at the RfC "partially because of the whitewashing ... and partially because of draconian crackdowns on dissent inside the project".

The admins were linked by the report to Croatian nationalist positions by their downplaying the UN war crime convictions of Croatians who fought in the 1991–1999 Yugoslav Wars, their use of biased unreliable sources and by their support of the World War II era Nazi-puppet state, the Independent State of Croatia (NDH), as well as the military group, the Ustaše which the report calls "terrorist".

The report echos earlier accounts of administrator abuse including a 2019 article in The Signpost, "The curious case of the Croatian Wikipedia", Croatian and international news stories going back to 2010, and complaints by Wikipedia editors starting about 2007. The report concludes that "Hr.WP had been dominated by ideologically driven users who are misaligned with Wikipedia’s five pillars, confirming concerns about the project’s integrity from the global community."

Articles are being re-written and disaffected editors are rejoining the project. The report notes this progress but warns that the transformation is not complete and that the banned admins may use new accounts to try to recapture the project.

The report also observes that this case highlights a "significant weakness in the global Wikimedia community and – by extension – Wikimedia Foundation platform governance."

The report

The WMF began its planning for the report in November 2020 as the RfC on banning Kubura was in progress, but the author's investigation started in February 2021. He is an external expert on the subject matter and provides three recommendations to the WMF. Jan Eissfeldt, Global Head of Trust and Safety at the WMF told The Signpost that the author is a native speaker of Serbo-Croatian with "decades of relevant international experience analysing ... patterns of organised disinformation." The report states "opinions expressed in this report are those of the author and do not necessarily reflect the official policy or position of the Wikimedia Foundation."

The Croatian, Serbian, and Bosnian Wikipedias are unusual in that they all separated starting in 2003 from the Serbo-Croatian Wikipedia, which continued to exist. All these languages are mutually intelligible variants of Serbo-Croatian, which is termed "pluricentric".

The report states that

this structure enabled local language communities to sort by points of view on each project, often falling along political party lines in the respective regions. The report asserts, furthermore, it deprived the newly-created communities of editorial diversity that normally guides and underpins the traditionally successful process of editorial consensus in other pluricentric language projects.
— Report, p.2

The limited number and diversity of editors on the new Croatian Wikipedia allowed it to become politicized, and allowed Kubura, his sockpuppets, and followers to capture the administrative structure of the project.

Evidence of this capture is shown in the report's section on "Key findings and case studies" (pp. 15–35) which makes up one-third of the report. It includes subsections on

This section is the core of the analysis and may set the standard if any future reports of this type are needed.

Based on these findings the report makes three recommendations to the WMF and the Serbo-Croatian communities:

  1. that the Croatian community seek to "continue re-establishing a robust local governance system, requesting oversight and support from the rest of the Wikimedia movement as needed;"
  2. that they seek to unify the selection of admins and functionaries with other Serbo-Croatian communities; and
  3. that they explore a full reunification into the original Serbo-Croatian language project.

Adding some urgency to these recommendations, the report warns that, as currently constituted, Hr.WP is at risk of being recaptured by nationalist editors and admins.

An additional observation – strengthen global governance

Following the recommendations, the report's author makes a statement that goes beyond the Serbo-Croatian community and the Hr.WP disinformation problem.

The evident failure of the Meta RfC system to resolve the structural misalignment of Croatian language Wikipedia and the lack of an adequate alternative pathway to resolution, points at the significant weakness in the global Wikimedia community and – by extension – Wikimedia Foundation platform governance. This is a problem for public and regulatory confidence in the self-governance model provided for within the framework of the Foundation’s policies ... While devising possible pathways to address this identified bigger challenge is beyond the scope of this disinformation assessment, it strikes the author of this evaluation as increasingly important to resolve in the light of heightened regulatory scrutiny of user-generated platform models, including Wikipedia ...
— Report, p.14

The Signpost asked Jan Eissfeldt of the WMF for his reaction to the report's observation. He recognizes that disinformation is a growing problem, and emphasized that the WMF would work with the communities as openly as possible. For cases where safety is a potential problem, they might need to work through stewards or other trusted users. "We are investing in our movement's capacity to identify and respond to all kinds of influence operations, including those led by government actors, to ensure the accuracy of the information shared on Wikimedia projects. An example of this was the taskforce we put together ahead of the U.S. presidential election."

While not promising to start any new program to systematically evaluate disinformation problems, he said "the Foundation aims to conduct project evaluations, in collaboration with volunteers in the Wikimedia movement, to explore potential issues in projects openly and transparently."

How well did the WMF respond in the case of the Croatian Wikipedia? He says that the WMF "did not adequately understand some of the unique risks now identified in the report," in particular the risks of having separate Wikipedias for parts of pluricentric language communities.

In this issue
+ Add a comment

Discuss this story

Well I did write "*may have been* resolved *with the help* of a report" (new emphasis) Yes the admins, folks at the RfC, the communities, etc deserve all the credit. Make that ten times all the credit (from 10 years worth of trying with little help from the WMF). At first when I saw the report, I thought "Here comes the WMF trying to grab some credit". I don't think so now. The official reason for publishing it was "for the sake of transparency" which aligns with what WMF employees are usually allowed to say. The report itself I think does deserve some credit - not for this round of the battle - but for the next time. So there's a method available to deal with this type of thing that might only take 3-4 years, rather than 10. But I shouldn't get started on this ... Smallbones(smalltalk) 22:13, 27 June 2021 (UTC)Reply[reply]
So if it was "may help resolve future hr-wiki issues" then that would be fine, but saying may have been resolved without knowing if it has seems...premature, at best Nosebagbear (talk) 22:51, 27 June 2021 (UTC)Reply[reply]
Well "*may have been*" does mean "*may have been*" Words do matter in journalism. But at the same, you do have me wondering whether I was being completely honest in my reporting here. There's a difficult step in writing a straight news story - and I do believe this needed to be a straight news story before any real analysis could be done. The step is to try to give up all your biases and POVs at the start, including some of your skepticism and assume the folks you're dealing with are acting in good faith. Then as the facts become clearer you can start asking deeper questions. Perhaps I could have concluded that this was a show report with no real meaning. But I didn't and still don't believe that. In any case I got the report on Thursday and some good info from the WMF on Friday, and more or less made sense of a 60 page report in 1200 words (or whatever), so I'm not going to be too hard on myself. Believe it or not, this is fun. Anybody who wants to try it, The Signpost needs some good news reporters. Smallbones(smalltalk) 02:15, 28 June 2021 (UTC)Reply[reply]
Croatian and Serbian languages are not variants of the so-called Serbo-Croatian language. There was a great political effort in the past to make Croatian and Serbian one language, and Croats and Serbs one nation. Neither occurred; regrettably if you ask some people today. A number of those people with regrets exist even among linguists; and also among wikipedia editors - mainly among the editors of Serbo-Croatian wiki.
You have there, though, closely related languages: something like Swedish, Norvegian and Danish, which are closely related, but NOT variants of the same language. The clue to understanding the "complication" with Serbo-Croatian is in Bosnia and Herzegovina (it is, between the territories of Croatia and Serbia) - where for a very long time the schools thought the children (Croats, Serbs, Bosnians) the same language. Well, this language ("Serbo-Croatian") is - more or less the modern Bosnian language. Speakers of Croat and Serbian languages understand Bosnian, too (like speakers of Slovak understand Czech language, but the attempt to make Czechoslovak language - failed). However, only Croats from Bosnia and Hercegovina and only Serbs from Bosnia and Hercegovina tend to be proficient in the Bosnian language.
I will cite the examples from here:
A slight difference is demonstrated by:
Sr. (Serbian) Sačekaj minut da uporedim tvoja i moja dokumenta.
Cr. (Croatian) Pričekaj minutu da usporedim tvoje i moje dokumente.
‘Wait a minute so I can compare your documents and mine’.
Greater differences are demonstrated by the following:
Sr. Što ga biješ?
Cr. Zašto ga tučeš?
‘Why are you beating him?’
Sr. U januaru sam rešio da uradim sve što me ranije mrzelo.
Cr. U siječnju sam odlučio učiniti sve što mi se ranije nije dalo.
‘In January I decided to do everything I didn’t feel like doing before’
We could even make up similar or identical phrases that have different meanings in the two languages, or in fact only one of them, while in the other they may sound as nonsense: suprotni pol Sr. ‘opposite sex’ Cr. ‘opposite pole’; Zemljina osa Sr. ‘Earth’s axis’; Cr. ‘Earth’s wasp’ ; prava stvar Sr. ‘straight thing’, Cr. ‘real thing’.
Sr. Odojče igra na zraku. ‘An infant is dancing on the ray’.
Cr. Odojče se igra na zraku. ‘A piglet is playing on the air’.
Or: Pravi zrak igra svoju igru.
Sr. ‘Straight ray is dancing its dance/Real ray is playing its game’.
Cr. ‘Real air is playing its game’.
Kad počinje slovenski čas?
Sr. ‘When does the Slavic lesson begin?’
Cr. ‘When does the Slovenian moment begin?’ RadioElectrico (talk) 15:22, 6 July 2021 (UTC)Reply[reply]


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