Two new Wikimedia fellows to boost strategies for tackling major issues

The Wikimedia foundation has just announced new community fellowships aimed at tackling two very different challenges faced by the movement. Both fall within the 2011–12 priorities of increasing participation and editor retention across the projects.

Tanvir Rahman is a steward and active editor of the Bengali Wikipedia, which has about 23,000 articles and 50 active editors—12 of whom make more than 100 edits a month. The project will concentrate on that site to experiment with "on-wiki strategies to encourage and grow the editing community on small language versions of Wikipedia", says Siko Bouterse, the foundation's head of community fellowships.

"Small", though, only refers to the size of the site: ironically, by some counts Bengali ranks sixth among the world's languages in terms of the number of native speakers, with nearly 300 million. We asked Tanvir why more Bengali-speakers aren't editing the site. He said, "Most Bengali-speakers get the English Wikipedia when they Google. A lot of them don't know about the Bengali Wikipedia, and still fewer that they can actually contribute to the site."

"Part of the problem is that Bengali is written in a non-roman script, and to edit you need software that supports this script. Many people don't have it, even though there are now increasingly satisfactory open-source tools, such as Avro Keyboard to which we have a link on our main page."

Wikimedia's increasing global reach is encountering an old phenomenon in new and different ways: languages often don't map onto nation-states. The Bengali Wikipedia is a profound example of this: 160 million speakers live in Bangladesh, where Bengali is the only official language, and about 80 million live in West Bengal, a contiguous state on the eastern edge of India; tens of millions more live in adjacent Indian states to the north. So the national border lies athwart this huge group of people who share long and rich literary and cultural traditions.

But there have been tensions, some of them even recent; for example, Tanvir told us that over the past few years, the Indian government has been blocking the transmission of Bangladeshi television into West Bengal, reportedly citing fears of "anti-Indian" propaganda and coded messages to separatists (the Signpost cannot reliably confirm the details or whether the blocking persists).

Despite these hiccups, Tanvir says, there are obviously strong links between the groups. We asked him whether the Bengali Wikipedia has the potential to bring the two groups together in terms of free information and intellectual pursuit. He says that both Wikimedia India and Wikimedia Bangladesh have been active in supporting the Bengali Wikipedia (Tanvir played a key role in establishing the Bangladesh chapter, which has already begun outreach in the country). He points out that although India has some 400 languages, Bengali is its second most spoken native language, after Hindi. Bengali matters to both countries, and articles on the Wikipedia concern the whole language region.

Furthermore, Tanvir says, West Bengalis are welcome to collaborate with the Bangladesh chapter—anything, he says, that will develop the editing community. "There are possibilities for joint ventures with Wikimedia India on workshops, and we would be glad to help, although so far this has not occurred." On the Bengali Wikipedia, there are "very few" editors from West Bengal, although "one of the four active administrators is from West Bengal".

The project will comprise a number of overlapping phases. Initially, information and feedback will be collected regarding editing patterns and the basic needs of the community, including technical and social issues that might have a bearing on editor engagement. These findings will be analysed and pilot projects designed on welcoming and training new editors, and on the creation of an outreach program. The pilot programs will then be run, and outcomes reviewed, measured, and reported, leading to implementation in the longer term.

Tanvir says the project will open opportunities for transferring to other small wikis what we learn about attracting and retaining editors on one. "Communities may live in very different sociocultural circumstances, but there are likely to be common factors in all small wikis that enable us to formulate more robust strategies."

Another dimension of the editor retention narrative is dispute resolution, highlighted by last week's release of data suggesting that a surprising proportion of Wikipedians find the social milieu problematic, with 23% of respondents rating their fellow editors "arrogant", 13% "unfriendly", 7% "rude", 5% "dumb", and fewer than half "collaborative", among their top two descriptors (question 17). Steven Zhang, according to Bouterse, has "a passion for resolving on-wiki disputes and helping others to do the same". In June last year he was instrumental in setting up the dispute resolution noticeboard (DRN) on the English Wikipedia, and his fellowship will use this as the starting point for developing new tools and strategies to improve mediation.

We asked Steven how he would score the noticeboard out of 10; he says 6–7 might be a fair assessment at the moment. Previously "there were many entry points for dispute resolution, as though they'd grown without any planning at all. Unless you were a seasoned editor, you wouldn't know which one to go to. My whole purpose was to draw the system into a 'single capsule' for low-end disputes, and if a particular dispute needs to go somewhere else, to direct it there subsequently."

The DRN has already made one of the low-end entry points somewhat redundant (the mediation cabal), but Steven is measured in his view of the task ahead: "Bad dispute resolution procedures", he says, "have a very negative effect on editor retention. However, there's no magic fix. The only thing that will maximise our success rate will be more people working in dispute resolution who've got the right knowledge. One of the advantages of the DRN is that it doesn't have to be as formal as the alternatives. People can just drop in and give advice without all of the strings attached to other forms of mediation. This format has been quite successful in attracting more editors into the field. Starting with just two people, we've now grown to more than 10."

What skills are necessary to work in dispute resolution, then? How do you get one religious or political group talking productively to their opposite number? "First, mediators need to know policy, because often disputants will quote policy, and you need to know your way around it yourself. But just as important, you have to look at a complex situation and clearly extract what the differences of approach are, separating the content issues from personal agendas. In the end, if they can't look at the situation objectively there may be nothing you can do, and it probably needs to go to another forum."

How much psychological insight is necessary? "You sometimes need to work out what the mind-sets of the parties are, and you can only do that by going into their edit histories. A lot of patience is required to succeed. Being aware of a number of typical scenarios might help both mediators and disputing parties to see their way through to resolution. For example, there's the ping-pong match, where participants don't listen to each other, the boomerang, where just bouncing back policy explanation does the trick, and the my source is better than yours dispute that can be solved by carefully working out how to present the different points of view in the article text. Then there's the compromise, where the final result is an amalgamation of opinions from parties, and the tough nut, disputes that would benefit from more formal mediation. You need to know when page protection may break the back of the dispute, and how to manage verbose editors, where tighter structure and intervention might work. Occasionally, a dispute has many issues and can be divided into components for better results.

Could admins play a greater role generally? "The problem for admins is that when a dispute is based on deep emotions—like some of the religious disputes I've mediated—they're reluctant to get involved. I think the admin and mediator hats are quite different. Being an enforcer and a mediator at the same time is a very hard thing to do. But having said that, experienced admins who have a detailed knowledge of policy and how to apply it can do a lot of good in resolving disputes."

"While on the issue of admins," says Steven, "I'll take this opportunity to say that I see a lot of RfAs where applicants are given a hard time for having less content creation experience than people think they should have; but my view is that dispute-resolution experience is a reasonable and valuable alternative to a lot of content editing."

As part of his fellowship, Steven will analyse the results of a survey of approximately 1,100 Wikipedians to determine the relationship between demographics and experiences of dispute resolution, and will collate opinions on how the process can be improved and who might be interested in participating as mediators. As part of the Wikimedia Fellow panel at Wikimania, he will outline the history of dispute resolution on the English Wikipedia, the results of his survey, and how they might feed into new strategies. In a related workshop at the unconference, he'll present developed case studies (anonymised) to compare how the audience believes the case should be resolved and how the case actually played out. He'll go on to create an online version so these test cases can be used by editors across foundation projects to improve their dispute resolution processes.

• Milestones: The following Wikipedia projects reached milestones this week:
  • The Korean Wikipedia has reached 200,000 articles.
  • The Cantonese Wikipedia has reached 20,000 articles.
  • The Scottish Gaelic Wikipedia has reached 10,000 articles.
  • The Mazandarani Wikipedia has reached 5,000 articles.