Community Capacity Development/Tamil 2016

Abstract

This page describes the capacity development project in the Tamil Wikipedia community in 2016.

Note that some of the relevant materials are to be found on the Tamil Wikipedia, for ease of access to the Tamil community. This page attempts to summarize and track the project overall, for comparative analysis and for the long-term record.

Context

The Tamil Wikimedia community is active, passionate, and well-organized. Most of its contributors are based in India, which was identified in the WMF 2010-2015 Strategy as one of the three focuses for the Foundation's "Catalyst" program. The "Catalyst" program, from the first, left the Tamil community relatively untouched, by preference of the Tamil community.

While the Tamil community has shown programmatic innovation (e.g. Wiktionary partnerships), strong community spirit (e.g. community-funded equipment grants for needy steady contributors), and good partnership-building skills (e.g. the Tamil Nadu state government, the Tamil Virtual Academy), the community's technical know-how has not kept up with the latest technological innovations across the Wikimedia movement.

While standard wiki technology like bot-assisted automation of wiki maintenance tasks was being accomplished by a couple of Tamil volunteers, newer technologies like Phabricator, Lua, Quarry, PAWS, PyWikiBot framework, and the entire Wikidata project, were not being actively used or contributed to by the Tamil community. Before the CCD program, there had not been a single active Wikidata editor from the Tamil community.

Process

During the research phase for the CCD program, the On-wiki technical skills capacity was identified as one particularly relevant for the Tamil community to grow, alongside an expressed interest in Partnerships. The former was chosen as the focus of the CCD pilot program with the Tamil community. A survey was run in November 2015 to determine community wishes and preference. Its results have been posted on the Tamil Wikipedia.

Training materials were developed in collaboration with WMF Operations Engineer Yuvi Panda, who joined Asaf Bartov and helped deliver a weekend-long training session in Chennai in April-May 2016. Active editors were invited to attend the training, and funding was provided for domestic travel for interested editors outside Chennai, as well as for several active Tamil contributors from Sri Lanka (a crucial element of the Tamil community). Approximately 25 people attended the training.

The in-person training took place in Chennai on Apr 30th and May 1st.

The key components of the training were:

A survey and demonstration of a range of Wikimedia tools
An introduction to PyWikiBot
An introduction to Phabricator
An introduction to Quarry and PAWS
A comprehensive introduction to Wikidata

Post-training survey of attendees has been done, with 22 respondents. Here are the results

Materials

Slides from the Wikidata introduction session

Slides from the Wikidata introduction session, in a slightly improved version delivered in August 2016 at WikiConference India 2016
- Video of training at WikiConference India: part1, part2
List of tools demonstrated or discussed (with links)

Evaluation

Funds spent (US dollars)

In addition to staff time, the following costs were associated with the program in Chennai:

Travel (international and domestic) and per diem allowances - ~$4800
Accommodation and transportation for non-local attendees - ~$760
Venue rental - ~$0 (donated by the Tamil Virtual Academy)
Food and supplies - ~$1015

Approximate total cost excluding staff time: ~$6575

Surveys

Qualitative interview in November 2016

This section summarizes responses given by community members Ravidreams, Tshrinivasan, and Balajijagadesh.

Expectations

The opportunity was welcome.
We were proud WMF selected the Tamil community for the pilot.
We looked forward for the opportunity to get together, which we don't get often.
We wanted to learn technical skills not from help pages.

Reflection after the training

We learned much more than we expected.
"Although I am en experienced programmer, I learned much from the training. Questions I had listed in advance were all answered."
"I expected mostly to organize; ended up learning a lot myself."
"I was aware of Wikidata, but found it complicated, too confusing to understand beyond interwiki. Now I think it is the future of Wikipedia. My mind was blown. I was inspired and started contributing massively." (since the training, this volunteer amassed ~200,000 manual edits to Wikidata in under a year --AB)
"Tamil Wikisource was progressing slowly, with too little knowledge. At the training, I was able to learn from my peers how to work with Wikisource, I updated the outdated and unclear help pages, and I adapted the ws_export tool to Tamil. Thousands of downloads are now being made through ws_export. It is the most active Indic Wikisource. I learned regexps; super useful with AutoWikiBrowser for example."
"Re Wikidata: I didn't know what I didn't know. I understood the beauty of it after the training. I started evangelizing for it on Wikipedia. I started two WikiProjects on Wikidata, around Tamil villages and Tamil temples, based on data provided by the TVA. We somehow knew something could be done with Wikidata and that dataset, but had no idea how to go about it before the training. I still enjoy working in Wikidata."

Observed outcomes in the community

"Previously, I couldn't get PyWikiBot to work. Resolved during the training. After the training, I wrote bots and encouraged others to automate tasks. I gave a talk at WikiConference India 2016 on how to write bots with PyWikiBot. I also helped non-Tamil Wikimedians (bn, te) automate tasks."
"I observed increase in systematic activity in [Tamil] Wikisource. Until then, activity was sporadic and disorganized. No updates to main page, etc. CCD was the cause for this growth."
"the training on Phabricator was useful; many people started filing bugs directly on Phabricator. Previously, the wikitech mailing list was our only recourse. It didn't always work; sometimes we would be redirected to a person and they wouldn't respond."
"More programmers (4-5 in Chennai) are now using bots and PWB. More people able to handle ad-hoc automation requests by the community."
"the less-technical people who attended did not engage much more after the training."
"there was great value in bringing in our Sri Lankan colleagues; we do not meet often, many have never met at all; it helped significantly to tighten the community. We found ourselves discussing Tamil grammar and pronunciation around the clock. (even instead of getting enough sleep.) :) Friendships were forged during the event, especially among those who were sharing accommodation."
"People commented that they enjoyed this hands-on training for two days. Most past events were largely celebratory or social. This was perceived as more useful than previous events, which were "all talk".

Does the CCD approach (high-contact, in-person investment by WMF in a specific community to build a specific capacity) seem worthwhile?

Very worthwhile; we don't know enough; learning by ourselves, from written documentation, is very tough. We learn a lot better in person; we benefit from the focus. I didn't believe I could learn as much as I actually did.
We should make more such training locally. For example, now one of us can conduct a local training on Wikidata, in Tamil.
"The quality and depth of the training by experienced WMF staff can't be matched by outsiders. I found the quality much higher than other wiki trainings I have attended, especially in India. Outside the CCD, we don't have many opportunities for in-depth workshops like that. Very few people can come to the international events. We need WMF to come to communities in their own countries, and ideally in their own language."
it was very useful that the audience was mostly technical people rather than a general audience. It is easier to learn in such an environment.

What other capacities would you be interested in WMF offering help with?

More training on creating gadgets and plug-ins.
An internship in San Francisco
A semi-regular hangout, to support continuing education. "It doesn't even have to involve WMF staffers every time."
Wikisource-specific training could help. Maybe several webinars, focusing on one skill each.

Additional Outcomes

The Wikidata tutorial delivered at this training was developed by Asaf on the basis of an improvised talk he gave in Tel Aviv University at the request of Shani who was teaching a Wikipedia course there. To teach Wikidata in Chennai, Asaf prepared slides and examples and expanded the range of techniques covered.

Since the training, Asaf delivered this workshop (using between 2 and 4 hours, as available) on several occasions:

Wikimania 2016 in Esino Lario
WikiConference India 2016
Wikimedia CEE Meeting 2016 in Armenia
WikiConference Ukraine 2016
WikiConference Bulgaria 2016
WikiBirthday 16 hosted by WikiSalon San Francisco, January 2017
Wiki Indaba 2017 in Ghana

Additionally, fellow staffers Alex Stinson and Stas Malyshev adapted Asaf's talk and delivered it at WikiConference USA 2016 in San Diego, and again at the WMF's own All Hands Unconference in 2017.

Each time, audience response was enthusiastic and grateful, and not infrequently (even long-term) Wikipedians would say they had never gotten the hang of Wikidata (beyond interwikis) until they attended this workshop. It is safe to conclude that this training met a genuine and widely-shared need across many communities.

Conclusions and recommended next steps

This pilot project was successful, in that it achieved its main goal: helping the Tamil community significantly gain confidence and skill in the application of technology on the Wikimedia projects, with a particular emphasis on using existing tools, bot frameworks, and embracing Wikidata.
- This conclusion is supported by both the follow-up community survey and the qualitative interview.
Asaf has been delivering two of the three components of this training (Wikidata training and tools demonstration) at every opportunity since the Chennai training (listed above), which allows us to draw the conclusion that it is effective to teach these skills in this way, having done so to both technical and non-technical audiences, both single-community and diverse-community audiences, and both complete beginners and intermediate users.
Recommended next step #1: It is definitely worthwhile to continue offering technical training directly to community members, at both international events (economically reaching delegates from multiple communities) and local or national events (economically reach multiple volunteers from single community; focus examples and questions). This has already been happening ad hoc (by Asaf, and recently by Alex Stinson as well), but can be built into more than one person's annual plan and conference plans. While the in-person aspect seems central to the observed efficacy, it still makes sense to record a high-quality version of these training talks (perhaps at WMF headquarters) and make it available for the use of volunteers (subtitles or dubbing) and for self-study, to help us reach volunteers we aren't reaching (and perhaps won't ever reach) at live events.
Recommended next step #2: WMF should lead a collaboratively-defined core curriculum of technical know-how we would like to ensure is well understood by every community above a certain activity threshold. (cf. List of articles every Wikipedia should have) Once the curriculum is set, an owner (not necessarily from WMF, but if no one else owns it, then a staffer) should keep it up to date with new "core curriculum" technology, and track individual communities' progress in integrating and getting comfortable with that new technology. Based on that matrix, specific training could be designed and carried out, in local, regional, or international settings.
- Most of our communities have at least one "bot person". We need all communities to have more than one "bot person" (redundancy). But most of our communities don't have a "Lua person". Quite a few communities don't understand Phabricator. Or Wikidata. etc. etc.
Recommended next step #3: Recognize that knowing things is not the same as being able to teach them well. This training was met with several comments along the lines of "I've heard talks about this before, but I never got it" or "it never motivated me until now". It would be good to be more proactive and more judicious about the selection of people tasked with teaching and training. This could take two main forms: 1. identify and utilize already-effective trainers among staff (and the movement); and 2. train prospective trainers among staff (and the movement) to make them at least moderately-effective trainers, to increase the pool of available effective trainers. Crucially, any "train the trainers" initiative must include some kind of post-training evaluation in practice, i.e. with the newly-trained trainer delivering training with observation or evaluation, so that only effectively trained trainers are subsequently asked to join the trainer pool. This is something most "train the trainer" events in the movement fail to do.

Links

project page on the Tamil Wikipedia