Global Collaboration/Reports/2017 October Offsite

The Global Collaboration team met for a week long offsite in Bengaluru, India from 23rd October to 27th October, 2017. We had 4 days of structured sessions during which the team members talked about the projects they have previously worked on, ongoing projects, their pet projects, and new ideas that they would like to explore together. We also had unstructured hours of quiet work time. The team also discussed about operational logistics and the impact of the merger.

We also had a day of excursion to see the Mysore palace, toymaking in Channapatna town and the silk cocoon auction center at Ramanagara.

Primary objectives

The primary objective of the offsite was two-fold: to share the legacy that the ex-Collaboration and ex-Language teams have brought into Global Collaboration, and to explore future opportunities as a combined force.

Legacy

On the first day, team members presented the projects that they had completed in the past as the Collaboration and Language teams. Joe and Roan led the presentation talking at length about New Filters for Edit Review (for RecentChanges and Watchlist), which provides a newly designed interface for users to provide helpful review using ORES-backed filters for assessing quality and user intent. This needed extensive testing. We also spoke about the revival of interest in the Structured Discussion project backed by the results from the survey led by Benoît. Discussion flowed around the core use case of using it as a viable alternative to the traditional talk page, and also exploration of other use-cases where this structured conversation format can be included to support interactions.

Amir, Santhosh and Niklas presented an overview of the history of internationalization (i18n) engineering for MediaWiki projects done by the ex-Language team since its formation (2011), particularly the generic libraries providing core support for language rules, grammar, i18n standards. These libraries are open source and already used in other projects. The official support status for fundamental i18n work was discussed with an agreement that this is important for this not to be dropped, especially since it aligns with our strategy for our products and workflows to be inclusive about multilingual needs. Universal Language Selector, Compact Language Links, Translate extension were discussed along with Content Translation.

Future

Discussions on future projects of the team were done in two parts - general exploration of ideas, and near term projects. With a view on the Movement Strategy document, the team explored the idea of products that can be built for audiences with features than can overcome geographical barriers, and establish participatory practices of building knowledge equity to provide a valuable platform for content procurement and distribution.

Existing projects and practices were discussed along with relevant sections of the Movement Strategy document in an effort to map ideas for exploration. The focus was particularly on the vision for Wikimedia to be a essential infrastructure for free knowledge by 2030 with the greater objective of providing knowledge as a service and knowledge as an equity. Evolving social norms of information sharing, for instance importance of verifiability of information from within our platform via appropriately designed tools for improving credibility of the content that people come looking for in Wikipedia could be key points in our vision for future products. It was also emphasised that we need to be mindful that our products and libraries ought to be suited for better reuse and wherever possible, extendable beyond MediaWiki platforms.

For the immediate future, our focus is expected to remain on ongoing projects that are nearing completion. Suggestions were made to make Structured Discussion more open for a global appeal like cross-language interactions.

Other discussions

Community interactions

We dedicated two sessions to talk about community programs and interactions, including technical partnerships with volunteer developers. Ravishankar A from the Partnerships and Global Reach team of WMF, joined in and introduced us to various local programs and the challenges in Wikipedia’s brand awareness in many places, including India, that has restricted our reach to a wider audience. He emphasised the need for in-person mentoring, that has helped during a local program to encourage teacher’s from Government schools to create new articles for Tamil Wikipedia.

We continued the discussion on how to have better interactions with the community when interfacing through community liaisons. This also includes product documentation and other resources that will help users’ better understand the functionalities and also provide feedback about the challenges and shortcomings that can be used by the development team in a meaningful day. Importance of data backed research and usage results were discussed.

Relevance of the volunteer developer community, especially their participation in our open source projects needs constant support. This requires matching their expectations of partnership in a way that their contributions are reviewed and responded to within a reasonable timeframe. The development team can declare their expected response times and level of support for particular projects as per priority and expertise. This would not only help us maintain a healthy partnership with the volunteer developers, but also help us prioritise projects that can be offered during community events like GSoC or outreachy, without distracting the team from its major deliverables.

Engineering practices for Global Collaboration

This session was a discovery session to understand engineering work styles currently prevalent inside Global Collaboration, and how they can be aligned for find a common development environment for future needs. In addition there were discussions of shortcomings in our practices which may need more attention. There was emphasis on better management of tech debt, scheduled release testing and unit tests. Documentation for both new team members as well as for general clarity was mentioned as key points, especially for unusual workarounds that are often not clear. The team also discussed communication channels and what could be a comparatively better medium for quick feedback between developers, designers, QA and PM.

Testing practices was a broad topic that was discussed with proposal for further focused discussions. Suggestions were made to prepare a protocol for periodic performance testing. Challenges like timely feedback on UI for quick changes, testing practices during code review or tooling overheads for browser testing, were highlighted.

Pet projects

Team members presented their pet projects which was a relevant exercise for individuals on the team to know what interesting problems their teammates are trying to solve. Projects showcased were:

Pau

1. A book for designers,
2. Adapting ‘Wiki loves monuments’ for an improved mobile view
3. An experimental mobile messaging app

Moriel

1. http://rtl.wtf/ - displays all content in RTL direction for experience and experiments
2. http://rtl.works/ - provides guidance about issues that can cause problems when displayed RTL text
3. http://www.neutrality.wtf/ - switches gender terminology around in text provided

Benoît

1. Documentation for users of French Wikipedia

Niklas

1. Continuous improvement to machine translation quality by soliciting specific kind of feedback from translators and deploying fixes more quickly (part of his PhD process). Several areas with possible improvements were highlighted: better sources for translation, content that matches the user’s interest, tooling that makes impact of work visible, and a continuous feedback loop.

Santhosh

1. Malayalam typography and font projects - listed at https://github.com/santhoshtr . Special note about the Meera font that has been engineered by him and which is used on the new series of currency notes in India (for Malayalam text). Also about a children’s science magazine which has switched to a font designed by him named Manjari.

Amir

1. A Telegram bot that makes it easy to translate messages from translatewiki.net using a mobile phone. This is particularly helpful for users who are heavy users of mobile phones and would like a seamless environment to continue translation when they are away from their desktops.