Commons:WMF support for Commons/Commons community calls/Discussion 5 - Tool investment priority

Wikimedia Foundation discussions with the Wikimedia Commons community

• Main page
• Upcoming discussions
  • There are no upcoming discussions for the moment. New meetings will be shown here.
• Past discussions
  • Wikimania 2024
  • Wikiconference North America 2024
  • Community call on Content organization (November 2024)
  • Community call on New media and new contributors (December 2024)
  • Community call on Tool investment priority (January 2025)
  • Community call on Impact and funding model (February 2025)
• Talk page

While participants in these calls highlighted specific tools or features for prioritization, much of the discussion actually focused on approaches to prioritization and principles for development and maintenance.

Support critical workflows, not specific tools

There was strong agreement on both calls that the Wikimedia Foundation should look at a smaller set of critical workflows or functionalities, rather than specific tools. Examples of workflows included upload, batch upload, campaigns, categorization, moderation, and linked open data. It was argued that if a tool is very popular, it is a sign that a vital feature is missing from the core product, and that consolidating these features could improve the user experience and make workflows easier to maintain. The Foundation was also encouraged to explore which improvements to the core product would actually reduce the need for volunteer-created tools, e.g. improved structured data on Commons could reduce the need for categorization tools.

However, one participant noted that developing and maintaining MediaWiki extensions is not easy, compared to JavaScript scripts.

One participant suggested that more tools and gadgets (such as BookReader) could be integrated rather than needing to be individually enabled.

Make long-term investments, not temporary fixes

There was a lot of appreciation for the contributions by volunteer developers, but a consensus that the Foundation should provide sustained, multi-year support for core functionalities, such as UploadWizard and the thumbnailing infrastructure. This was contrasted with the past, when the Foundation had implemented major features and then seemed to move on without addressing bugs or adding new functionalities.

However, it was noted that specialized tools with a smaller user base could be served by the community, with grants, mentorship, and structured support for volunteers and affiliates wanting to make technical contributions.

Maintenance challenges within our ecosystem

Many of the participants in the discussion acknowledged the legacy issues that both the Foundation and volunteer maintainers face.

• Many of the tools and gadgets were originally written more than 10 years ago. Continuing to patch them, while making major changes to MediaWiki (such as new skins, or dark mode) makes them much harder to maintain.
• There aren’t enough people who understand how things work. There is a need to build knowledge and make sure there is capacity to fix legacy bugs that are not owned by anyone.
• We don’t have a good understanding of all of the interactions and dependencies between software. For example, when PetScan breaks, several tools break.
• It was also suggested that when the Foundation used time-limited contractors to upgrade the critical thumbnailing infrastructure (e.g. Thumbor), an opportunity to build and transfer knowledge was missed.

Design and development approaches

There was some discussion of design and development approaches, such as:

• An increased focus on accessibility and usability to lower barriers to contribution, minimize the social investment required for training, and reduce the creation of workaround tools.
• Optimizing new features and tools for mobile use (it was also suggested that the Foundation could invest in Android and iOS apps).

Prioritized features

SPARQL endpoint

The most-discussed and supported request was for an open SPARQL endpoint to enable querying and reuse of structured data:

• “I’d give up ALL the tools I use to be able to have this happen.”
• “There is no point to adding structured data if you can’t then use an endpoint to get a result when you search that structured data.”
• “It’s also essential for adding the structured data as you are using queries all the time for finding what data already is/what data differs/finding errors.”
• “I agree with prioritising the open SPARQL endpoint for SDC, as it unlocks reuse.”
• “Commons Query Service being authenticated has problems for downstream applications.”

Other requests related to structured data included:

• Media search needs to be more multilingual and support hierarchical searches (e.g., ‘trees’ should show related categories like ‘oaks’).
• Categories should be migrated to structured data in a way that leverages all of the work contributors have put into image categorization.
• Adding structured data is currently slow, error-prone, and lacks essential features, making it difficult for users to contribute effectively.
• It is difficult to make reports of issues or missing data for structured data.

UploadWizard

A couple of participants talked about UploadWizard as the main priority, because it is used by new contributors. Specific requests included:

• Adding chunked upload support to Special:Upload because it touches all critical areas for upload support and removes the need for a third-party tool.
• Handling video support, in order to replace video2commons.

However, there is not a community consensus for improved video support, with some saying that we should optimize for images first, and others wondering if we have the bandwidth to support more video, especially in a context where we’re seeing increased scraping of our sites.

Moderation

There were also requests for improved moderation support, which often relies on Wikipedia tools that are not optimized for Commons. Specific requests included:

• Adopting and integrating the small javascript-based deletion request tool.
• Making VisualFileChange more stable.
• The ability to make large maintenance/rollbacks for structured data.

There was also some discussion about whether AI models could be useful.

Reuse

• A couple of participants explained that being able to measure how many times videos are played on Commons and Wikipedia is essential to incentivize continued partnerships with public broadcasters.
• It was suggested that we should improve the presentation Commons to foster creative reuse and public engagement. Examples included an audio field guide and Wikidocumentaries.