Some notes from the session (about 10 people):

Presentation:

• Introduction and quick overview of https://wikimedia.biterg.io/ - what we have (Gerrit, a bit of Github/Phab activity/mailing lists indexed in https://gitlab.com/Bitergia/c/Wikimedia/sources/blob/master/projects.json ), some statistics on tech community growth/health in those covered areas
• What we don't have statistics on: Wikimedia Phabricator Diffusion, User scripts / gadgets, Lua modules, templates stored on-wiki, extension code stored as content on wiki pages, unknown code repo locations of some tools
• We probably want to start indexing GitLab code review once we start using that
• See https://www.mediawiki.org/wiki/Developer_Advocacy/Metrics#Technical_Contributors_Map for the complexity of our ecosystem
• https://www.mediawiki.org/wiki/Community_metrics

Discussion:

• What do you use the statistics for?
  • Annual statistics; was used to understand contributor challenges in https://www.mediawiki.org/wiki/New_Developers/Quarterly ; etc
• Is this used for eligibility of tech contributors for board elections in 2021 - https://meta.wikimedia.org/wiki/Wikimedia_Foundation_elections/2021 ?
  • Current guidelines are outdated from SVN times, https://meta.wikimedia.org/wiki/User:KCVelaga_(WMF) has been looking into updating this
• Do you have any estimation why number of contributors are stable ?
  • Numbers are only for Gerrit; limited data sources indexed, activity outside of what we index (tools), maybe Gerrit's learning curve, our docs, etc
• Is git authors a good way to estimate number of contributors?
  • Probably not, because we also import upstream Git repos from projects outside of Wikimedia to deploy their software on our servers without doing changes ourselves - we import their Git repo and the activity in there, so those contributors get indexed but they don't contribute to Wikimedia software but to software used by Wikimedia instead
• Translations?
  • Could be an area, unclear how much to consider translators technical contributors
• We should probably remove external repos from these dashboards more generally no?
  • maybe checking the git remotes for upstream or whatever name it may have to exclude them automatically
  • Would be manual work, plus some mixed cases (repo imported from upstream but some custom changes, e.g. Mailman3), so a mix of outside-Wikimedia-upstream contributions plus a few inside-Wikimedia custom changes in the same repo
• Do we have any statistics about the diversity (age, sex, nationality etc.) of technical contributors? Or maybe these information is not on developer profiles so we don't
  • maybe only as aggregate anonymous data, but definitely not linked to names and such?
  • We don't have info on age, gender, sex and don't index that. We would have info on IP addresses (location) in some systems, but we do not index that and I am not sure if there is even code to index that in the platform. Plus Privacy Policy might collide (90 days), though this is third-party hosted.
• maybe from Dev satisfaction survey?
  • Could be, but that would be outside of this system