Page MenuHomePhabricator

A proposed modification to pageviews data protection for Wikistats
Open, Needs TriagePublic

Description

Since the July 2023 update, (relevant link is here: https://phabricator.wikimedia.org/T338033) there has been a hiding of 35 countries for privacy reasons on the pageviews list, after the protection for number of contributors (which is a right decision by me).

Although this is an improvement in privacy, I think that this change is too far-fetched, in the way it was done. This decision doesn't allows the viewing of pageviews statistics for multiple countries, which, in the previous form were useful for research and analysis reasons, while there was anonymisation thanks to the fact that the statistics were rounded on the million in the map. In my opinion this causes more problems than it solves, particularly when it comes in the communities affected by this block. For example, Uzbekistan, has around to 30 million pageviews, and Wikipedia is promoted by the government through WikiStipendiya, and there aren't any risks in the Uzbek community now. There is no security risk in my opinion for the readers there given the large numbers (as we can restrict public data only in the major languages), the previous security standards on the numbers shown, which I think they're adequate, while every other more specialised data is hidden, and stats pre-May 2023 are available. It's an asymmetry, plus the disruptions I think this will cause due to regular removals and additions of countries.

I propose a partial revision of the change. For example, in Pakistan, statistics only for English, Urdu, Punjabi (Western), Sindhi and Pashto projects to be shown, or only Spanish and English in Venezuela, which means keeping statistics available only for the languages native in the blanked out countries. This means, not showing, for example pageviews of Greek Wikipedia in Pakistan, but showing Bengali in Bangladesh where it is national language. I mean, for cases like Bangladesh, Rwanda or Uzbekistan, the strictness of the protection can be revised for the most visited Wikipedias and projects, if it's feasible.

Regarding China (were people have been even arrested for using Wikipedia), North Korea (where only a very small portion has internet access), Myanmar (also blocked), Russia or Belarus (and possibly Syria, Venezuela or Saudi Arabia/Iran) the block can be kept because of security threats for Wikipedians in the aforementioned countries.

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald Transcript
Nikosgranturismogt renamed this task from Wikistats protection list to A proposed modification to pageviews data protection for Wikistats.Jul 2 2023, 4:30 PM