Change Details

In upcoming OS updates (iOS 15, iPad and Mac OS likely to be released in mid-september 2021) Apple has created a new service for users of iCloud to anonymize their internet requests. Although not a VPN service, the impact is similar, in that users IPs will no longer be transmitted with requests, instead an IP from Apple's service will be used. Initially this feature will be in "Beta", even when the rest of the OS is released, but it will be available on all apple devices, and will impact both desktop and mobile traffic. Given that IPs are a primary signal and only way to prevent repeated account creation and sock-puppetry, this means many Wikis will likely need to or will at least consider blocking all edits from Apple's IPs, as we do with VPN. English Wikipedia has already discussed this [[ https://en.wikipedia.org/wiki/Wikipedia:Administrators%27_noticeboard/Archive334#Upcoming_Apple%27s_iCloud_Private_Relay_%28sort-of_VPN%29 | here ]]. Because this is a new feature for Apple it is not clear how big the uptake and impact here will be. The goal of this task is to estimate how severe these impacts might be to determine what if any mitigation strategies or conversations are needed. Below are questions we intend to answer, all in service of understanding the potential loss in editors and edits from a wholesale block of these users, and to come up with ways to allow those users to easily continue contributing. * **Analytics questions** ** **Editors**: Of the unique non-bot editors who have published at least one successful edit in the past 90 days, what percentages of them have published at least one with Safari on mobile? With Safari on desktop? We want to know these two percentages grouped by these attributes: *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Country (e.g. United States, Indonesia, etc.) *** User tenure bucket (e.g. Under 1 day, 1 to 7 days, etc.) ** **Edits**: Of the successful non-bot edits that have been made in the past 90 days, what percentage of them were made with Safari on mobile? With Safari on desktop? We want to know these two percentages grouped by these attributes: *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Country (e.g. United States, Indonesia, etc.) *** User tenure bucket (e.g. Under 1 day, 1 to 7 days, etc.) ** **Impact so far**: How many non-bot edits have been blocked for belonging to one of the IP ranges that are being used for Relay? We want to know these counts grouped by these attributes: *** Date *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Platform (e.g. desktop, mobile, iOS app, Android app) *** Operating system (e.g. iOS 14, iOS 15, etc) *** Whether the edit was from a logged-in account or not. ** **Expected timeline**: In the past, how much time elapses between a new version of iOS being released and the majority of our iOS traffic originating from devices using that version? * **Market info** ** [Roughly from public info] What proportion of Apple editors are iCloud users? ** [Roughly from expertise/past experience/public info] what proportion of iCloud users are likely to use this feature? * **User experience**: Given our issues with unclear, undisplayed or unusable block messages, what user experience will these editors have when they try to edit after installing the OS update? * **Community**: Is the IP exemption policy and admins dealing with IP exemptions aware of this change and able to handle additional requests? What would their general stance to these requests be? === References - [Superset: Edit Attempts by Browser and OS](https://superset.wikimedia.org/r/736)

In upcoming OS updates (iOS 15, iPad and Mac OS likely to be released in mid-september 2021) Apple has created a new service for users of iCloud to anonymize their internet requests. Although not a VPN service, the impact is similar, in that users IPs will no longer be transmitted with requests, instead an IP from Apple's service will be used. Initially this feature will be in "Beta", even when the rest of the OS is released, but it will be available on all apple devices, and will impact both desktop and mobile traffic. Given that IPs are a primary signal and only way to prevent repeated account creation and sock-puppetry, this means many Wikis will likely need to or will at least consider blocking all edits from Apple's IPs, as we do with VPN. English Wikipedia has already discussed this [[ https://en.wikipedia.org/wiki/Wikipedia:Administrators%27_noticeboard/Archive334#Upcoming_Apple%27s_iCloud_Private_Relay_%28sort-of_VPN%29 | here ]]. Because this is a new feature for Apple it is not clear how big the uptake and impact here will be. The goal of this task is to estimate how severe these impacts might be to determine what if any mitigation strategies or conversations are needed. Below are questions we intend to answer, all in service of understanding the potential loss in editors and edits from a wholesale block of these users, and to come up with ways to allow those users to easily continue contributing. * **Analytics questions** ** **Editors**: Of the unique non-bot editors who have published at least one successful edit in the past 90 days, what percentages of them have published at least one with Safari on mobile? With Safari on desktop? We want to know these two percentages grouped by these attributes: *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Country (e.g. United States, Indonesia, etc.) *** User tenure bucket (e.g. Under 1 day, 1 to 7 days, etc.) ** **Edits**: Of the successful non-bot edits that have been made in the past 90 days, what percentage of them were made with Safari on mobile? With Safari on desktop? We want to know these two percentages grouped by these attributes: *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Country (e.g. United States, Indonesia, etc.) *** User tenure bucket (e.g. Under 1 day, 1 to 7 days, etc.) ** **Impact so far**: How many non-bot edits have been blocked for belonging to one of the IP ranges that are being used for Relay? We want to know these counts grouped by these attributes: *** Date *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Platform (e.g. desktop, mobile, iOS app, Android app) *** Operating system (e.g. iOS 14, iOS 15, etc) *** Whether the edit was from a logged-in account or not. ** **Expected timeline**: In the past, how much time elapses between a new version of iOS being released and the majority of our iOS traffic originating from devices using that version? - Major releases: For example when iOS 14 was released on Sept 16, 2020, how long until the unique editors on iOS (defined above) were more than 50% iOS 14? How long until 80%? - Point updates: We expect that any change in opt-in status would occur with a sub version update, the first of which generally occur 4-6 weeks after the initial release. For example when iOS 14.1 was released on Oct 20, 2020, how long until the unique editors on iOS (defined above) were more than 50% iOS 14.1? How long until 80%? * **Market info** ** [Roughly from public info] What proportion of Apple editors are iCloud users? ** [Roughly from expertise/past experience/public info] what proportion of iCloud users are likely to use this feature? * **User experience**: Given our issues with unclear, undisplayed or unusable block messages, what user experience will these editors have when they try to edit after installing the OS update? * **Community**: Is the IP exemption policy and admins dealing with IP exemptions aware of this change and able to handle additional requests? What would their general stance to these requests be? === References - [Superset: Edit Attempts by Browser and OS](https://superset.wikimedia.org/r/736)

In upcoming OS updates (iOS 15, iPad and Mac OS likely to be released in mid-september 2021) Apple has created a new service for users of iCloud to anonymize their internet requests. Although not a VPN service, the impact is similar, in that users IPs will no longer be transmitted with requests, instead an IP from Apple's service will be used. Initially this feature will be in "Beta", even when the rest of the OS is released, but it will be available on all apple devices, and will impact both desktop and mobile traffic. Given that IPs are a primary signal and only way to prevent repeated account creation and sock-puppetry, this means many Wikis will likely need to or will at least consider blocking all edits from Apple's IPs, as we do with VPN. English Wikipedia has already discussed this [[ https://en.wikipedia.org/wiki/Wikipedia:Administrators%27_noticeboard/Archive334#Upcoming_Apple%27s_iCloud_Private_Relay_%28sort-of_VPN%29 | here ]]. Because this is a new feature for Apple it is not clear how big the uptake and impact here will be. The goal of this task is to estimate how severe these impacts might be to determine what if any mitigation strategies or conversations are needed. Below are questions we intend to answer, all in service of understanding the potential loss in editors and edits from a wholesale block of these users, and to come up with ways to allow those users to easily continue contributing. * **Analytics questions** ** **Editors**: Of the unique non-bot editors who have published at least one successful edit in the past 90 days, what percentages of them have published at least one with Safari on mobile? With Safari on desktop? We want to know these two percentages grouped by these attributes: *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Country (e.g. United States, Indonesia, etc.) *** User tenure bucket (e.g. Under 1 day, 1 to 7 days, etc.) ** **Edits**: Of the successful non-bot edits that have been made in the past 90 days, what percentage of them were made with Safari on mobile? With Safari on desktop? We want to know these two percentages grouped by these attributes: *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Country (e.g. United States, Indonesia, etc.) *** User tenure bucket (e.g. Under 1 day, 1 to 7 days, etc.) ** **Impact so far**: How many non-bot edits have been blocked for belonging to one of the IP ranges that are being used for Relay? We want to know these counts grouped by these attributes: *** Date *** Wiki project (e.g. English Wikipedia, Indonesian Wikisource, Commons, etc.) *** Platform (e.g. desktop, mobile, iOS app, Android app) *** Operating system (e.g. iOS 14, iOS 15, etc) *** Whether the edit was from a logged-in account or not. ** **Expected timeline**: In the past, how much time elapses between a new version of iOS being released and the majority of our iOS traffic originating from devices using that version? - Major releases: For example when iOS 14 was released on Sept 16, 2020, how long until the unique editors on iOS (defined above) were more than 50% iOS 14? How long until 80%? - Point updates: We expect that any change in opt-in status would occur with a sub version update, the first of which generally occur 4-6 weeks after the initial release. For example when iOS 14.1 was released on Oct 20, 2020, how long until the unique editors on iOS (defined above) were more than 50% iOS 14.1? How long until 80%? * **Market info** ** [Roughly from public info] What proportion of Apple editors are iCloud users? ** [Roughly from expertise/past experience/public info] what proportion of iCloud users are likely to use this feature? * **User experience**: Given our issues with unclear, undisplayed or unusable block messages, what user experience will these editors have when they try to edit after installing the OS update? * **Community**: Is the IP exemption policy and admins dealing with IP exemptions aware of this change and able to handle additional requests? What would their general stance to these requests be? === References - [Superset: Edit Attempts by Browser and OS](https://superset.wikimedia.org/r/736)