Change Details

T389897 introduces a tag when Tone Check is presented during an edit session that results in a //published// edit. This task involve the work of logging when ≥1 Tone Checks are shown within edit sessions that do **NOT** result in a published edit. This task emerged in response to the following risks that volunteers [i] and staff [ii] independently raised: > Tone Check might nudge people away from more obvious peacock words (e.g. "visionary") and towards subtler forms of biased writing that is more difficult for the model and people to detect. === Story === Open question(s) - [ ] 1. How might we log / expose data of this sort in a way that aligns with existing project and Foundation security and privacy protections/policies ? -- //Note: AbuseFilter offers similar functionality to people with specific rights/permissions: https://www.mediawiki.org/wiki/Extension:AbuseFilter#User_rights // ---- i. Volunteers at en.wiki raised this in off-wiki discussions (Discord); thank you for making us aware of these discussions, @Soda ii. This risk surfaced during the pre-mortem the Editing and Machine Learning Teams held (T325851)

T389897 introduces a tag when Tone Check is presented during an edit session that results in a //published// edit. This task involve the work of logging when ≥1 Tone Checks are shown within edit sessions that do **NOT** result in a published edit. This task emerged in response to the following risks that volunteers [i] and staff [ii] independently raised: > Tone Check might nudge people away from more obvious peacock words (e.g. "visionary") and towards subtler forms of biased writing that is more difficult for the model and people to detect. === Story === Open question(s) - [ ] 1. How might we log / expose data of this sort in a way that aligns with existing project and Foundation security and privacy protections/policies ? -- //Note: AbuseFilter offers similar functionality to people with specific rights/permissions: https://www.mediawiki.org/wiki/Extension:AbuseFilter#User_rights // - [ ] What – if any – information about the account making this edit will be included in the scope of this logging? ---- i. Volunteers at en.wiki raised this in off-wiki discussions (Discord); thank you for making us aware of these discussions, @Soda ii. This risk surfaced during the pre-mortem the Editing and Machine Learning Teams held (T325851)

T389897 introduces a tag when Tone Check is presented during an edit session that results in a //published// edit. This task involve the work of logging when ≥1 Tone Checks are shown within edit sessions that do **NOT** result in a published edit. This task emerged in response to the following risks that volunteers [i] and staff [ii] independently raised: > Tone Check might nudge people away from more obvious peacock words (e.g. "visionary") and towards subtler forms of biased writing that is more difficult for the model and people to detect. === Story === Open question(s) - [ ] 1. How might we log / expose data of this sort in a way that aligns with existing project and Foundation security and privacy protections/policies ? -- //Note: AbuseFilter offers similar functionality to people with specific rights/permissions: https://www.mediawiki.org/wiki/Extension:AbuseFilter#User_rights // - [ ] What – if any – information about the account making this edit will be included in the scope of this logging? ---- i. Volunteers at en.wiki raised this in off-wiki discussions (Discord); thank you for making us aware of these discussions, @Soda ii. This risk surfaced during the pre-mortem the Editing and Machine Learning Teams held (T325851)