Editors want to be able to patrol edits on pages that they care about. Currently they use: 1) Watchlist 2) Article History 3) Recent Changes The mechanism for "fixing" a problem with an edit is doing a revert. Reverting allows you to go back to a revision of the article that existed before the change that you want to "fix". Specifically it populates an edit with the content of that previous revision and then you are actually able to make any additional changes on top of that old content and then save. There are two special cases of reverting that are especially useful to users: 1) Undo - this is when you want to "fix" an edit but there have been edits since the problematic edit that were productive. Undo tries to just undo that specific edit in question instead of reverting all the way to the revision before the edit. The reason to use undo is that sometimes there was a problematic edit but since then there have been productive edits. Specifically, what undo does is it tries to revert to the revision before the problematic edit and then computationally add back in the edits since then. Sometimes this isn't possible. Sometimes it is. When it is possible to undo automatically the user gets the revision plus the "productive edits" that occurred since the edit being undone, all of that content is populated into an edit interface and the user can make any additional edits (sometimes necessary to make the article make sense after the undo) and then save. 2) Rollback - this is when you take all of the edits of the last user and revert to the revision before those edits. The purpose of this is when there is a user that has been committing vandalism you can quickly rollback those edits. This is a one step process because it just does the revert and saves automatically. note 1: generally speaking vandalism gets caught quickly and is often the most recent or most recent set of edie by a single user i.e. the situation that rollback is designed for note 2: undo occurs on an edit, revert and rollback operate on revisions. on desktop the list of edits and the list of revisions is the same interface but it may make sense to divide these on mobile. Thus on mobile we may have separately a list of revisions (possibly grouped by user) that may allow reverts and rollbacks, and a list of edits that allows for unto. We will likely prioritize revert/rollbacks because that covers the biggest use case (vandalism on articles that are changing at a moderate velocity. Also, this may have implications for how we display watchlist items: considering grouping edits by user, and only displaying most recent edits (i.e. only rollback eligible edits) There a detailed view of revisions in addition to a list view of revisions and. We need to understand what goes into a detailed view of a revision. Maybe we show the diff to current version because this is what would be affected by a revert, actions, username, time, other details.We may want to consider changing the interaction of reverts a bit (maybe should be 2 click action instead of putting user into edit). Let me know if I missed anything. -- Kenan Wang Product Manager, Mobile Wikimedia Foundation

There are two special cases of reverting that are especially useful to users: 1) Undo - this is when you want to "fix" an edit but there have been edits since the problematic edit that were productive.

that specific edit in question instead of reverting all the way to the revision before the edit. The reason to use undo is that sometimes there was a problematic edit but since then there have been productive edits. Specifically, what undo does is it tries to revert to the revision before the problematic edit and then computationally add back in the edits since then. Sometimes this isn't possible. Sometimes it is. When it is possible to undo automatically the user gets the revision plus the "productive edits" that occurred since the edit being undone, all of that content is populated into an edit interface and the user can make any additional edits (sometimes necessary to make the article make sense after the undo) and then save. 2) Rollback - this is when you take all of the edits of the last user and revert to the revision before those edits. The purpose of this is when there is a user that has been committing vandalism you can quickly rollback those edits. This is a one step process because it just does the revert and saves automatically. note 1: generally speaking vandalism gets caught quickly and is often the most recent or most recent set of edie by a single user i.e. the situation that rollback is designed for

note 2: undo occurs on an edit, revert and rollback operate on revisions. on desktop the list of edits and the list of revisions is the same interface but it may make sense to divide these on mobile.

separately a list of revisions (possibly grouped by user) that may allow reverts and rollbacks, and a list of edits that allows for unto. We will likely prioritize revert/rollbacks because that covers the biggest use case (vandalism on articles that are changing at a moderate velocity. Also, this may have implications for how we display watchlist items: considering grouping edits by user, and only displaying most recent edits (i.e. only rollback eligible edits) There a detailed view of revisions in addition to a list view of revisions and. We need to understand what goes into a detailed view of a revision. Maybe we show the diff to current version because this is what would be affected by a revert, actions, username, time, other details.We may want to consider changing the interaction of reverts a bit (maybe should be 2 click action instead of putting user into edit). Let me know if I missed anything.

From a 10,000 foot perspective, here's what happens on desktop Wikipedia today:

Also, this may have implications for how we display watchlist items: considering grouping edits by user, and only displaying most recent edits (i.e. only rollback eligible edits)

> >note 2: undo occurs on an edit, revert and rollback operate on

> >desktop the list of edits and the list of revisions is the same

I'm confused -- edits and revisions are synonymous. In both cases, we're talking about atomic changes to a single document that are stored sequentially as a version history. Am I missing something?