User:Cenarium/On FlaggedRevs

From the various and numerous community discussions on FlaggedRevs, I think two themes are dominant:
 * The need to improve our monitoring of blps, using FlaggedRevs-like systems.
 * The need to use FlaggedRevs-like systems to protect certain articles, as an alternative to regular protection.

But there is a strong opposition to use strict flagged revisions for a large number of articles. By strict, I mean similar to what is done on the German Wikipedia, where for all articles, edits by IPs or users without special rights are not showed to the public by default until flagged by a user with appropriate rights. But FlaggedRevs is a very versatile extension and can be adapted in a number of ways,

For example, it's possible to use passive flags, they are only check points, and don't affect the version viewed by readers, it's always the latest one. They still allow to monitor pages, even though they are not acting preemptively on edits.

However, sometimes, an article will require a form of protection that cannot afford a passive flag. Using flags affecting the version viewed by readers is a form of protection, as it protects readers from unwanted content (and the blp subject for that matter). We do have a protection mechanism, acting by limiting the group of users who are able to edit the article, but a FlaggedRevs-based protection mechanism would be a very interesting addition.

Thus, my proposal is to implement a passive flag for all articles, in order to monitor blps, and an active flag that can be enabled for certain articles, with a use defined by the protection policy.

The BLP issue
Many users would like to turn on strict flagged revisions on all blps, but it's also opposed by many, for the traditional reasons classic FlaggedRevs is opposed. We do not have consensus to do this at this time, nor even for a small number of arbitrary blps, and we'll likely never have it any time soon. Now, it doesn't mean that all is lost for blps. A passive flag, would allow a monitoring of blps, especially little-trafficked ones, and wouldn't receive the opposition an active or strict flag would get, so that may be worthwhile to try that. It won't prevent an edit from being visible to readers, but it will allow to bring it to the attention of reviewers, and they'll act on it.

Special:Unpatrolledpages lists all pages that have never been patrolled, and this can be filtered by categories, so by Category:Living people... it's an unmatched way to detect and take care of problematic blps. Now when a blp has been patrolled at least one and the latest revision is not, it'll end up in Special:OldPatrolledPages that can also be filtered by categories, so changes will also be able to be monitored.

Even if purely from a blp enforcement point of view, it's not as efficient as a strict FlaggedRevs implementation, it's a huge improvement compared to the status quo, and this has a chance to be adopted !

As for blps that have repeatedly suffered from vandalism or blp violations, we can use flagged protection for them (an 'active' flag). Precisely when we should use it shall be determined by the protection policy, the requirements would be presumably lower than for normal semi protection, consensus will say.

Flagged protection
The original flagged protection proposal was to use FlaggedRevs on an article when it matched the criteria for protection. Any autoconfirmed user would be a reviewer, that is, their edits are automatically reviewed when the previous revision is already reviewed, and they can review ('sight') revisions. This has been opposed for the following reasons:
 * too lenient for certain articles, especially blps
 * granting reviewer rights to autoconfirmed, but still new users would make editing more difficult for them (need to analyze diffs, etc) and would be a responsibility they may not want
 * in cases of mistakes of inexperienced users while reviewing, they may be blocked or have their rights removed
 * it wouldn't be efficient in case of sockpuppetry

Thus I modified the proposal this way: edits by autoconfirmed users are autoreviewed, but this can be disabled on a per-page basis, and they do not have reviewer rights. As the case where an autoconfirmed user edits a flagged protected page with an unreviewed latest revision, this won't change much in practice from the original version. Cases where autoreview should be disabled for non-reviewers would be defined by the protection policy, this would be essentially in cases of persistent sockpuppetry or major blp violations. An auto promotion to reviewer status would also be implemented, but the requirements would be higher than autoconfirmed, the right could also be added and removed manually by administrators.