Flagged protection background
An extended look at how we got to flagged protection and patrolled revisions
The current "flagged protection and patrolled revisions" trial has a long history under various names. The concept originated in the supposition that users should be able to rate particular article revisions, and was thus closely linked with the rating system now found on article discussion pages at its early inception. This was quickly followed by the idea that highly rated revisions should be prioritized for the viewer in some way. The variant before the recent approved proposal was referred to as "flagged revisions", while "stable versions", "article validation", "page validation", "article rating" and "Wikipedia 1.0" are all terms used for the idea of vetting particular article revisions.
According to a March 2005 article in this newspaper, User:Magnus Manske announced MediaWiki 1.5 would likely have "the first implementation of a feature for use in rating or validating articles", with one editor responding, "Wikipedia 1.0 is starting to look less and less like vaporware." However, The Signpost writer notes,
In its preliminary stages, however, such a feature is likely to remain somewhat in the background. Besides the issue of whether and how it works in a purely technical sense, a rating system would benefit from testing in social terms, by seeing how Wikipedia contributors interact with it. As a result, the initial use of an article rating feature will probably be purely for gathering data, and evaluating whether the end product actually makes sense.
A May 2005 article notes software requests for a "page-ranking feature, where users can indicate how accurate and complete they have found an article. Magnus Manske has developed a page validation feature, and developers hope to incorporate it into a putative MediaWiki 1.6 release later this year." However, the feature remained problematic. In August 2005, User:Jimbo Wales was forced to issue a clarification after a garbled translation in a discussion about Wikipedia:Pushing to 1.0 in German language media "in which he was quoted as saying that Wikipedia planned to 'freeze' the content of pages once they reached a state of 'undisputed' quality". This newspaper noted that the
article rating feature designed by Magnus Manske is still on hold. David Gerard, who has been active in pushing for it, reports that implementing the current code "would bring the wiki to a screeching overloaded halt". As a result, further improvement is needed before the feature could even be considered ready for a trial run.
In December 2005, this newspaper reported, "Article validation is set to be enabled on Wikipedia 'very soon', according to David Gerard. Currently, Brion is 'cleaning it up' for usage." Later that month, a poll approved semi-protection with Jimbo once again attempting to calm fevered media coverage in a post stating, "This is one of many changes to the software which are coming soon, including the ability to put pages into a 'validated' state (better name should be determined) and so on." [frb 1] The Signpost reported semi-protection as "part of the broader effort to add quality-related features to Wikipedia, such as article ratings and stable versions of articles". Looking back over the year, this newspaper summed up these developments as "new measures were taken toward article validation, although no such feature has been implemented yet"
In a February 2006 Signpost interview of Jimbo, he asserted,
We should be tightly focused on the quality of our coverage and content. The goal of Wikipedia is to create and distribute a freely licensed high quality encyclopedia. The path to that goal will require us to be flexible and thoughtful. The first steps will come soon with the article review system, which will initially be used simply to gather data. After we have data, we can begin to work on how we will focus our attention to improve quality.
In July 2006 and in the continued lack of dedicated software features, Gmaxwell proposed Wikipedia:Stable versions now. The proposal involved moving articles into a development subpage until a version that "must contain no obvious factual, grammatical, or typographical errors and must contain at least some level of referencing" is generated. At that point the development version would have been moved to the actual article title and protected. Despite some enthusiastic supporters, the extra responsibility this proposal put upon administrators proved to be a sticking point and the proposal failed to gain consensus support.
At Wikimania in August 2006, Brion Vibber gave a presentation announcing stable versioning likely "in one to three months". This newspaper reported,
Vibber suggested different ways that stable versioning could be used, including a short delay before changes appear, or explicit approval, in order to reduce the impact of vandalism. Experiments will be conducted on the German Wikipedia, from which a formal stable versioning policy may be adopted. Possible uses for stable versioning include the marking of stable versions for semi-permanent display, marking reviewed versions to aid in publishing Wikipedia articles, and allowing editing on protected articles while publicly showing the stable or reviewed version.
Board of Trustees member Erik Möller (User:Eloquence), in response to an October 2006 interview question on how to get the scientific community involved with Wikipedia stated,
I want us to have a focused discussion to define very precisely what functionality we will implement to identify "stable" versions of articles. Solving the well-known problem that we don't distinguish between an unreviewed edit from 5 minutes ago and an article that has grown over 5 years should be pretty high on our agenda to gain credibility with academia, I think.
That same month, a German media outlet first reported on the planned adoption of a flagging system by German Wikipedia, though Wales cautioned that the details of the implementation had not yet been worked out. User:Michael Snow wrote a December 2006 article examining the ramifications of the Siegenthaler incident and noting,
One step taken in the aftermath, on December 5, was removing the ability of unregistered editors to create new encyclopedia articles, such as had been done in Seigenthaler's case. This was characterized as an experiment but has remained in place since that time, although it is unclear how successful this was. The issue was recently raised again on the English Wikipedia mailing list, and Wales stated his opinion that the experiment "did not achieve the intended effect". He suggested that the restriction should be changed when the planned feature to flag "stable" or "non-vandalized" versions of articles is available.
This feature is supposed to be tested initially on the German Wikipedia, and was discussed at Wikimania, but is not yet ready for implementation. What to call flagged revisions remains a matter of debate, but the ability to flag an article version is expected to be widely distributed. One possibility is after a small number of edits or a brief waiting period such as the time needed to edit semi-protected articles.
Little was heard about the stable version features for several months though this newspaper noted in April 2007: "A huge amount of work was done on Flagged Revisions and ConfirmEdit ... although neither are enabled on the English Wikipedia." mw:Extension:FlaggedRevs had been started in late March 2007 though it took some time for the term "Flagged Revisions" to overtake in usage the previously used terms.
Jimbo Wales gave another interview in September 2007, including on several related topics:
WS: Gwern asks, "When anonymous page creation was forbidden, it was said that there would be a study of its effects. That was a very long time ago. Was any type of formal study ever done? If it was, when will it be released? Have you considered allowing anonymous page creation again?"
JW: I am unaware of any formal study. My own feeling is that anonymous page creation should be re-enabled when the stable versions feature is available. I do not consider the experiment to be a success or a failure. It seems to have had very little impact on anything, all things considered.
JW: What I support with stable versions is that whatever it is, it is the default view for anonymous visitors, but that it should be used sparingly, not on all articles; that it needs to be a "state" that an article is in, like semi-protection, but more open.
WS: Ta bu shi da yu asks, "A number of attempts at setting up a way of marking stable versions have been proposed, and at least one MediaWiki extension has been created. None of these proposals have taken off, and it looks like the extension has stagnated. What's your opinion on stable versioning? How do you personally think it should be used – as a default when available for anonymous users, or as an optional view?"
As a sidenote to coverage about a proposed lifting of the restriction against anonymous page creation, this newspaper noted in October 2007, "Some people have looked to the introduction of a MediaWiki feature for stable versions or flagged revisions as a step that would allow article creation to be reopened. However, ... progress in developing that feature remains unclear (a planned test on the German Wikipedia has yet to begin). The occasion did prompt some renewed debate about which version, stable or live, should be displayed as the default."
Not until May 2008 was there significant progress, when German Wikipedia introduced "sighted revisions", a feature made possible through the extension FlaggedRevs in which "any page with a 'sighted revision' shows the sighted revision by default to non-logged-in users; logged-in users, by default, view the most recent revision, whether it has been sighted or not." The Signpost noted,
The German Wikipedia is acting as the testing ground for this feature, as that community has been very vocal in their support of such a feature. It is expected that their experiences will affect how the English Wikipedia uses the feature. Testing there is expected to last for several weeks at the least before it is enabled here.
The FlaggedRevs extension is customizable, and the Wikipedia community considered proposals that fell under Wikipedia:Flagged revisions, namely sighted versions and quality versions, the latter of which would take precedence over sighted version flagging. This newspaper reported, "At least on the English Wikipedia, the entire process is proposed, with details still to be hammered out on many important issues. However, with the German community embracing flagged revisions, it is likely that the feature will be enabled on the English Wikipedia within the next few months."
The English Wikipedia community continued to discuss possible implementation of some kind of configuration of FlaggedRevs, while German Wikipedians issued a report of their experiences in December 2008.
In a poll that began on 2 January 2009, proponents of flagged revisions proposed a variety of different trials they wanted permission to carry out. The poll ended three weeks later with 59.7% in support, with this newspaper reporting, "Many opposed the extension in principle, as a contrary to the 'anyone can edit' spirit of Wikipedia. Another significant portion of those opposed were concerned about the particular proposed implementation, and editors are still working to forge a compromise on the details of the trial to satisfy that group." The dispute grew emotional, with several editors threatening to leave the project over the issue. Specific concerns of editors who opposed include that the proposal was not for a specific trial, was vaguely worded, or would appear to create a large backlog of "unsighted" revisions. The three-fifths split was important as Erik Möller, who was now the Deputy Director of the Wikimedia Foundation, had previously stated that "a very large majority, at least two thirds, is generally necessary".
As The Signpost reported, things grew more dramatic when, at the unsuccessful close of the poll,
Jimbo Wales asked the Wikimedia Foundation to turn on Flagged Revisions on the English Wikipedia on his "personal recommendation"... Wales' interpretation of the poll is limited to a technical request to the developers to turn the extension on, and place control of the extension in the hands of the project's bureaucrats. Thus, the proposal does not detail what pages will have flagged revisions enabled, just a code change in the site's configuration file. Some proposed trials are detailed here, but were not a part of the poll. As a result, it remains to be decided how the community will proceed if Wales' request is granted. Brion Vibber, the Wikimedia Foundation's Chief Technical Officer, stated on Jimbo's talk page that Flagged Revisions would not be enabled by the foundation "before working out some very specific parameters for the test first".
This prompted a request for arbitration to decide if Wales had the authority to close a poll in his own favor, Wales having expressed strong support for Flagged Revisions on several occasions, that was unanimously declined by the Arbitration Committee.
There was further coverage of German Wikipedia, which was reported to have done an initial "sighting" of 97% of its articles. In mid-February 2009, German Wikipedia announced that it had completed sighting every article.
In early March 2009, this newspaper wrote, "Although discussions about potential trials of the flagged revisions extension have essentially ground to a halt, Jimbo Wales has indicated on his talk page that he is "shopping a very premature proposal around", with "[n]ews to come soon".
In mid-March, a two-week poll was started on a new proposal: Wikipedia:Flagged protection and patrolled revisions. Unlike the January proposal, which was criticized for covering a variety of different trials and being overly vague and generally applied, this proposed a two-month trial of a specific configuration of mw:Extension:FlaggedRevs:
On the issue of biographies of living people, discussions have demonstrated the need to improve monitoring of these articles, and that flagging systems could help us to do so. But there is no consensus to use an active implementation (in which new edits are not shown to readers unless made by or flagged by trusted users) for all biographies of living people or an arbitrary subset of them, preemptively. Thus, it is proposed to enable patrolled revisions, which uses a passive flag that reviewers can use to mark a revision patrolled, for monitoring purposes, but that has no effect on the version viewed by readers. This passive flag is available for all articles. Flagged protection is a proposal to allow administrators to enable an active flag on a given article, 'flag protecting' it. Reviewers can flag revisions, and the version viewed by readers by default on (semi) flagged protected pages is the latest confirmed revision. During the trial, semi flagged protection is intended to be used with the same requirements as for semi-protection, and full flagged protection (see below), with the same requirements as for full-protection.
This restriction was well understood by editors in the poll that ended on 1 April, whose comments in support included:
- Not nearly as good as semi-protection of all BLPs and liberal use of full protection but a small step in the right direction;
- More protection is needed, particularly for BLPs;
- Qualified support.... Two months with the flagging as a protection-style option (which is honestly how I think it should be used) will give us some (but not all) of the data we need to work with; and
- I don't see how replacing article protection with flagged article protection should cause idealogical [sic] controversy.
The Signpost noted, "In addition to those who oppose any form of flagged revisions, the proposal has been opposed by some who argue that it is essentially toothless with regard to biographies of living people. It is yet to be seen whether Jimmy Wales will present his own compromise proposal to apply some form of flagged revisions to such biographies, as he indicated he would before the current proposal gained momentum." The poll ended with 259 in support, 61 in opposition and 4 declared neutral, for a 79.9% support percentage.
In May, Vibber gave an update on the technical aspects of flagged protection and patrolled revisions, stating "Yes, we do plan to roll out an English Wikipedia test setup for Flagged Revs." and "There's not yet a fixed schedule for it, but I'd like to see it up and running in production before Wikimania. :) [August]"
The next update came on the Wikimedia Tech blog on 23 August 2009, which stated, "A few highlights from the last week… Test wikis with Flagged Revisions and ReaderFeedback configurations have been set up to shake down UI and workflow before we prepare to deploy these extensions on English Wikipedia in the coming weeks. The test sites have been populated with featured articles, and should be getting some decent front pages soon. ;)" This post apparently was the catalyst for the subsequent media firestorm. (See full Signpost coverage.)