User:Jéské Couriano/A brief history of AfC

A common argument used on pages where article creation or deletion are brought up is, "Well, then, why was <foo> approved?!"

I find that usually, the answer is, "It wasn't."

The reason these are brought up as examples is in part because the user mistakenly assumes every article on Wikipedia had to go through the same process as their draft currently is, mainly because these users haven't the time or inclination to seek out the various requests for comment on the topic. This essay thus seeks to summarise the history of the drafting process on Wikipedia.

Early history

From the inception of the project until about 2011, there was no formal drafting process. Userspace sandboxes being used to draft articles were very much a thing, but there was no actual formal review process. Wikipedia operated more on a "Publish and be damned" system, especially after the introduction of WP:Biographies of living persons in 2005 and WP:Requests for arbitration/Editing of Biographies of Living Persons in 2008. During these days, it was assumed that if you created an article, than you accepted whatever consequences arose from that action, whether it be the article's deletion, being relentlessly edited, or even your account being blocked for egregious or persistent issues. In addition, the only restriction on creating a page was having a registered account.

At the time, the only sort of "review crew" that existed was the New Page Patrol (NPP), a faction that still exists today on Wikipedia and serves the same function now as it did then - patrolling articlespace for new articles and making sure they're up to standards. The usual result of an article not meeting their standards prior to 2011 was deletion of some stripe, either summary or debated, with the figures for articles being deleted as a result of NPP intervention being placed at approximately 80%. NPP itself had a few dedicated users, but the rate of article creation even back in those days was high, resulting in the new pages feed becoming backlogged consistently once the more prolific patrollers were offline for the day. NPP was understaffed for its purpose, and was struggling to keep up with a firehose of new articles.

While WP:Articles for creation technically existed during this time (the page was created in 2006), it behaved more like WP:Requested articles as opposed to a formalised drafting process, and wouldn't start resembling its present form until mid-2010.

Autoconfirmed creation proposal

On 10 March, 2011, The Blade of the Northern Lights started a thread on WP:Village pump (miscellaneous), laying out the situation and proposing two solutions: Getting more people to assist with new pages patrol and/or requiring autoconfirmed status in order to create articles directly into mainspace. An informal discussion showed overwhelming support for the latter option, which in turn resulted in a formal request for comment on the topic opening in early April of that year. After the standard month for requests for comment, the closer noted that there was a consensus to trial this change, with length of both the trial and the evaluation period to be determined later.

The subsequent RfC determined the trial should last six months, with a one month evaluation period before determining if it should permanently be implemented. However, during the duration RfC some issues were noted. The Article Wizard was seen as not fit for purpose and would need an upgrade. The Pending changes "trial" was, at the time the duration RfC started, still being litigated (and would be terminated during same). Page creation was all-or-nothing in MediaWiki's codebase and did not allow for specific namespaces to be set up to disallow article creation while allowing others.

While the first two could (and ultimately would) be largely dealt with on Wikipedia proper, the third one required an edit to the underlying MediaWiki software itself, something that the Wikimedia Foundation at the time baulked at, citing that this would essentially require a major overhaul of the code.

Wizard 2.0 and AfC

Meanwhile, other technical developments were going on around the actual article creation process itself. The Article Wizard would receive a major overhaul and re-coding, merging it with a similar wizard that created Articles for Creation entries to create Article Wizard 2.0, which would be implemented in 2010.. Articles for Creation itself shifted from a pipeline for unregistered users to make suggestions to a means to allow them to create their own drafts, initially done as talk page subpages (as IPs would be able to create those without any issue).

In 2013, a new RfC was created proposing that AfC be made a requirement for new users. While most of the RfC's proposals would ultimately be rejected at that time, there were signs that a new namespace and/or userright would be acceptable, provided they were supported to a more in-depth RfC first. Taking this result, TheOriginalSoni started an RfC on creating a new namespace - what is today known as the Draft namespace. The proposal involved consolidating all existing drafts (with the exception of userspace drafts due to that proving to be too controversial) into this new namespace and making it the default location for working on draft articles. This proposal also received support from within the WMF, and ultimately ended with support for creating draftspace.

After the RfC, two different bugzilla (now known as Phabricator) threads were opened to implement the RfC. The first, by Steven Walling in his capacity as a WMF employee, was to propose making adjustments which would lay the groundwork for a draftspace to be created in wnatever MediaWiki project that wanted one, Wikipedia or otherwise.. The other, by MZMcBride, was strictly about creating the namespace. Both wound up facing some pushback, with Walling's falling into the same issues as the initial trial proposal (i.e. limiting by userright or registration status), but would be implemented within a month.

The Trial and Aftermath

For the next few years, things largely remained quiet as far as article creation permissions and tools went. However, in June of 2017, a face-to-face meeting between WMF employees resulted in the WMF finally working towards implementing the article creation trial, primarily as a research experiment. This led to a discussion about the implementation that lasted for a couple of weeks, after which discussion was moved to WT:ACTRIAL to thrash out the minor details. After a few brief delays, the trial began on 14 September, and would run for six months, during which time autoconfirmed status would be required to create articles directly in mainspace.

After the trial ended on-time on 14 March 2018, the data from the trial would be analysed via the lens of m:Research:Autoconfirmed article creation trial, with the WMF focusing on new-editor retention and the quality of articles that went through the drafting phase during this time. One month later, their findings were that it had no effect on editor retention and kept low-quality material out of mainspace (and thus off of Google), but that the burden shifted from NPP to AfC reviewers, who were (and at times still are) trying to keep up with a firehose of new drafts. Nonetheless, an RfC held immediately after the trial's end was overwhelmingly in favour of making the trial restrictions permanent, and in April 2018 that is precisely what happened.