Wikipedia:Autoconfirmed article creation trial

In April 2011, a proposal was put forth to require autoconfirmed status in order to create articles. (Previously, no registration had been required; since 2005, an experiment was ongoing where registration was required.) In May 2011, the proposal was closed with a consensus that a trial should be implemented to test out this new idea. A second RfC was then opened to determine duration of the interventional phase of the trial, and it was decided that the interventional phase should run for 6 months, and then the changes should be reversed for a period of one month while their effects are discussed. The Wikimedia Foundation prevented the idea from going forward in 2011, but in June 2017 agreed to conduct and help implement ACTRIAL as a research experiment. In July 2017, after a post on his talk page, Jimmy Wales issued a personal statement regarding ACTRIAL supporting the use of limited trials to gather data to make Wikipedia policy decisions.

The discussion on the duration and terms of the assessment of the interventional phase was announced at the Village Pump and at Centralized Discussion, and was officially closed by independent admin summary 18 August 2011, in favour of the proposal with "... a clear consensus for a six-month trial, followed by a one-month period of discussion to determine the trial's effects. Given the wide support and uncontroversial nature ..."

ACTRIAL will serve as a trial of the original proposal that achieved community consensus. Information will be gathered during the trial period and reported to the English Wikipedia community at its conclusion, and the community will decide if any additional steps that should be taken based on the results. The trial began on 14 September 2017 and lasted for 6 months.

The autoconfirmed article creation trial (ACTRIAL) ended on 14 March 2018. The results of the research collected can be read on Meta Wiki. The community discussed the results and their impact on the English Wikipedia at Wikipedia:Autoconfirmed article creation trial/Request for comment on permanent implementation, which closed on 18 April 2018 with support for making the changes permanent.

Basics

A trial is a study in which you change something in order to determine what the effect of that change is.

The intervention is what you change. In this case, the intervention is requiring autoconfirmed status for page creation. In this case, the trial consists of the interventional phase plus the non-interventional phase. During the non-interventional phase(s), autoconfirmed status will not be required for page creation.

The trial length is the length of the whole study: the length of the interventional phase (in this case, six months) plus the length of the non-interventional phase(s) (the time periods being used for comparison data).

Technical changes required

Requiring developer action

For autoconfirmed status implementation

Currently MediaWiki's settings would only allow us to prevent non-autoconfirmed users from creating pages in any namespace, including their own userspace. (This is the createpage userright - see mw:Manual:User rights - which in MediaWiki all user groups have by default. On en.wp's MediaWiki configuration this is currently removed from anonymous users, with the result that everyone but anonymous users can create pages.) Possibly the desired result can be better achieved another way, but the most obvious solution is:

1. A support for namespaces to user rights. That needs a modification to MediaWiki.
2. For en.wiki.x.io the createpage right would be set to false for the content namespace(s) – for the anonymous ('*') and non-autoconfirmed ('user') user groups. That needs a modification to wikipedia's MediaWiki config, once 1. is done.
3. Create MediaWiki:Noautocreatetext as a variant of MediaWiki:Nocreatetext shown to non-autoconfirmed users. See #User interface messages below.

For article creation flow implementation

Tracked in Phabricator
Task T170354

1. Update 2011 landing page to be simplified and work with current MediaWiki code.

    Done: Wikipedia:New user landing page (WP:LANDING)

For trial data

• Set up data pipeline to collect information about page creations (T150369)

Updates to Wikipedia help pages and guidelines

When the change is made, a lot of Wikipedia's help pages and guidelines will be outdated. We should make a list of the ones that will need to be changed (at both ends of the trial). WhatamIdoing (talk) 19:56, 12 August 2011 (UTC)

1. Wikipedia:Your first article—point #2, "You can create a new article once you have registered an account..."  Done
2. Wikipedia:Starting an article—in the lead, "Articles may only be created by registered users..."  Done
3. Wikipedia:User access levels#New_users—"may immediately create pages in any namespace..."  Done
4. Wikipedia:User access levels#Autoconfirmed_users—add creating pages to the list of restricted activities  Done
5. Wikipedia:FAQ#How_do_I_create_a_new_page.3F  Done
6. Post a notice on MediaWiki:Watchlist-details. Requested.  Not done (removed as requested)

Updates to Article Wizard

Wikipedia:Article wizard/Ready for submission will need to be changed to only show the second option (create article in main namespace) to autoconfirmed users and only the first option (create article in draft namespace) to non-autoconfirmed users.

 Done

Statistics

Main article: meta:Research:Autoconfirmed article creation trial

In order to measure the effect the intervention has had, various statistics should be collected before, during, and after the trial. This section is a place to request the collection of well-defined statistics which you think will be useful in determining whether the intervention affects Wikipedia. Metrics to be measured should include: new users registered and retained, quality of new articles created, new page patrol workload, deletion process workload, etc.

The Community Tech team at the Wikimedia Foundation requested in July 2017 to receive funding to hire a data analyst contractor that will be devoted to this project. The Foundation will provide an update on funding by 15 August 2017.

Questions we could ask

Effects on users

Question	Predictions
Does the number of users registering during the intervention change?	Might go down, if users know that registering does not permit instant page creation and instant page creation is the sole goal.
Does the number (proportion?) of new accounts making their first edit each day change?	25% of new accounts start by creating a new page. Therefore, this might go down by as much as 25% (especially if the number of account creations remains steady).
Does the number (proportion?) of accounts reaching autoconfirmed status change?	Needs a time limit: number reaching status within a week, a month, a quarter? This is likely to go up, if the ability to create pages is motivating.
Does the typical length of time that it takes accounts to reach autoconfirmed status change?	New users might reach autoconfirmed status sooner, if the ability to create pages is motivating.
Does the number (proportion?) of accounts making 100 edits in the first month (three months?) change?	If new users have a better initial experience, they might be more active than previously.
Do new editors work on a greater number of articles (main namespace only?) in the first month (three months?)?	To make the initial ten edits, new users might be more likely to work on a greater number of articles than previously.

Effects on articles

Question	Predictions
Does the number of articles being created on an average day change?	Does our stats process treat page creation at AFC (in the Draft namespace) the same as page creation in the main namespace or in a user subpage? Might go down, because it's harder to create articles.
Does the number (proportion?) of articles being listed and/or deleted through CAT:CSD, CAT:PROD, CAT:BLPPROD, or WP:AFD change?	Both listings and deletions might go down absolutely, especially if fewer articles are created. The proportion of tagged articles being deleted might go down, if greater experience means that higher quality articles are being created.
Does the proportion of articles surviving the first (third?) month after their creation change?	The proportion surviving might go up, if greater experience means that higher quality articles are being created.
Does the overall rate of growth of (non-deleted) articles change?	The overall rate of growth might decline. (Whether this would be a bad thing is disputed.)
Are fewer unsourced-but-undeleted BLPs created by new editors?	With more focus on AFC, we might see fewer unsourced BLPs created.
Does pent-up demand during the intervention result in a spike in article creation immediately after the end of the intervention phase?	It might, especially at the end of the intervention phase. New editors might choose between doing "all that work" to get autoconfirmed status and waiting until the intervention phase is over.

Effects on the community

Question	Predictions
Does the trial change the number and activity of (new-ish?) editors in future months?	It might decrease activity, if making article creation harder means fewer people ever edit. It might increase activity, if greater breadth of experience and fewer BITEy deletions mean fewer newbies are offended.
Does the change affect DYK?	If fewer articles are created, it might decrease the number of noms at DYK. However, since most DYKs are not newbies, it might have no effect. (Major changes proposed for DYK, if implemented, might make comparisons impossible.)
Does the number of requests made at WP:AFC increase?	If AFC is used as a significant alternative to editing other articles and waiting four days, then AFC requests might increase significantly.
Are more requests for confirmed status made at Wikipedia:Requests for permissions/Confirmed?	This might go up, if users want to bypass the four-day/ten-edit waiting period.
Does the intervention affect the length of the backlog at Special:NewPages?	Fewer pages being created by newbies might decrease the number of pages requiring review.

Results and followup

On 14 March 2018, after 6 months, the trial ended. The results of the research can be found at meta:Research:Autoconfirmed article creation trial and Wikipedia:Autoconfirmed article creation trial/Post-trial Research Report. Three major themes of the report were that:

1. New user activity and retention is largely unaffected
2. Creation of pages by new users shifted from article space to draft space
3. Fewer low-quality and inappropriate pages have been created in article space

The independent consultant's research generally showed that the concerns (hypotheses) that were raised before the trial about the restrictions having a negative impact on retention of new editors did not occur, the trial and the research conducted received generally positive feedback.

The community held an request for comment at Wikipedia:Autoconfirmed article creation trial/Request for comment on permanent implementation. The RfC was closed on 18 April 2018 with the community being in favour of requiring autoconfirmed or confirmed status, to directly create a new article. The discussion had 233 participants, and closed with 207 editors (88.8% of those participating) supporting the proposal, and 26 editors opposing.