Jump to content

Wikipedia:Wikipedia Signpost/2018-10-28/Special report

From Wikipedia, the free encyclopedia
Special report

NPP: This could be heaven or this could be hell for new users – and for the reviewers

Kudpung joined Wikipedia in 2006. His focus is on policy changes concerning deletions/notability, and the improvement of the new page patrolling and AfC processes. He was de facto coordinator for NPP for many years before retiring from it in March 2017. He was acting Editor-in-Chief of The Signpost for several months this year. The views expressed in this article are his alone and do not reflect any official opinions of this publication.

Just what is NPP/NPR? Why does it need YOU?

Animated depiction of a red-and-white Christmas stocking
Fill me?

How many of our readers actually know what goes on behind the scenes? Thousands of content contributors and maintenance workers go about their tasks without ever being involved in contentious issues and noticeboards, and may only be fleetingly aware of what a huge mechanism is required to keep Wikipedia clean and sailing on calm waters. New Page Patrol and New Page Review are probably an enigma to many, and those who know most about it are its operators and the creators of new pages whose submissions of new articles have been subject to the process. It's a complex operation that relies on complex and sophisticated software – and a lot of skill. It needs more reviewers, but above all, improved tools for them to use which only the Wikimedia Foundation can develop.

New Page Patrol, with its associated Page Curation tools and feed, is the only firewall that exists to protect our encyclopedia from becoming a platform for spam, paid editing, attack pages, and gratuitous graffiti. It's arguably the most important backroom function of all – even before the work of admins who are responsible for the physical deletion of unwanted pages and maintaining some form of harmony among the deck hands and landlubbers. But let's be clear about the meaning of a firewall: in today's computer and Internet parlance the term is generally related to some security algorithms conjured up by highly skilled software engineers; but NPP is a 100% human task, one that requires other skills, such as recognizing and mentally analyzing the articles in the daily tidal wave of new content that arrives as a result of our very open policy towards editing the encyclopedia.

New Page Review wasn't invented yesterday

NPP has a long history. It was probably first created, or at least the original Special:NewPages, as part of the MediaWiki software at the very beginning when it was realised that some form of system would need to be instigated not only to prevent the incursion of totally unsuitable material into the encyclopedia corpus, but also to accept new articles that meet our basic criteria.

The first iteration of a guide for patrolling new pages was created March 2004 by Sethant, admin since 2004 but pretty much retired now for the last eight years. It was a far cry from the detailed set of instructions the reviewers are expected to be familiar with today, but it was start.

By 2007, the first rumblings of a need to formalize the process began, and it soon became apparent that the quality of reviewing was uncontrolled, unsupervised, and very often a hit and miss affair, frequently exploited by spammers and their socks; and that the backlog was growing wildly out of control (20,000–30,000 pages), not to mention that unpatrolled pages just slipped neatly into the corpus after just 30 days. Like many maintenance areas that are a magnet for enthusiastic new users, much of the patrolling was carried out by inexperienced editors – and even today, some of the tasks, such as maintenance and deletion tagging, are still open to all comers. The number of bots for NPP had gone from 0 to 16, somewhat reducing he workload of the 50 or so experienced users who were patrolling new pages, but still not able to stem the tide and prevent a large number of articles from lurking in the deep waters after 30 days. The process became a serious challenge to those who were striving to do it properly.

What became to be known as ACTRIAL (The Signpost, Special report, March 2018) was in fact first proposed by Beeblebrox in in 2009, but the discussion petered out without any consensus being reached.

By March 2011 the situation had become critical and The Blade of the Northern Lights raised the issue again at the Village Pump in a section titled "We need more New Page Patrollers – requiring autoconfirmation to create articles", seeding the idea of the actual trial, that took place seven years later after initial and prolonged resistance from the WMF to implement the change.

Video tutorial for Page Curation

In June 2011, Aaron Halfaker of the WMF initiated a research program, which largely confirmed the importance of the process and the issues he had clearly identified:

The research concluded with "...this analysis refute[s] the hypothesis that page patrollers' work is increasing."

In the aftermath of the Foundation's blunt rejection of the consensus for ACTRIAL, as a consolation prize they developed the Page Curation system and its feed in close collaboration with the community. They also produced excellent documentation and a very well made and compelling video tutorial.

Graph depicting the backlog of unreviewed new pages between 2016 and 2017: rapid increase in the second half of 2016, flattening until February 2018, decrease during February, and then resumption of a steeper incline until mid-May 2018
Backlog of unreviewed new pages
between 2016 and 2017

Crisis

In early 2017 the backlog had again reached crisis proportions with a backlog in the tens of thousands. Reviewers finally decided that they had the technical means to implement ACTRIAL locally and announced they would go ahead, whereupon the WMF finally relented and even provided in-depth research and support. The 6-month trial was carried out, and refuting their initial claims and concerns, the results of the Foundation's post-trial analysis conceded that there had been no detrimental effect to editor retention, nor had any significant loss of potential appropriate new articles been recorded. The trial was ratified by a further RfC (The Signpost, Special report April 2018), and the restriction on article creation in mainspace by new users was made permanent.

Who are New Page Reviewers?

The questions that arose therefore were: Who is sufficiently qualified to accept articles or label them for extinction; and how can the patrollers, or nowadays more aptly called reviewers, go about this task which all agree is onerous and often depressing? As a consequence, a new user group, New Page Reviewers, for qualified reviewers, was created in October 2016.

There are currently 807 New Page Reviewers, which makes the total number of users with this permission 1,655 (the rest are administrators).

It needs a near-admin level of understanding of our hugely complex system of policies and guidelines for notability, conflict of interest, copyright violation, and just simply What Wikipedia is not. After adminship, New Page Reviewer is possibly the most important of all the special user rights.

As hinted on by WMF executive director Katherine Maher (The Signpost, Special report, June 2018), one of the problems of NPP was that the system was accessible to all registered users regardless of whether they were qualified to assess and accept or reject new content. The present situation is that while anyone can still tag pages for maintenance or deletion by using Twinkle, overall control over accepting new pages is the realm of the authorized reviewers who are able to mark pages as 'patrolled'. Although this ensures that new articles are given an equitable assessment, it cannot prevent new users from being bitten by rough handling by less experienced editors who do not have the New Page Reviewer flag on their accounts.

New Page Review needs YOU

...and for more reasons than one!

Photograph of a brown English Cocker Spaniel sitting and wearing a red-and-white Santa cap in a snow-themed set
The Community Tech mascot:
a dog wearing a Santa hat.
  1. Development: According to their page, Community Tech states: "It's important to us... To work on projects that have a big impact". The NPR community has been asking the Wikimedia Foundation for nearly three years for essential updates and maintenance to the suite of Page Curation tools that were developed back in 2011. However, while this is a vital core process, Community Tech—a Foundation department and for additional confusion, also at MediaWiki's Community Tech, a slightly different page which additionally lists a team—insists that such development work will only be accepted following a successful annual poll, or "Wishlist" as they call it, for the most wanted comfort and convenience gadgets. Some of these developments have proven to be of minor benefit to editors and readers, but the broader community outside the reviewers is not certain to show much interest in NPP, a niche, but important Wikipedia function.

    "It is near impossible to get any useful software development out of the WMF in any reasonable amount of time."

    — MER-C

    Voting takes place 16 to 30 November at the "Wishlist" and the NPP community is hoping for a good turnout in support of the requests to Santa for the tools they need. They are counting on significant support not only from their own ranks, but from everyone who is concerned with maintaining a Wikipedia that is free of flotsam, tom-foolery, flagrant financial exploitation, and other pollution. Read what the people in the pointed red hats at San Francisco's North Pole have in store for the community's stocking fillers at Community Wishlist Survey 2019 – noting that there is not an available category for requests for work on core software extensions.

  2. Backlog: Since its inception, around 650 users have been granted the New Page Reviewer right. Many editors think perhaps, "Oh, I don't need that user right, there are plenty of reviewers already", but in reality, 90% of the reviewing is done by a small, regular corps of less than 10% of that number. Many of the reviewers were doing this work already before the user right was created and among them are some of Wikipedia's most experienced editors. Additional experienced users are needed to apply for the right and help out. It may seem harmless compared with the horrendous 2016–17 graph above, but while it has been proven this year that the backlog can be reduced almost to a day's intake, anything much over 1,500 is not sustainable. The chart displayed below updates in real time – check again tomorrow for an update.

Further reading