User:WereSpielChequers/Newbie treatment at NPP

Preface

This Essay is primarily based on the results from Wikipedia:Newbie treatment at Criteria for speedy deletion (NEWT), but has also been influenced by my own experiences at CSD and at RFA, by discussions at Strategy, by some of the presentations and discussions at wikimania 2009, and by essays by User:SoWhy and others. My thanks to all who took part in NEWT and for the various related discussions, my apologies to those who were offended by our methodology. what follows is a personal analysis of the situation and some potential ways forward.

Summary

At the moment submitting new articles to wikipedia is a bit of a lottery, especially if you do so from a new account. Whether your experience as a newby is good, bad or indifferent depends on your subject, level of completeness when you start the article and crucially on which new page patrollers happen to look at your article. My own experience in the test was actually quite reassuring - both the articles I submitted survived unscathed, as I believe the great majority of paragraph length new articles on Encyclopaedic notable subjects. But there are some mistakes happening, and an awful lot of people adding stuff in good faith that frankly doesn't belong in an encyclopaedia.

We do have several problems with the Article Creation/New Page Patrol processes. Simply improving the accuracy with which people apply the current speedy deletion criteria will not solve all those problems.

Wikipedia is an Encyclopaedia that anyone can edit, and that is not constrained either by the physical space limitations of a printed Encyclopaedia, or by the deadlines of a printed media. But it has set some rules on wp:notability to determine what sort of articles do and do not belong in Wikipedia.

This can at times lead to conflict on several fronts:

1. Every day, Vandals, spammers, cyber bullies and extreme inclusionists create hundreds of articles that don't belong in Wikipedia and need to be deleted.
2. Inclusionists and Deletionists disagree on several boundary issues such as whether all tenured Professors are notable, and whether competitors at certain levels in certain sports are "professionals".
3. Eventualists and Immediatists clash as to whether the notability of a subject should be decided on the basis of the first sentence written at the creation of an article, or on the potential of the topic.

Modus Operandi

WP:NEWT was setup with the following instructions, which survived almost unaltered for the duration of the data collection phase:

This page was created as a response to this critique of wikipedia, which poses two challenges: can we avoid the problem of new articles by newbies being tagged for speedy deletion within two minutes? And, can such new articles survive for 7 days?

To take up the Let's all create an extra account challenge:

1. Create a new wikipedia account (remembering of course to inform Arbcom per WP:Multiple Accounts by emailing (arbcom-llists.wikimedia.org)
2. Write an article that doesn't meet the deletion criteria
3. After seven days add it to this page by creating a row in the results table.
4. Create a subpage to this page in your user name and transclude it here whilst discussions are ongoing
5. When you are ready to unveil your article notify all involved parties of this page, but please take care to be as discreet and gentle as possible when informing fellow editors of any mistakes they may have made.

In hindsight it was probably a mistake to link to the speedy deletion criteria, as some articles submitted met the deletion criteria but not the speedy deletion criteria (we also had the problem that a site that seeks to undermine wikipedia misrepresented this project as deliberately creating articles that should have been deleted). We should also have emphasised from the start that pointing out other editor's mistakes should always be done sensitively.

As with any new project there is a risk of something unexpected happening, I was surprised to discover that some people consider Mystery shopping unethical. Personally I would be shocked and disappointed if I discovered that the guide I buy when I go backpacking to a country had been written without using mystery shopping techniques, I use such techniques in real life and would think a business that didn't was inept. However I now know that the technique is controversial in Wikipedia, it was sufficiently controversial that we had to stop NEWT and I've since advised people against similar ventures. To me it is a bit like pork. I eat pork, I serve pork to guests if I know they are pork eaters, but I would suggest leaving it off the menu at Wikimania and similar events as I know that many people don't approve of it.

Findings

Problems

Some articles are being deleted that should not be deleted, and as a result Wikipedia is losing contributions that should be part of the encyclopaedia. The article creation process is unnecessarily stressful and maybe contributing to our failure to keep as many of our first time editors as we used to. There is a division within the community as to whether certain work is useful or not, and some Newpage patrollers don't find out that their work is seen as problematic until it is reviewed at their Request for adminship. This exacerbates our existing problems of a cadre of active admins who are dwindling by 1% monthly, and upsets and risks losing some of the committed Wikipedians whose offer to take on an adminship role can be the first time they discover that their work is not appreciated, or considered incorrect by many in the community.

Scientific validity

Several editors expressed concerns about the scientific validity of the study and expressed concerns about the validity of any data arising from it. I would like to add that the study combined features of a Focus group and along with some specific tests. We have established that though errors take place at speedy deletion, articles by newbies are not always tagged for speedy deletion within two minutes, nor are they guaranteed to be deleted within 7 days. We have not established the statistical probability of an article being incorrectly tagged or incorrectly deleted, but there have been a number of discussions at Wikipedia Talk:NEWT as to how that could be done if it were deemed useful.

Recommendations

I've come to realise that my concerns about the Newpage Patrol system fall into several areas: It bites editors and especially newbies, it is inefficient at using our volunteers' time. It fails to give feedback to taggers who make mistakes. These were the recommendation that came out of NEWT, some I've subsequently expanded and updated.

1. strategy:Proposal:Speedy deletion - 24 hour pause for some articles
2. strategy:Proposal:Welcome all useful new users, if necessary by a bot
3. Wikipedia talk:Criteria for speedy deletion#Suggestion - authors must be informed

   Wikipedia:Bots/Requests for approval/SDPatrolBot II a bot that would do this

4. Wikipedia talk:Newbie treatment at CSD#Proposal/Idea for NPP changes - Twinkle and Huggle scripts for newpage and construction tags
5. Wikipedia talk:Criteria for speedy deletion#Should unsourced become a speedy deletion criterion?
6. Wikipedia talk:Newbie treatment at CSD#Informing taggers that you've declined their tags
7. MediaWiki talk:Welcomecreation review the welcome screen that you briefly see when you create a new account.

Redesign the deletion process

My experience is that many patrollers start off choosing which speedy criteria something belongs to, and only after a while do they move on to using a mix of Prod, AFD and Speedy. I think it would be better for all concerned if people started out learning the deletion policy for an area like sports, film or music and concentrated on that area before moving on to others.

This would require a radical redesign of the deletion process to retain its functional complexity, but reduce its apparent complexity. Prod, CSD, MFD and AFD all have their virtues, but if one were designing a deletion system from scratch would one choose quite the complexity of names and templates that we do? With the current documentation and dropdown menus the hierarchy of decision making for patrollers is upside down; Instead of first choosing among the prod criteria or among the speedy criteria patrollers should have dropdown menus that are more like:

• Music
• Film
• Sport
• Companies
• Biographies
  • Sourced or unsourced?
  • Attack?
  • Admitting unimportance
  • Not asserting importance or significance
  • Asserting importance but not notable?
  • Notable?

The patroller wouldn't need to know that if they tick "admitting unimportance" it tags the page with {{db-bio}} whilst "Asserting importance but not notable?" gets a {{prod}}, and each generates a slightly different template for the author.

This approach also lends itself to integration with our categorisation system, so Newpage patrollers could be adding some relevant categories as well as the tags.

Making the process more efficient

Auto-Reviewer (Now Autopatrolled)

Articles by admins and wp:autopatrolled editors are automatically marked as patrolled. There are a lot of editors out there whose articles always get patrolled or (or are only speedied in error). Appointing more of them as admins or autopatrolled would reduce the workload at New Page Patrol. Unfortunately WP:RFA is deeply flawed if not broken and our number of active admins is declining by 1% a month. However appointing Autopatrollers is easy and painless and it is rarely controversial to appoint lots more of them. One way of doing this is to check out the good authors you encounter at New Page Patrol and if they meet the criteria nominate them for autopatrolled (or if you are an admin set the flag yourself). A more organised autopatroller recruitment drive is detailed at Wikipedia talk:Autoreviewer#Bot to find article creators.

Since NEWT there have been at least two big drives to recruit Autopatrollers, and the guideline has been lowered from 75 articles to 50.

More Colours at Newpage Patrol

Currently Special:Newpages works on an overly simple system, unpatrolled articles are highlighted in yellow, patrolled ones in white. A slightly more complex colour system would have advantages, and I suggest:

1. Yellow: - neither patrolled nor tagged for deletion
2. Dark pink: - tagged as {{|G3}} or {{|G10}}
3. Light Blue: tagged for other deletion other than by {{|G3}} or {{|G10}}
4. Dark Blue: no longer tagged for deletion but not patrolled.
5. White: patrolled and not tagged for deletion

This would have the following advantages: It would save the faff of marking a page as patrolled when you tag it for deletion. It would clearly differentiate between speedies declined by patrollers or admins and speedies declined by the author logging out and editing as an IP. It would bring G3s and G10s to the immediate attention of any admins at Special:Newpages

Antechamber

The community is currently divided between those who consider A1 and A3 and sometimes even A7 tags in the first few seconds of an articles existence as overhasty, and those who consider they are judging the article as submitted.

This could be resolved by splitting the queue and NPP into three rather than the current two. An Antechamber where only bad faith articles such as attack pages and vandalism are deleted. Then after the article has been up for a certain period of time (at least one hour perhaps 24) it would go through Special:Newpages as at present. (The back of the unpatrolled queue would be unaltered). Those inclusionists who want to accept as much as possible could concentrate on the antechamber, whilst templaters and deletionists would be encouraged to evaluate the articles when they leave the antechamber and enter Special:Newpages.

A further refinement if possible would be for new articles by authors who've previously had articles deleted as G3 or G10 to bypass the Antechamber and appear immediately at Special:Newpages.

Make NPP more friendly

Wikipedia has become a less welcoming fun place and as a result we are not retaining as many of the volunteers who try us out as we once did - and we lose a steady stream of experienced editors. The New Page Patrol process contributes to this by biting several different types of user, much of this could be prevented.

Welcome more newbies

Though some of our "newbies" were welcomed several weren't - we didn't do a large enough sample to quantify this, but a cursory glance at the back of Special:Newpages will usually find a few redlinked talkpages amongst the authors of articles submitted thirty days ago.

Some Wikimedia projects welcome all new editors at least with a bot message, this has been rejected several times on EN wiki, but if we continue to fail to welcome a significant proportion of newbies then perhaps this should be re-evaluated. More seriously only a tiny proportion of the people who create an account ever do more than a handful of edits. If welcoming more of them would change this then we should seriously consider doing so.

This is a Wikipedia:Bot requests/Archive 31#Welcome Bot - advanced version proposal that in my view answered most of the common objections to a welcome bot on EN wiki. But what I think we need is a proper scientific test - take the new editors with redlinked talkpages from a particular period of time divide them randomly into several groups, give all but one group a different welcome message and leave the control group unwelcomed. Then compare how many edits each group subsequently made and how many blocks each received. Whichever group edited most and was blocked least we should consider implementing that bot (or lack of it - we don't know if these welcomes work or not).

Revise the templates

Some of the templates that we use for deletion are overly bitey - in particular we shouldn't use the same or similar warning signs that we use for previously warned vandals for people who have made a good faith contribution that we consider doesn't really belong.

Resolve some policy differences

Part of the reason why New page patrol is a perpetual trainwreck is that we have New page patrollers, admins and article creators with contradictory ideas as to what the minimum standard is for a new article.

At one extreme you get articles deleted for "poor formatting", at the other extreme we have hundreds of totally unreferenced articles created every day.

While I prefer our policy of verifiable over the alternative policy of verified, I think that a workable solution to the current mess would be to adopt a policy of verified for all new articles. If we alter the creation process to require a source for all new articles and automatically prompt article creators accordingly then we will have a more workable system at newpages.

We are not good at picking up on and dealing with newpage patrollers whose error rate means that they do more harm than good. I'm not sure what the best solution is to this, but for some individuals a topic ban from special:newpages may be the sort of last resort that we need.