User:Femke/AfC improvements

Last big improvement drive was 2018: Wikipedia talk:WikiProject Articles for creation/AfC Process Improvement May 2018? Not sure what was changed then.. Ask meta:User:MMiller (WMF) for outcome? Automatic copyvio test seemed part of the plan. More background: Wikipedia:The future of NPP and AfC.

Expected outcomes of ideas
Change	Expected outcome	Possible negatives	Progress
Inactivity criterion	up to 1% more reviewers	-
Better decline messages	-3% resubmission non-notable + resubmission SNG notable		Half done
Explain notability a) pre-submission template b) SIGCOV in Article Wizard c) improve user talk message	-5–15% non-notable submissions	-2% notable submissions	Half done
Bots for reviewers	+10% efficiency
Bots for writers	-5–10% declines, more engagement from writers	New editors might be frustrated if a bot gives a hint, and an AfC reviewer does not give them time to rectify the mistake and declines the submission.
Progress bar Or option to put articles on hold	+2% resubmission for notable topics, -2% for non-notable topics	May require too much coding for expected outcome
Suggested articles	+3% retention
Deprioritising Spam	+up to 1% retention good-faith / -1% retention spammers	Spammers may become autoconfirmed when waiting long, and patrollers may lose out on decline/reject red flags

Possible change to afc reviewer tool: some 'are you sure' when submitting with very little difference from original

Mark reviewers as inactive after 12 months of inaction / 3 months not editing

Just a bit less bureaucracy. AWB also uses 12 months.

Decline message

Our notability decline message is a wall of text, and difficult to parse. Proposed a solution in (format source requirements in bullet points for clarity) this discussion . After long wait, discussion was closed. Done

Our SNGs were poorly communicated too: most profs won't meet GNG, but the prof decline message still puts more emphasis on GNG, and does not tell the reader WHY they should click on the SNG link. The CORP decline message equally doesn't capture the spirit of the stricter rules. The second discussion (first) led to a consensus for change. Done

Other SNGs are also poorly communicated, but the relation to GNG is in flux.

Notability communication

There are (at least) three places where new editors should/can learn about notability. The most important is probably the Wikipedia:Article wizard, which does not touch upon the 'significant coverage' aspect (essential), nor the secondary source (slightly less essential, as independent reliable sources are often secondary). The second one is Help:Introduction, which has the amazing quiz. And the last one is when they are writing their draft, which shows the {{AfC submission/draft}} template.

Change the pre-submission draft template Done

I can edit test edits at: Template:AfC submission/draft/sandbox

The {{AfC submission/draft}} template is very focussed on copyvio (90% of editors understand this pre-Wikipedia), and not writing with COI (most people would be able to guess these rules). No explanation of notability. Given most of our drafts are declined for not showing notability, this is a great miss. I propose these changes: A. Remove sentence Note: The submission-received box will appear at the top of the page. If it's there, your draft has been submitted correctly. -> not sure what its function is. How do people incorrectly submit their draft? Replace text with

Show the subject qualifies for a Wikipedia article. Typically, this can be shown by using a few sources that meet four criteria. The sources need to be (1) reliable (2) secondary (3) independent of the subject (4) talk in depth about the subject. For some topics, there are alternative ways to show they qualify for an article.
Write from a neutral point of view
Respect copyright laws, and never plagiarize from any sources.

It is strongly discouraged to write about yourself or your own business. If you do so, you must declare it. {{

Improve explanation notability in Article Wizard

Will need some good design, only needs a few extra words, but goal is to keep it very succint. The text is now

The topic of an article must already be covered in reliable sources that are independent of the subject. These include academic journals, books, newspapers, magazines, and websites with a reputation for fact checking. Social media, press releases or corporate/professional profiles do not qualify.

Which means it omits the 'in depth' or sigcov aspect of notability, as well as the secondary aspect.

Change the user talk messages after decline

The Template:AfC decline is a wall of text, which does not point the editor to concrete things they have to learn to be more successful the second time around.

Get rid of deletion bullet point or points Done
Add a blue button to the notability quiz
Add a blue button to the neutrality quiz?
Add a blue button to 1 min explanation reliable sources.

Better use of bots

Two tasks take up a lot of time, but bots exist that could do this:

Initial copyright check via Earwig (was done by Wikipedia:Bots/Requests for approval/EarwigBot 1 from 2009 to ??. Says it’s replaced by web tool, which requires reviewer to manually click and wait.
Changing bare url into fuller citations (citationBot only works when called by editor, should be automatic)

AfC Assistant Bot

Quite a few problems with articles can be bot-detected, and when solved, make reviewing easier

Detect when overcitations occurs
Detect when there are external links in the article
Detect citations to unreliable websites
Has no references
Uses Twitter or other social media websites
Uses YouTube (multiple times

It would be nice if we can develop a bot to give feedback quite fast after submission. It's more likely that draft writers are still around then (compared to a few months later). Probably start with around three problems, and look whether new users actually respond to the bots (they often don't listen to reviewers, so will require getting some baseline there)

Progress bar or putting drafts on hold

In general, we’d like to stimulate new editors to continue to work on articles that are notable, and stops working on articles where they can’t show notability after a decline. It may be an idea to put a two-step progress bar on the article, which shows to new editor when their subject qualifies for an article

Instead, we could introduce a fourth decision: hold: for articles that are notable, but not yet there (usually small NPOV issues). The expectation is that the first reviewer re-reviews (which should take less time as they've already done most reviewing work). Now, we often leave comments to be addressed when an article isn't that bad. The article stays in the queue, meaning that others will land on the draft before the author(s) have a chance to address the faults. After a week on hold with no action, the article could go back in the normal queue.

Deprioritise spam

When clicking next random draft, articles that are tagged as possibly spam are deprioritised (say 3x less chance to appear).