User:Aymatth2/SvG clean-up/Audit notes

User:Aymatth2/SvG clean-up/Guidelines gives gives guidelines for the clean-up of 16,104 articles about living people started by User:Sander.v.Ginkel (SvG) and debated at Wikipedia:Administrators' noticeboard/IncidentArchive941#User:Fram. This process was closed on 24 April 2017 and all remaining draft articles were deleted. Following that, an audit of results will be undertaken to check that articles moved back to mainspace were in fact checked and corrected where needed.

This approach was challenged on 25 April 2017 and was confirmed at Wikipedia:Administrators' noticeboard/Archive288#SvG articles – mass deletion.

Overall approach

The original definition of the audit approach was:

The results will be audited after a few weeks. This approach will be considered to have failed if a significant percentage of the articles that have been moved to main space turn out to still contain BLP violations and/or to contain information that is not confirmed by a citation to a reliable source. We will then revert to the letter of the close at the ANI discussion. The decision there was to give one week for rescue attempts followed by bulk deletion:

Editors whose restorations to mainspace were problematic will be banned from participating
Other editors will be invited to move articles they want to save from main space or draft space to their user space
After the bulk deletion, these editors may fix the saved user space articles and move them to main space.

It was pointed out in subsequent discussions that one or two busy but careless editors would cause the careful work of many other editors to be lost. To address this issue,

If the audit determines that most of the problems were introduced by just a few editors, all the articles that those editors moved will be deleted from mainspace, other errors detected will be fixed, and the project will be considered complete.
If the audit determines that the problem is more widespread, we will revert to the original audit approach: one week's notice for userfication, then delete all the restored articles.

Notes on audit

2,930 articles had been restored to mainspace as of 19 March 2017, so probably the total restored will be no more than 4,000. A 5% sample (200 articles) should give an accurate measure of overall quality. Possibly the simplest approach will be to select every 20th article from the mainspace list, and divide these up between audit volunteers.
A sortable list should be created giving article title, editor who restored it, auditor, and any problems found
Minor errors should be noted and fixed, but are not cause for alarm.

The "is/was" problem is common, as in: "She is a member of the national team" or "He was an athlete". We do not know that she is still on the team, and probably do not know that he is no longer an athlete. The wording should be fixed: "She competed on the national team in 2016. He competed as an athlete in 1976".
It is fairly common for the article to give information that is not backed up by the cited source, but that is correct. In this case a source should be added to back up the information.

Where an editor appears several times on the list with articles that have problems, including minor problems, this may be evidence of inadequate care in reviewing

A separate list should be created for that editor, giving date/time of move, article, edit comment and any problems found
If it seems clear that the editor has been restoring articles without checking them, they go on the blacklist and all the articles they moved will be deleted.

If there are widespread non-trivial problems missed by non-blacklist editors, we revert to the original approach: one week's notice for userfication, then delete all the restored articles.

Working lists

Counts by editor

Counts by editor	Add your name to entries you plan to audit, and add results after completed

High volume editor moves

High volume (50+ moves) editor lists, with audit notes. The purpose of the audit is not so much to detect serious errors as to find failures to check articles against the sources. For example, an article may say Jane Doe was born on 3 March 1954 and is a former weightlifter, but the source says she was born on 7 March 1954 and does not say she has retired. The error in the birthdate is minor and given her age she probably has retired, but that is a guess. If the article had been checked these errors would have been found and fixed.

List	Editor	Count	Move log
Arielslytherin list	Arielslytherin	613	Move log
Beatley list	Beatley	1530	Move log
Gtrbolivar list	Gtrbolivar	101	Move log
Hmlarson list	Hmlarson	324	Move log (Some moves after cut-off)
Inwind list	Inwind	100	Move log
Lugnuts list	Lugnuts	860	Move log
MFriedman list	MFriedman	337	Move log
Onel5969 list	Onel5969	161	Move log
Raymarcbadz list	Raymarcbadz	820	Move log (Some moves after cut-off)
Sillyfolkboy list	Sillyfolkboy	178	Move log
SuperJew list	SuperJew	122	Move log
ThiagoSimoes list	ThiagoSimoes	50	Move log
XyZAn list	XyZAn	93	Move log
all other editors	–	555
Total	–	5,844

Master lists

User:Aymatth2/SvG clean-up/Moves	Breakdown of all editor activity as of 25 April 2017‎ (thanks to user:JJMC89)
All moves by editor and date	Breakdown of all editor activity as of 25 April 2017 sorted by user, date/time
Mainspace complete list	List of articles remaining in mainspace after moves back to draft and deletions