User:Risker/Content moderation

Methods of content moderation

There are multiple types of content moderation carried out on Wikipedia, described below:

Editing

• Most common type of content moderation, used by any editor (whether or not confirmed or registered) to modify the content of a page.
• Used by preference by some editors when added content is at least partially appropriate.
• If added content is entirely inappropriate (e.g., vandalism, test edits, poorly sourced fact information such as sports scores), revert of the edit is normally used. See below.

Revert (undo/rollback) of edit

• See Help:Reverting
• Second most common type of content moderation, used by any editor (whether or not confirmed or registered) to return a page to the previous version.
• Used for vandalism, certain types of content removal
• "Undo" is considered the gentlest form of reversion, and allows the reverting editor to include an edit summary explaining the reason for reversion. This is the most common method of reverting good-faith edits that are not appropriate to the page on which they have been made.
• Various tools such as Twinkle add tabs or options to rollback an edit, usually with an automatic edit summary or no edit summary. This is primarily used for vandalism.

Revision deletion

• See Wikipedia:Revision deletion
• Deletion of one or a series of edits (or log entries) to remove specific content/material from public view. Sometimes called "hiding" or "changing visibility".
• Can be carried out only by administrators or persons with Oversight permission
• Once content has been revision deleted, the content of the selected edits cannot be seen by anyone without administrator/oversight permissions
  • This can create challenges when reviewing the history of an article edit-by-edit, so its use is restricted. See criteria.
• Revision deletion is reversible by administrators/oversighters
• Revision deletion can also be used to redact username or action summary information in logs. Its use for this purpose is less common and is more likely to be subject to review or questions.

Suppression (Oversight)

• See Wikipedia:Oversight
• Deletion of one or a series of edits or log entries to remove specific content/material from the view of both the public and administrators. Called "suppression" and "oversight" interchangeably.
• Can be carried out only by users with oversight permissions.
  • Oversighters are appointed by the Arbitration Committee (Arbcom), and are (a) current Arbcom members, (b) former Arbcom members and (c) selected community members.
• Once content is oversighted, the content of the selected edits cannot be seen by anyone (including administrators) except for other oversighters.
• Use is regulated and restricted. Oversighters are expected to review each other's activities, and act as a self-regulating group.
• Used mainly for material that is privacy-violating and/or potentially harmful (to a person or their reputation, to the project, to a person or place off-wiki, etc.), or that could make an individual excessively vulnerable (e.g., self-identification by minors, certain mental health issues).

Deletion

• See Wikipedia:Deletion
• Removal (deletion) of the entire article or page.
• Can only be carried out by administrators. Once deleted, the deleted contents of the page can be viewed by administrators/oversighters.
• Deletion is carried out using established criteria, see proposed deletion, speedy deletion and the various "XFD" pages (articles, files, miscellany, redirects, etc.) for applicable criteria.

Suppression Deletion

• No good documentation for this process, although it is briefly discussed at Wikipedia:Oversight
• Can only be carried out by oversighters. Once suppress-deleted, the contents of the page can only be viewed by oversighters.
  • Oversighters delete the page while also checking the "suppress data" box.
  • Oversighters may occasionally undelete a previously deleted page and then immediately redelete it using the suppression function as an alternative to suppressing each deleted edit individually. Most frequently occurs when the page has a significant (25+ revisions) history.
• Used when every revision (or many discrete revisions) of the page qualifies for suppression.
• Least common method of content moderation.

Page protection

• See Wikipedia:Protection policy
• Protection is applied by administrators, but editors may request various levels of protection at requests for page protection
• For descriptions, see the overview. Chart is reproduced below:

Protection levels and their impact on various Wikipedia user groups^[α]

F	Protection level	New or unregistered editors	Confirmed	Extended confirmed	Template editor[β]	Admin	Interface admin[γ]	Appropriate for...
Editing	None	Normal editing						The vast majority of pages.[δ]
	Pending changes	Can edit Changes are only visible to logged-in users until reviewed by a pending changes reviewer or administrator.[ε]	Can edit Changes are visible to everyone if there aren't any unreviewed pending changes. Otherwise, they are only visible to logged-in users until reviewed by a pending changes reviewer or administrator.[ε]			Can edit If there are any unreviewed pending changes, the administrators will be required to review them before they can edit the page.[ε]		Infrequently edited pages with high levels of vandalism, BLP violations, edit-warring, or other disruption from unregistered and new users.
	Semi	Cannot edit	Normal editing					Pages that have been persistently vandalized by anonymous and newly registered users. Some highly visible templates and modules.
	Extended confirmed	Cannot edit		Normal editing				Contentious topics authorized by ArbCom, pages where semi-protection has failed, or high-risk templates where template protection would be too restrictive.
	Template	Cannot edit			Normal editing			High-risk or very-frequently used templates and modules. Some high-risk pages outside of template space.
	Full	Cannot edit				Can edit[ζ]		Pages with persistent disruption from extended confirmed accounts.
	Office	Cannot edit				Can edit[η]		Pages that the Foundation has determined to be exceptionally sensitive.
	Cascade[θ]	Cannot edit				Can edit		Particularly visible pages, such as the Main Page, to prevent vandalism to pages that are transcluded onto them.
	Interface[ι]	Cannot edit					Normal editing	Scripts, stylesheets, and similar objects fundamental to operation of the site or that are in other editors' user spaces.
Creating pages	None	Cannot create[κ][λ]	Can create					The vast majority of page titles.[δ]
	Create	Cannot create[λ]		Adjustable Protection may be applied to neither, either, or both groups.		Can create		Pages that have been repeatedly and problematically re-created. This form of protection is often called "salting".
Moving pages	None	Cannot move	Can move					The vast majority of pages.[δ]
	Move	Cannot move		Adjustable Protection may be applied to neither, either, or both groups.		Can move		Pages that have been the subject of move wars. Pages that are edit-protected are usually also move-protected at the same level.
Uploading files	None	Cannot upload	Can upload					The vast majority of file names.[δ]
	Upload	Cannot upload		Adjustable Protection may be applied to neither, either, or both groups		Can upload		Files that have been repeatedly uploaded after deletion
Notes: Details about historical protection levels and user groups no longer in use are available at Wikipedia:Protection policy § Retired protections. This table assumes that template editors are also extended confirmed, which is almost always the case for non-bot accounts. This table assumes that an interface administrator is also a "regular" administrator, which is almost always the case in practice. This is the default protection level. However, if any editors (including unregistered editors) revert all unreviewed pending changes back to the latest accepted version, that revision is automatically accepted and pending changes reviewers and administrators aren't prompted or notified. Only noncontroversial changes or requested changes following an achieved consensus should be performed. Only with the approval from the Wikimedia Foundation. Cascade protection extends to all pages that are transcluded onto the protected page, unless the transcluded page is at the same protection level or higher. Cascade protection can only be applied to pages that are fully or office-protected because otherwise it creates a workflow flaw. The interface protection level is automatically set by the MediaWiki software to a specific set of pages, such as pages in the MediaWiki namespace, system-wide CSS and JavaScript pages, and personal CSS and JavaScript pages of other users. It is not a protection level that an administrator can manually apply to any page, nor is it a protection level that can be modified on pages currently under interface protection. Because of this, administrators also cannot cascade-protect pages that are Interface-protected. This has been in effect for unregistered users since 5 December 2005. The restriction was extended to newly registered users on a six month trial basis starting on 14 September 2017. The extension became permanent on 18 April 2018. Under the default no protection, unregistered and newly registered users can create talk pages in all namespaces and draft articles in the Draft namespace. For these namespaces, it would therefore be possible for the creation protection to only apply to unregistered and newly registered users

Moderating potentially harmful content

The primary methods of identifying problematic content are:

• Recent changes (RC) patrol - carried out by all types of editors based on an ongoing log of changes to all pages.
  • Focus is primarily on "article space" - in particular, articles - which is where potentially harmful content is most likely to be widely disseminated. These are the pages that show up on google searches.
  • RC patrollers do not normally review edits to noticeboards or user talk pages, but may review edits by "other editors" to user pages or to templates or pages in the Wikipedia space that are rarely edited (such as policy pages), particularly if the edit is made by a new/unregistered user
• New Page Patrol (NPP) - carried out by administrators and editors with a special "new page patroller" permission
  • Main focus is on newly created articles, which are reviewed for quality but will also flag potentially harmful content.
  • A few editors (mostly administrators) will do periodic sweeps of other new pages, particularly those in "user space" or sometimes "draft space", specifically to identify problematic/inappropriate/potentially harmful content.
    • Most common example is minors writing detailed pages about themselves/their families/their friends, etc. These are normally flagged for oversighter review/attention.
• User contribution logs - carried out by all types of editors on a case-by-case basis when an account is identified to have made a problematic edit. Particularly focuses on new/unregistered accounts, although "experienced" accounts that have been dormant for a long time and suddenly start making problem edits will also be examined. Often these accounts will be brought to the attention of an administrator, who may well block the account.

In addition:

• If a non-admin editor identifies material that should be revision-deleted or suppressed, they can (a) revert and privately notify an admin/oversighter or (b) email Oversight (an OTRS queue where oversighters receive requests for suppression and/or revision deletion).
• If an admin identifies an edit that should be suppressed, they will normally revision-delete and (a) privately notify an admin/oversighter or (b) email Oversight

The chart below discusses various potentially harmful content types and the manner in which it is addressed on English Wikipedia. Preferred method is bolded. Examples are provided where appropriate (using quotation marks), and are all fictional.

Content type/example	Edit/Revert	RevDelete	Suppression	Other/Notes
Insulting content directed at a third party or article subject	"John Smith is a jerk"	"John Smith jerks off in the school washroom"	"John Smith rapes babies"	Dependent on circumstances, may also result in a warning or block. If edit qualifies for suppression, user will probably be blocked. User may be "Oversight- blocked" if they continue to post the suppressed material after warning/block.
Insulting content directed at an editor	"You're a dick"	"You're a f**king dick who wouldn't know his ass from a hole in the ground"	"You're a f**cking dick who rapes babies in your spare time"	(As above)
Non-public information about article subject, third party, or editor	(revert and report to OS or admin)	(revdelete and report to OS)	Examples: Phone numbers, addresses, non-public names/DOB of offspring, "outing" of editors,	(As above)
Threat of self-harm ("I am going to kill myself")	(revert and report to OS or admin, can also notify emergency@)	(revdelete and report to OS, can also notify emergency@)	Suppress and notify emergency@	emergency@ is an email address monitored 24/7 by WMF Trust & Safety staff, who analyse the edit and contact appropriate Emergency Services.
Threat of harm to other individuals/locations ("I am going to shoot John Smith", "XYZ School will be bombed on Thursday", "I will be waiting outside your workplace tonight")	(revert and report to OS or admin, can also notify emergency@)	(revdelete and report to OS, can also notify emergency@)	Suppress and notify emergency@	T&S will likely attempt to communicate with editor if editor is the subject of the threat, other editors who are close to the threatened editor may also reach out.