Public outreach/Academy/RfC/John's talk

Wikipedia's system of quality management

Underlying messages of the talk

• That you hardly ever see vandalism is not just because good editors outnumber bad ones, or have good tools, but because the immune system is so multi-layered.
• That Wikipedia takes itself seriously: quality matters.
• That as an editor, this complexity affects you mostly because it protects articles that you've edited from being worsened by other editors, but it's something you don't need to understand in order to be a successful editor.
• That there is inconsistencies and incompleteness to even such things as measuring quality; that's because no one is "in charge" of deciding what editors do, and because Wikipedia (always) needs more editors.

Presentation

The environment

• Millions of articles, with several thousand added every day
• No barriers to editing: registration not required, valid email address not required for registration
• Tens of thousands of new editors every month
• About 50% of edits from the U.S., remaining 50% from across the world
• 100% of writing done by volunteers, for both cost and legal reasons.
• Absolutely flat editing structure: no "super-editors", "senior editors", "topic editors", "managing editor", etc. No one can tell another editor what to work on.

Vandalism and spam (the easy stuff)

Keeping the bad stuff out: Wikipedia's immune system against vandalism and spam * Blocks

• • Identifying problem editors
  • Actions by admins
• Page protection (semi, full)
• Automation
  • Edit Filter
  • Spam blacklist
  • New/IP editor restrictions:

*** Can't rename articles

• • • CAPTHA for external links
  • Anti-vandalism bots
    • Use en:Cholangiocarcinoma (a main page article, June 2009)
• Humans
  • RCC patrolers
    • Wiki: revert to prior version
    • Ease of reversion (undo, rollback)
    • Automated tools like Huggle
  • Watchlists

*** Tailorable

• • RSS, IRC feed options
• WikiProject Spam
• Other bots
• Regular editors (checking page history before editing)
• Readers who notice vandalism

* Semi-protection

Content disagreements

Norms

• Wikipedia's core (policies about quality): WP:V, WP:NPOV, and WP:NOR
• Resolving content disagreements by consensus
  • Takes longer, often fails in the short run, very difficult to dislodge determined POVers
  • No alternative, given that Wikipedia has a flat (as in, no) hierarchy of editors
  • Wikipedia has a flat hierarchy because anything else doesn't scale, given the ambitions of the project and the funding (de minibus)

Minimization: content policies and guidelines

• WP:NOT
• Naming conventions

Shaping the process of reaching consensus

* Conflict of interest

• Discussion and explanation
  • Forum: Article-related discussion on article talk pages
• No excessive reverts:
  • 3RR: If it's just one editor trying to have his/her way, can't keep reverting forever
  • Edit wars: Full protection of individual articles
• Formal mechanisms for escalation:
  • Third opinion, Request for Comments
  • Mediation (informal, formal)

** Arbitration Committee

Enforcement

• Graduated warning system, with emphasis on prevention, not punishment
  • Done by regular editors

• Roughly 1500 administrators.
  • Janitors, not supervisors.
  • Selected by the community, deselected by super-admins (bureaucrats) if necessary

* Monitoring/alarms:

• • WP:AIV
  • Many, many noticeboards, including WP:AN and WP:ANI
  • Individual admins respond to specific problems

Community and quality

* Defining appropriate behavior:

• • Civility, no personal attacks
  • Other behavioral guidelines

• Helping editors learn about editing
  • Massively extensive documentation
  • Discussion/talk pages for policies and guidelines
  • Help desk, village pump, etc.
  • Mentoring, editor review, etc.

• Formal mechanisms for collaborations and cooperation
  • WikiProjects

**Peer reviews

• • Reviews of good article and featured article candidates

Closing

• Flatness: the ratio of good to bad matters greatly
  • For vandalism and spam, technology has been a significant force-multiplier for the good

Incrementalism as a philosophy of quality

• "Improve pages wherever you can, and do not worry about leaving them imperfect."

** For other content issues, eventualism has been a winning approach so far. It will continue to be, provided that the ratio remains high enough.

• • • Eventualism as a personal philosophy: If you're dealing with an idiot with a lot of time on his hands, work elsewhere. Let him lose interest. Let others deal with him. Nothing is at stake but your stress level.