Yesterday I set up a first unofficial demo of stable versions in the configuration which I would like to see used on the English Wikipedia. It’s based on the FlaggedRevs extension by Aaron Schulz & Jörg Baach. Unofficial, because Brion is currently reviewing the code for security and scalability. I’m still slightly worried that we might hit a snag, as the extension goes pretty deeply into the way MediaWiki serves pages, and of course Wikipedia can’t afford to run something that causes major slowdowns.

Still, if you want to get a feeling for the kind of configuration that I, Jimmy Wales and Florence Devouard have expressed support for, play with it. The main new thing you’ll notice is a top right corner icon which indicates the status of the page you’re looking at. On selected pages, the last revision shown to unregistered users is the most recently vandalism-patrolled one, while the rest of the wiki behaves as before. Essentially, in this configuration, it accomplishes four things:

  • Allows us to open up semi-protected pages to editing, since edits have to be reviewed and vandalism does not affect what the general public sees;
  • Allows us, as a consequence, to also use this kind of “quality protection” more widely — e.g. when an article has reached a very high quality, and is unlikely to be substantially be improved by massively collaborative editing
  • Improves our vandalism patrolling abilities, since vandalism doesn’t have to be repeatedly checked: we record who has looked at which changes. Also, changes by trusted users don’t have to be reviewed. The tagging system ties into “recent changes patrolling”, an old feature that has never scaled well.
  • Allows us to make processes like “Featured article candidates” and “Good article candidates” revision-based rather than page-based. Thus we can more easily track when an article reached a certain quality stage, and can better examine whether changes past that point have increased or decreased its quality.

Some wikis would like to use more restrictive configurations than the one shown here. For example, a group in the German Wikipedia community is advocating to let all edits by unregistered users be reviewed before applying them, rather than only doing so on selected pages. I’ve argued against this at some length here. I think the current configuration strikes a good balance of openness, quality, and transparency to the reader. Of course there’s already a big wishlist, some of which should be addressed before taking this feature live.