On Wed, Apr 3, 2013 at 3:30 AM, Ori Livneh ori@wikimedia.org wrote:
On Tuesday, April 2, 2013 at 5:45 AM, Quim Gil wrote:
I have been drafting a proposal to attract new contributors, help them settle in, and connect them to interesting tasks. It turns out that many of these problems are not unique to new contributors. We suffer them as well and we are just used to them.
The proposal has evolved into a deeper restructuring of our community spaces. We're still drafting it, but a round of wider feedback is welcome before opening the official RFC at
https://www.mediawiki.org/wiki/Requests_for_comment/Wikitech_contributors
In summary:
- wikitech.wikimedia.org (http://wikitech.wikimedia.org) would become
the one and only site for our open
source software contributors, powered by semantic software and an ontology of categories shared across wiki pages, Bugzilla and hopefully Gerrit.
That seems wrong. Of the two, MediaWiki.org is clearly the more successful wiki. It is larger by all measures, and draws a wide pool of active contributors. This is reflected in the quality of the documentation, which (pace self-deprecating humor) tends to be quite good. Up until recently a good portion of the links on Wikitech's main page pointed to content that was flagged as obsolete. It has come some way since then, but not quite enough to subsume mediawikiwiki.
mediawiki.org will still exist to document MediaWiki. The domain name itself makes it fairly ill-fit to document our non-MediaWiki software documentation.
Wikitech, till recently, was a fishbowl wiki that required an operations-team member to create an account for anyone wanting to edit. Since being merged with labsconsole it has open registration and its authentication is integrated with a number of our developer and operations tools.
All contributors must become at least partially familiar with wikitech anyway, since it's the registration location for commit access. That, along with its more generic name, better suit it for hosting our documentation for non-MediaWiki products.
The core of MediaWiki is in my mind still radical and exciting: you make or find a page, click edit, and just type into it. This seems to have gotten buried over the years under a pile of bad ideas and bad implementations, and skewered from the outside by MegaTronEditKillerLaserBot2000 and the like. The solution is not to pile additional layers on top, but to excavate MediaWiki from underneath them and make it fast as hell and simple, so that it once again feels like a dangerous, underspecified, liberating idea. Semantic this-and-that and ontologies of categories entails putting up rails everywhere and signs with arrows on them indicating which way you should go. That approach will inevitably end up reflecting a narrow, inflexible view of what a wiki is and what you do with it. And all the while MediaWiki will creak and groan from underneath.
The biggest freedom a wiki provides is the ability to create your own structures and to change and modify those structures quickly and easily. Adding semantics gives you more freedom to create more interesting structures, especially if this is combined with proper templating.
One example of how semantics could improve mediawiki.org is the extension matrix https://www.mediawiki.org/wiki/Extension_Matrix. That list is maintained by a bot. Its listing is completely inflexible. The bot is required because MediaWiki has no mechanism for handling this. Adding semantics to the extension pages would make the matrix a simple query. We could also add some statistics about extensions. It wouldn't require anything from the editor's side as the semantics are hidden away via templates. Editing extension pages could also be considerably easier, since the extension page could have a proper form for the templates. We will hopefully be able to use wikidata for things like this in the future.
Spend some time editing a well designed Semantically enabled wiki. Web Platform is a good example: http://docs.webplatform.org/wiki/Main_Page. There's a high degree of structure there. That wiki is way above average quality from the point of view of a reader specifically because it enables the editor to easily make the content consistent.
I think the level of structure on webplatform doesn't well suit something like wikitech or mediawiki.org, but adding some extra structure or at least adding semantics to existing structure will be very useful.
- Ryan