Wikimedia Research Network/Meetings/2005-06-18

The second meeting of the Wikimedia Research Network was held on June 18, at 20:00 UTC on the channels #wikimedia-research, #research1 and #research2 on irc.freenode.net. The meeting lasted almost 3 hours. A full log is available at [1].

Erik reminded all interested individuals to help with the Proceedings of Wikimania, written at http://en.wikibooks.org/wiki/Wikimania05. After that, we decided to split the meeting into two parts, held in parallel in separate channels:

1. about the software needs of the Wikimedia projects (moderated by Erik)
2. about research into the content and communities (moderated by David)

This proved to be a reasonably efficient way to go into details without worrying about scaring away people more interested in other aspects of research, so a similar split will likely be used in future meetings. However, a more flexible approach to the agenda may also be a good idea; meetings could generally start with brief introductions where everyone can say what their favorite topic is. This could then be used to finalize the agenda, since typically, it is not edited much while it is still on the wiki.

Commitments and interests

People have committed to the following:

• Rob Lanphier: GUI mock-ups for single login (will have to wait for finalization of process)
• James Forrester:
  • e-mail notification text for single login (will also have to wait a bit)
  • perhaps coordinate translation of said text at least for all languages with more than 10K articles
  • rough wishlist of features for a next generation category system
• Erik Möller:
  • finalize single login specs
  • organize Wikibooks community meeting
  • research Wikiversity (late)
• David Gerard: invite Stirling Newberry to join
• Brion:
  • single login ;-)
  • improve database schema and/or code so that user name lookups are done on a single table
• Ilya: write proposal for server-side data client
• Erik Zachte: write proposal for survey and post to foundation-l etc.

People have expressed interest in helping with:

• Odile: educational metadata for Wikiversity
• K1v1n: surveys, possibly Wikimedia COTW

Technical meeting

Participation

The following people made comments in this discussion.

High participation (more than 50 lines): Erik Möller Brion Vibber James Forrester David Gerard

Medium participation (between 10 and 50 lines): Rob Lanphier Ilya Haykinson Odile B. Magnus Manske

Low participation (less than 10 lines): Austin Hair Ævar Arnfjörð Bjarmason Dariusz Siedlecki Kim Bruning Nikerabbit Tomasz G. Sienicki

Results

Single login

Together with Brion Vibber, MediaWiki release manager and Chief Technical Officer of the Foundation, we made an attempt to reach a consensus on the procedure for migrating authentication on the Wikimedia wikis to a single login system. "In the context of Wikimedia, this refers to a single, persistent account and session for all projects and all languages," Erik explained.

Brion has committed to working on single login after the successful deployment of MediaWiki 1.5 which, including post-deployment bug fixes, should be finished at some point in July or August.

The current draft specifications can be found at Single login specifications. This will be edited to reflect the discussion, which may already have happened by the time you read this.

The discussed model is different from Brion's original proposal insofar as local user accounts will stop working after the migration date. Because of this, the aspect that was discussed in most detail was the best strategy for migrating accounts to a global database if there are different users sharing the same username.

The presently agreed upon strategy can be summarized as follows: For any set of accounts sharing the same name, subsets that have email addresses associated with them are identified. These subsets are compared to each other. The subset with the highest number of total edits is the one which will get privileged access to the username in question. The username will be "locked" in the global user database.

The associated email address will receive a notification stating that the account can be migrated to the global login system, and that the user can log in using the account password of his account with the highest number of total edits. Only a user supplying this password will be allowed to log in. If they successfully do so, all accounts sharing this password will be migrated automatically. Further accounts can be migrated manually by supplying the password for them. (The scenario of the email address not being valid needs to be resolved.)

Due to the importance of email addresses, users should be informed before the migration about the need to set a valid email address at least in their most used account in order to still be able to merge it into their global account.

Users without an email address associated with their account are not part of this privileged resolution process. They can, however, still register accounts on a first come, first served basis unless they are locked as part of the aforementioned process. Any registered account in the global system can, during the transition period of 6 months, hook up old user accounts by supplying the password for them.

Hooking up a user account in this scenario means changing its attribution from the old account to the new global one. How this will be accomplished in terms of the row changes in the database has not been finalized yet; however, Brion has committed to changing the code in such a way that user names in a MediaWiki installation do not have to be redundantly stored across multiple tables, which will make changes much easier.

An important requirement is that it will have to always be possible to reset the password of a local account, so that users who have forgotten their passwords can still access and hook up these accounts.

The global account system will rely on the MediaWiki authentication plugin, and not on the existing code support for a single user database. Only minimal information will reside in the global account database, while most user information will continue to be stored locally in all the wikis a user edits. Erik voiced the desire to have a global set of preferences that can be used as the default prefs in case no local preferences are set.

Wiki syntax changes

Lee's proposed changes to the MediaWiki syntax were discussed only briefly, as Lee was not present. Brion stated that he prefers gradual improvements to the existing syntax and worries about the transition pain. Erik suggested that the existing proposal could be split into components, which could then be treated as individual development tasks.

Magnus and others proposed an XML "Esperanto" between wikis that could exist alongside a new syntax. Erik noted that, due to the different features in various wiki engines, such a schema would require long term commitment by a developer or group of developers to get off the ground and be usable. He believes that it should not be a priority for targeted development in terms of allocation of funds by the Wikimedia Foundation.

David made the good point that any new syntax has to be tested with non-techie users before it is deployed.

Other

Erik briefly described the evaluation work that needs to be done as a first step towards the Wikiversity/Wikisophia project. Odile B. volunteered to help with the standards part of the evaluation, particularly IEEE.LOM for metadata.

Erik also reported that the first project community meeting will involve the Wikibooks community and their technical needs. He will organize this meeting soon. David suggested contacting Theresa Knott for help.

Ilya, who is highly involved with Wikinews, joined us to discuss the needs of that project. He noted that there was still no sufficient solution for listing the most recent pages in a set of categories, or for providing an RSS feed of the same nature. He also explained his vision of a server-side tool that would fetch certain data, like weather and market data, and import it into a database, to be made available for an extension that could then render this data to the user. He committed to writing a proposal for this project, which, as he pointed out, could be useful in a variety of scenarios outside Wikinews itself.

The weaknesses of the existing category system were briefly discussed (also on the second channel), and several desirable changes, such as the ability to query multiple categories with Boolean operators, or to display a category navigator on article pages, were mentioned. Erik asked James to formulate a wishlist of the most needed features to work on a next generation design.

As a high priority task that needs to be worked on soon and that might be eligible for funding, improving the handling of licensing data for materials on the Wikimedia Commons was identified. Such data should be passed on transparently to the using projects, which is also important for using projects. Erik also mentioned the idea of "InstantCommons", the instant availability of any file on the Commons to any public MediaWiki installation. He stated that, due to the bandwidth needs of such a feature, its very idea should be discussed with the Board and the community.

Content / community meeting

Participation

High participation (more than 50 lines): David Gerard Erik Zachte

Medium participation (between 10 and 50 lines): Erik Möller k1v1n Magnus Manske

Low participation (less than 10 lines): Ævar Arnfjörð Bjarmason Kim Bruning Tomasz G. Sienicki

Results

Magnus Manske joined us to discuss his article validation feature. David briefly went through his wishlist for the feature, primarily the easy editability of all aspects of the survey. However, due to the fact that the feature is not likely to be deployed as part of MediaWiki 1.5 for technical reasons that were revealed after the meeting, this discussion is only briefly summarized here.

Erik Zachte introduced his idea of a community-wide survey on questions such as (quote verbatim):

1. Is syntax becoming too technical and unwieldy (e.g. wiki tables) or is functionality and flexibility valued above easy of use?
2. Are search facilities satisfactory?
3. Is it difficult to keep up with new developments (technically and organisationally) More precise: Do we have too many forums, like goings-on on meta, many mailing lists, village pumps, etc. If so, what would help to improve the situation?
4. Are there issues that keep you from contributing more? Like response time, edit conflicts, etc?
5. Would you be willing to tell more about yourself on your user page (country of residence, age, sex, education, etc) provided no one else can see this, it is optional and only used for demographic statistics ?

Such a survey could guide and influence the development process and the research network. He wants to work on a formal proposal for this project, but also would like assistance with the technical implementation, especially if the survey component is meant to be part of MediaWiki (if so, should it be delayed until the rollout of single login?).

Erik Möller briefly summarized the status of Wikimedia COTW. He stated that the first project was likely far too large an undertaking, and that a "Collaboration of the Month" might make more sense in general. David suggested that one way to get more people interested would be to ask the topic submitters for new collaborations to help with the organization of the project itself. K1v1n expressed some interest in the project.

On a final note, Erik Zachte made the interesting observation that "every conceivable user error occurs at least 100 times on [the English Wikipedia]." Hopefully the Research Team can help to identify the most common errors, and eliminate them by addressing the causes in collaboration with the developers and the community. (Note EZ: Project en:Wikipedia:WikiProject_Wiki_Syntax proves there is huge interest in syntax validation, see list of contributors down the page)