Learning Points from Wikimania 2015

25 July 2015

This piece of writing is licensed under CC BY-SA 3.0 and a similar post is also available in Bahasa Indonesia at Indonesian Wikipedia

I attended Wikimania 2015 conference in Mexico. There are some learning points that I want to highlight and share with Indonesian Wikipedia community:

  • Using data from Wikidata at Infoboxes: You can easily use it using Parser Functions-like syntax or using Lua. Example of a template getting data from Wikidata when parameters is not supplied can be seen at https://fr.wikipedia.org/wiki/Mod%C3%A8le:Infobox_Fromage
  • Charts in Wikipedia: Vega.js: You can embed a chart using vega.js syntax in Wikipedia. https://www.mediawiki.org/wiki/Extension:Graph
  • Tool Labs Directory: Hay’s directory (http://tools.wmflabs.org/hay/directory/#/) Tool directory with descriptions, source code location, tags.
  • Usability of Wikimedia: Evidence that Wikitext editor has usability problem (e.g. uploading image). I laughed but that laughter is painful. So, do usability testing!
  • #100wikidays: like #100happydays movement, but creating 100 articles in 100 days; personal challenge.
  • Video in Wikipedia: Uploading video will be easier (planned to use video.wikimedia.org currently is at wikimedia.meltvideo.org); auto-transcoded to acceptable format.
  • The Wikipedia Library: Collection of open resources for each Wikipedia; good to have one :)
  • Wikimedia India: 1-to-1 edit training (someone request to be trained; an editor go to that person and teaches them in person; as a result number of active contributors has increased a lot, according to them).

And there are some major ones that I want to highlight:

Wikimedia Asia Forum


Josh (Wikimedia Philipines) felt that Wikimedia Movement can be split into around 6 regions: North America, Central and South America, West Europe, Central and Eastern Europe, Africa, and Asia (Australia included here). Also, he felt that Foundation treat Asians the same as what they treat Americans and Europeans but the conditions in Asia are different hence it should have some other treatment (e.g. Wiki Loves Monument was a great success in Americas and Europe, but not really successful in Asia).

Besides that, there is no one representing Global South at the recently elected WMF Board of Trustees. This may be the case that votes from Asian countries are really low; hence what we can do in the next election is to promote more voters OR to gather the Wikimedians to support one candidate from Global South (but we’re not sure of doing this since we felt ethically incorrect).

I asked around some other Wikimedians and some of them (Thailand, India) doing photo contest (Wiki Loves Earth, Wiki Loves Food), could we consider this? Taweetham from Thailand said that there is a problem though since the photo contributors only contribute photo and not article and what he did for this year is to require the contestants to write article alongside contributing photos.

The forum results in doing idea from Addis (Wikimedia User Group China) called Wikipedia Asian Month <https://meta.wikimedia.org/wiki/Wikipedia_Asian_Month> in November 2015 to create articles about Asian countries topic but not your country and will be sent postcards from participating countries. Well, for Indonesia, I need communicate with Wikimedia Indonesia on this.

Hackathon, Revision scoring as a service, ORES, and Raun

So on this Wikimania, I joined their Wikimania Hackathon and I was paired up with Aaron Halfaker, a research scientist from WMF doing research on Revscoring project.

What I did was basically help him extend the revscoring project to support Indonesian language: I found a list of stop words; I added the list of Indonesian bad words (with help of User:Ladsgroup who gives us the list of top 250 words that got reverted in idwiki); he found a Indonesian language dictionary (from aspell library); we also found an Indonesian language stemmer in ntlk style but unfortunately was not properly documented with a license. But that was enough he said, he started doing the work to generate the model but unfortunately he said that the prediction model was really bad and he did not know why (he said to me that he may look into FlaggedRevs “reverts” since currently he did not do that)

The other thing that I did was to integrate ORES (the online service that provides revscoring score) with Raun! Now on some projects (en, pt, tr, fa, and es?), Raun will show a fire icon showing the probability of that edit being vandalism. Halfaker’s project was so cool and I really hope that it can be integrated with idwiki (and its sister project) really soon as we’re really lacking of admins.

Effect of blocking IP editing: Evidence from Wikia


I joined this talk because it reminded me on https://id.wikipedia.org/wiki/WP:NOANON

The talk focuses on what happens when a wiki blocks editing from anonymous users. They analyzed before and after data from Wikia wikis that disables anon edits.

The results?

  • Damage:
    • Words removed: declined ~70%
    • Revered edits: declined ~55%
  • Quality:
    • Unreverted edits: declined ~30%
    • Persistent word revisions: declined ~60%
  • New editors:
    • Not really a big change

Their conclusion of whether this is bad or good is depending on how expensive it is to patrol and revert those anonymous edits.

So, what I’m proposing here is to create a counter proposal for WP:NOANON, i.e. to let anonymous users create new articles on Indonesian Wikipedia because

  • As pointed out by Ricky Setiawan of Wiki-ID, we have Flagged Revs! At the time of WP:ANON proposal, well FlaggedRevs proposal is approved by community but went unimplemented till 2010. So might as well use this feature to handle anonymous users.
  • Research on giving more insight on revision scoring is ongoing. (i.e. automatically determining if one edit is damaging or not) [See below]
  • Number of admins have increased to 26 since introduction of WP:NOANON (it was ~15 at time of approval) although only little of them are actively patrolling every edits.
  • VisualEditor is now deployed, there should be very minimum “wikitext test” vandals.
  • Indonesian Wikipedia already have rollbackers that help rollback edits quickly

Current situation:

  • Vandalism is not only from anonymous users, some of the users have developed to a point where they can insert false information and went undetected.
  • Workaround model from anon users since they can’t create new article: they create that page at the talk page and is easily deleted due to WP:CSD#G8.
  • No meaningful number of registration of new users observed [stats.wikimedia]
  • Articles per day? Roughly still the same before and after.

I hereby propose Indonesian Wikipedia to turn anonymous article creation back on and to mitigate the effects of more vandalism coming, we should:

  • Increase number of admin [ongoing]
  • Push Revscoring project to support idwiki
  • Turn on “half-admin” rights (“eliminators”, implemented in vi.wiki) so that one can help admins to do admin jobs in fighting vandalism (including deletion of articles) without having other admin rights (like editing global JS and CSS, etc)


  • More vandal articles from anon users
  • More non-vandal articles from anon users: This is what we hope to achieve by unblocking the article creation


Note: This piece is finished on end-July 2015 but I forgot published it until September 2015.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.