File:Thiemo Mättig.jpg
Thiemo Mättig (WMDE)

Hi, my name is Thiemo Mättig and I work as a software developer and engener at Wikimedia Deutschland in Berlin. My main project is Wikidata, as well as tools and user scripts that empower all Wikimedia communities. This account is solely used for the editing of Wikimedia projects and on behalf of Wikimedia Deutschland.

Thiemo Mättig
Wikimedia Deutschland e.V.
Email
thiemo.maettig@wikimedia.de
IRC
#wikidata Thiemo_WMDE
Twitter
@maettig

Talk to me about

  • Wikidata
  • German Wikipedia community
  • Code quality, good code, software architecture and code reviews
  • Tools, gadgets and user scripts
  • Indie game development
  • The power of regular expressions
  • Wikimedia Deutschland
  • Berlin

My personal schedule

Tuesday, 15 July

  • 19:00 - Arrived at the airport after being in the air for about 12 hours.
  • 21:00 - Meetup in the hotel bar.

Wednesday, 15 July: Hackathon Day 1

  • Met Reuben from wikiHow. Explained Wikidata, it's concepts and limitations.
  • Ladsgroup and Legoktm asked what a bot framework should do? Always call wbeditentity, even if only a label is edited? Or prefer the more specific API modules? We agreed that a bot, if it can, should always prefer the more specific API modules. The same time the wbeditentity API module should try to forward simple requests to the more specific modules. I will check if we have a ticket for that or create one.
  • Found my buddy Viswaprabha. Talked about Malayalam, how images, language and text is used to transfer knowledge and how text will become obsolete in the near future. What role Wikidata will play in a future of concepts, with no text.
  • 14:00-15:00 - Meeting: Wikidata - Find a project, ask your question, Don Genaro.
  • Dug into the rabbit hole T85340 again. Figured that MediaWiki core's Status class is just broken.
  • Realized that the HTTPS switch broke my local repository. I can't edit sitelinks any more. Still don't know how to fix.
  • Learned about neural networks in revision classification and spam fighting.

Thursday, 16 July: Hackathon Day 2

  • Learned about a project to use the friendly space policy on content.
  • Talked with User:Mxn about OpenStreetMap and how Wikidata and OSM should link to each other. OSM does not have stable identifiers, that's why a way property probably is not a good idea. But OSM should link to Wikidata a lot more.
  • Said hello to User:Amire80 and User:Pginer-WMF.
  • Talked with User:Tgr about the new "vertical" way the WMF sections are split, how "mobile first" redesign is much more fun than trying to work on the desktop design, the Vector redesign and why a sans-serif font stack should end with serif (because Japanese). He is also working on a logging extension that can log JavaScript errors in a nice way.
  • Talked about Wiki Loves Monuments, what defines a good photo and how different photos in the contest are from encyclopedia illustrations.
  • 18:00 - Welcome reception.
  • Finally met Denny.

Friday, 17 July: Hackathon Day 3

  • 10:30-12:30 - My shift at the Wikimedia Deutschland booth, talk to me about tech and Wikidata.
  • 16:00-16:30 - Visually editing Wikipedia: Why VisualEditor is different, Don Diego.
  • 16:30-17:30 - Can Conflicts of Interest (COI) be aligned with the Wikimedia project? Don Americo.
  • 18:00-22:00 - Essential Mexico City Downtown walking tour, Lobby.

Saturday, 18 July

Sunday, 19 July

  • 09:00-09:30 - Making Infoboxes easier to edit and maintain with the help of Wikidata, Don Alberto.

Monday, 20 July

  • Pyramid tour?

Tuesday, 21 July

  • 20:35 - Flight back to Germany starts.

Notes

Future of structured documents: VisualEditor, Citations and Wikidata, oh my!

Brion Vibber talks. I missed the first slides. Want to use the same system across media types (PDF, spoken Wikipedia, videos). Shows an example of a pronunciation audio with an IPA pronunciation string that should have a source. But Wikidata references are currently confusing. Shows an <ref>{{harvcoltxt|...}}</ref> template that basically can't be used by humans. Suggests to start with putting stubs in Wikidata. "Here is a bit of text that's referenced by this bit of text. I don't know how to encode these details but that's what I have for now."

Questions start:

  • Project "source metadata".
  • Wikisource author. Deep linking sources, link to a specific sentence in a Wikisource page.
  • User asks for tools or if new tools will come? Suggest a parser function to wrap referenced text snippets. Thinks it's nicer to link to the beginning of the sentence. But that's easy. The hard part is creating the reference. "Let the Wikidata devs deal with that. We will use what they come up with."
  • User explains Wikipedia articles can show Wikidata references. Brian prefers references as their own items instead.
  • Question about adressability of sentences and paragraphs. Suggestion to automatically have this in the DOM. Answer is that it will change constantly, hard to have stable identifiers basically. Also very ugly in Wikitext.
  • An example of a template that got deprecated by a deeply confused Wikipedia community, because they could not understand where the data comes from.
  • Mobile team, alternative ways of presenting articles. Likes the idea of creating stuff from Wikidata because it's so much easier to target on different devices and such.
  • User brings up a use case of fuzzy ankering of fragments of content across revisions. Brian asks for link.

User(s) Talk(ing): The future of wiki discussions

Danny Horn talks. Great presentation of the current state how to start a deletion discussion. Almost a rap.

  • We possibly reached a platoe of the number of people that are really interested in the topic.
  • Flow mostly copies the process on large English discussion pages that are split over subpages and transcluded in a big board overview. My personal experience: Almost no other wiki does this!
  • User shouts: We already have this via scripts. Why are you fixing something that's not broken? Answer is that what the Flow team is targeting new users. "We are building the next generation." He basically says what the Flow team creates will fit the need of all communities.
  • They tried to understand the existing workflows, split it in elements, identify common elements to let the community recreate the workflows they are used to via the new elements. They call it a "workflow pattern builder". Shows raw mocks of a form builder, looks like a Wikitext wizard to me. I see a gazillion nitpicks about each detail of this proposed "builder" pop up. It tries to be suitable for every use case but in doing this it restricts everybody to a few most common cases.
  • Shows features that are all marked as "do not exist". First is remix. Next is search. Awesome filter options. They all do not exist. Just awesome looking mocks.
  • Mock shows one of these "intelligent" input boxes one can not edit any more, instead of a simple "user:username" string.
  • User says: I remember Wikimania Hong Kong. Lots of exciting mockups, nothing happened and now not even talked about any more. "You are biting off way more that you can chew."
  • User says: "Clearly a degree of hostality associated to this. The other being VisualEditor." Asks for expertise from VisualEditor. "Please learn from that."