Wikimedia Developer Summit/2017/Media, Visualizations, and Layout
Session Overview
[edit]
- Title: Media, Visualizations, and Layout
- Day & Time: 2:50 pm – 4:00 pm (22:50 – 00:00 UTC)
- Room:Ventana
- Phabricator Task Link: T151952
- Facilitator(s):
- Note-Taker(s):
- Remote Moderator:
- Advocate:
Slides and Video
[edit]Slides are at Media,_Visualizations,_and_Layout.pdf on commons.
Video from this session is at Media, Visualizations, and Layout.webm on commons.
Session Summary
[edit]- Chronology: [Capture the gist of who said what, in what order. A transcript isn't necessary, but it's useful to capture the important points made by speakers as they happen]
- TheDJ: Story builder
It's just an idea, a thought experiment. It's a human thing to tell stories, it's real communication, not just "regurgitating facts".
Not every country is blessed with written history. You can upload an audio file to Commons, but then what?
MediaWiki was written for "pros' world", without thinking of (then-)future media formats.
Example from Washington Post: integration of text, video, blurbs, ads—UI concepts that we aren't using.
- Brion: Video and other media support
We should finish existing work on audio and video support, which got stalled after 2014 RfC on MP4
Revisit or work around it? I don't have strong opinion about it (although I have invested quite a bit of work in the "work around" option)
Mobile media playback is more in flux
have everything for OGV to work on iOS Safari (uses a bit more battery than it should, but works fine)
but mobile apps need more work
we use a format that almost nobody produces content in, and don't offer conversion tools [outside Labs]
VisualEditor doesn't currently have a way to set the playback beginning time, to show a preview, to let the user search for existing media conveniently, etc. There's no search by technical characteristic, such as resolution or size.
Transcoding
two problems that make them overflow: 1. not enough servers for peak upload times
2. serious bugs in job queue with backlog/ large files
C Scott: Syntax
We don't quite have enough syntax for semantic layout. We make guesses, for example, show the first image on mobile, but it's very weak.
Example idea for semantic syntax: "[[File:Name.jpg|type=illustration]]
Localization could work through (future) Shadow namespaces ( https://phabricator.wikimedia.org/T91162 )
Yuri: I dream of content
Google paid people to build a great search engine in existing content
Facebook paid people to build a platform to post cats, and reward them with likes
We built a platform that lets some intellectually-inclined people to write stuff
But what we did is not interactive enough. We don't have something like a "Talking globe"
To build interactive stuff, you need the data, and the tool that converts it to something visual.
We now have a tool that draws a map—e.g. takes data about the party of each U.S. state's governor, and shows a colored map. Or a map that shows how much does each state contribute to GDP each year.
We have data about women's salaries, but it becomes useful only when shown as a chart. Same for weather in New York—it would be too hard to edit table cells by color manually, but with the current chart code it is done automatically.
Or a star map, that is auto-translated to each language.
Discussion
[edit]CScott: Brion, which part are you going to do first?
Brion: mobile web, which requires finishing desktop
Q: There's a big project called OpenStreetMap. It sounds like a natural map.
Yuri: We are doing it already. It's a perfect synergy. A year and a half ago we started the Maps project. Some of the examples I showed are based on OSM. We have a service that is usable for third parties, and the bigges third-party traffic is from chinese-language pokemon site.
TheDJ: we do a lot more than maps, e.g. enhance OSM content via Graphs extension
Yuri: OSM is a base layer. You need to tell a story on top of that.
Q: What is your take on Cargo extension and Semantic MediaWiki?
C. Scott: I'm not familiar without, see the next session.
multi-content revisions; some of these [content types?] might be frameworks
Yuri: Data sources are important for visualizations.
already support pageview API, mediawiki API - easy to integrate more
CScott: concern from Parsing team: better separation
Andy: How to integrate it into Wikipedia?
TheDJ: Why Wikipedia? Maybe an encyclopedia is not really appropriate. Maybe another site will complement Wikipedia in this regard.
C. Scott: The Nutcracker article is all text. There could be a video synchronized with text. And what is an article about the Great Wall of China without actually seeing the wall?
Yuri: A lot of people complained that my tool is too complex. It's indeed complex because it allows doing a lot of things.
Tilman: (to Yuri) Does it work on mobile web and apps? (about half of our traffic now)
Yuri: No effort till now. Graph extension was mostly a volunteer project. Raising issues upstream (Vega). Events like taps will be supported if you as [graph] designer support them. We are thinking about showing stuff as full-screen, rotating, etc.
Yuri: Data sets exist separately from pages, and one article can combine them.
Votes
[edit]Story builder: 2+
Video: 5+
Syntax: 5+
Visualizations: almost everybody!