User:JHernandez (WMF)/Wikimania 2011 - Mobile web talk notes
I watched the talk from Wikimania 2011 about mobile wikipedia to understand the past and compiled some notes.
A lot of interesting content in there, I recommend watching the talk.
The notes are written from the context of 2011, so take any information in here as if it came from 2011 unless stated otherwise.
The notes are written in order of appearance in the talk.
Year is 2011. Mobile is going to be biggest growth area in the next years. Wikimedia's two main goals are going to be editor retention and mobile -as Sue Gardner said-.
Estimations from market research say there will be 2 billion mobile web subscribers in 2015.
We estimate that by 2014 mobile web will overtake desktop web (page views) on all types of web properties on the internet.
A lot of research being done.
2011 stats for comparison:
- 20% of visits to Facebook from mobile
- 0.03% of visits to Wikipedia from mobile site
First mobile experience, WAP Wikipedia
- Built around 2006 / 2007
- WMF had no bandwidth to create a mobile gateway
- Community provided this experience
Current mobile web: MERB
- Built with Ruby by a contractor
- No PHP, can't benefit form foundation expertise
- Many things needed from core need to be implemented, localization (i18n) for example
- Since the content has to be accessed from the MediaWiki core cluster, there is an extra HTTP request. So it can't leverage the full site caching layer, which means it is not as efficient as it could be
- Completely different setup for the servers. It has a smaller number of servers available to serve requests
How it works
- Accesses the content from the core cluster, brings it back (via HTTP request)
- Scrapes the page, rips out different elements that aren't appropriate for mobile rendering
- Has some caching mechanisms via memcached
Wikimedia Mobile App
- Lives in github
- Features geolocation
- Translated via translatewiki too
- Over 150.000 downloads per month
- Talks to the ruby gateway (MERB)
- One way experience, can't edit
- Wikimedia made APIs available and external actors created mobile experiences
A lot of organic traffic coming in. Between 2-15% growth each month is being observed, which we believe is underestimated because of missing redirects.
Future mobile strategy (more robust)
Strategic goals: Keep up with mobile growth
June of 2012 Should have around 2 billion page views on mobile. And we need more growth.
Mobile site rewritten in PHP
Rewrite MERB to PHP as a MediaWiki extension.
- Just use the output buffer capability from core
- Remove the need for an extra HTTP request
- Then remove things from the content
- Easier use for MediaWiki users, just install extension, should work out of the box
- Could be expensive, using a DOM parser
- Done profiling, seems performant enough
- Use a device detection database (WURFL) for detecting mobile devices capabilities and redirecting more devices
Q: Why not build a skin
A: We thought to build one, but an extension gives us more flexibility for reformatting or rewiring the UI as needed for a better experience on mobile devices. We talked to Brion Vibber and others about it and it seems appropriate
Building a bigger mobile ecosystem
Using PHP allows us to leverage community developers to help.
We need to get partners and provide resources for other experiences (Android, Symbian, Augmented reality, ...) from their developer networks.
What we're learning (research)
- Research projects ongoing (qualitative)
- India, Brazil, US
- After that, ethnographic studies and quantitative surveys
Find out what's scalable across languages/projects that we should do.
- By the end of 2012, 68% will access internet via mobile
- More than 30% will only interact with the web on mobile
Current mobile web
Decided on the caching layer (squid) as the place to add ACLs for redirecting. From the WURFL DB of devices we want to redirect many more to it.
Also looking into other features like using web fonts. Siebrand community member is helping with that.
Research results and findings
Getting results from the research now. Some findingsː
- An example of persona, 20 year old from India, looks to most content in punjabi
- Articles from his place are better in punjabi than in hindi. We are thinking about disambiguation and transliteration of content across languages
- Interoperability across devices for reading (desktop & mobile). People want print functionality, but are oblivious of is existence. Offline consumption and sharing are very important
- There are no sharing tools or anything similar in any Wikipedias
- People want/expect precise and concise summaries on mobile, not too much information. People want summaries not full articles in mobile
- The context, devices, change the information that people want to get
- Updated content: People want richer experiences, more pictures, not only tables and theory
- People say wikipedia looks like site of the 1990s. We need to think about this
- Autocomplete needs to search in all languages, not only english. People in other countries don't speak english and expect autocomplete to work because it is very useful
- Mobile has changed user behavior
- We focus on being as open as possible, but you can't change user behavior
- We need to use native closed tools to integrate with the OS that people use and how they use them
- People don't access on their phones because it uses their phone credit
- We may have to focus on partnerships, zero rating
Other platforms research say that people on mobile contribute more than people on desktop, but the quality is usually lower.
For example, Yelp doesn't allow writing reviews from mobile because of quality issues.
We need to ask, how are people going to need to contribute in mobile? (block level editing, picture uploading...)
We need to use the community for testing, there are too many devices out there. Please test and send screenshots.
We're making interface improvements. With the WURFL database we're going to make improvements for the devices and the UIs.
We'll be looking to mobile editing, and uploading photos from mobile.
Maybe partner with other orgs like Flickr for CC uploads and then import in commons, etc.
Apps, we have iOS app, we need presence in other OS, like Symbian, Windows mobile. Going to change the iOS app for a Phonegap app so that we can distribute for all OS. We'll be talking to the creators of Phonegap for that.
Going to change the iOS app for a Phonegap app so that we can distribute for all OS. We'll be talking to the creators of Phonegap for that.
We'll be talking to the creators of Phonegap for that.
Qː Why are apps needed?
People want and have apps and we need to be there. We need to be practical. We are going to focus on mobile web and phone gap and not develop native apps all across the board, we will leave that to third parties.
Also Phonegap wants to disappear and provide web apis as the web platform improves, so in the future that should bring us closer to the web from native platforms.
Future plans (continued)
Research results are coming and will bring more information.
We will create a wiki page with mobile features to talk about and vote for.
We will reach to operators -as many as possible- to expand reach with partners and device makers. Eventually lead to Wikipedia on all devices.
In the future:
- Get wikipedia for free (zero) on partners
- Donation campaigns in mobile
- Special tools for other partners, specific to local communities etc.
- Maybe content partnerships
Who are we
Team: Amit, Tomasz, Mani, Patrick, Kul, Parul
Wikipage for the mobile project, #wikimedia-mobile and mobile-l, etc.
We need to make sure mobile is on everyones radar.
Languages and fonts
Siebrand is looking into it, there will be a languages team on the foundation
Work with partners for getting things like speech recognition, offline, tools for people with disabilities, ask for community support.
Not only technical stuff, there needs to be a lot of awareness and marketing efforts
What would you want to see on mobile
- Offline content on my phone
- Full article editing
- Access to other projects
- Will work with the extension
- Article feedback for mobile
- Watchlist and revisions/history
- Offline wiktionary
With editing the biggest problem is the UX on mobile. Full page editing is hard and doesn't make sense
There is block level editing, or voice editing, it will need to be a different experience. For example select a sentence and just edit it.
For example select a sentence and just edit it.
Concurrency may be a problem since people may take longer to edit on mobile. So how to present conflicts or which UI to edit through is important. Maybe add markers to see when other people are editing.
Markers to see when other people are editing.
Maybe collaborative editing at some point.