Offline/Offline Wikimedia projects
Modified from an email from Erik Moeller dated Sept 8, 2009
The desire to get Wikipedia into other formats than "as a website" was expressed pretty early in the project's history. There are a couple of reasons why people have cared about this early on:
- prestige - it's nice to see your work in other formats, and to make it more tangible;
- reach - it helps to reach audiences who either can't or don't want to use the Internet;
- convenience - it's simply nice to have for those times when you aren't online, as rare as they increasingly are for people in developed countries.
There were some early projects both to develop print and offline editions. The most notable of these were the early German WikiReader projects, including a commercial print-run by a German publisher, and early Wikipedia editions on CD and DVD.
Some historical background on these: http://de.wikipedia.org/wiki/Wikipedia:WikiReader (first issue March 2004) http://de.wikipedia.org/wiki/Wikipedia:WikiPress (first issue December 2005) http://de.wikipedia.org/wiki/Wikipedia:DVD (first issue September 2004) http://it.wikipedia.org/wiki/Wikipedia:DVD (first issue September 2007)
Interestingly the early English attempts to create print editions or DVD copies never got much traction: http://en.wikipedia.org/wiki/Wikipedia:WikiReader
Erik Zachte, as a volunteer, lovingly created one of the first "Wikipedia in your pocket" editions, which is often forgotten in this early history: http://www.infodisiac.com/Wikipedia/
It had a niche of passionate users but was for-pay and depended on the proprietary "TomeRaider" application and never reached mass distribution.
The most widely distributed English language DVD, to my knowledge, is the selection of articles created by SOS Children, first in 2006, and a more recent issue in 2008: http://www.soschildrensvillages.org.uk/charity-news/education-cd.htm http://www.soschildrensvillages.org.uk/charity-news/2008-wikipedia-for-schools.htm
The English Wikipedia 1.0 Team has been working with WikiProjects since 2006 to organise and assess articles. This scheme is also used to assist groups such SOS Children's Villages and One Laptop Per Child in their offline selections. In 2007, the team put together a manual selection, released as a small test CD of 2000 articles, called Version 0.5. This was followed in 2010 by another test version containing 31,000 articles, Version 0.7, which used automated article selection but manual vandalism checks.
Finally, there are numerous smaller local efforts that we at best have ambient awareness of; for example, SchoolNet Namibia has had copies of Wikipedia installed on their school computers since at least 2005, and the Wizzy Digital Courier project is literally based on a guy on a motorcycle bringing Wikipedia to schools in South Africa: http://www.linuxhomenetworking.com/forums/showthread.php?t=15687 http://www.wizzy.org.za/
Most of these efforts happened in a completely bottom-up manner, with occasionally tiny bits of funding and promotion from the Wikimedia Foundation. This is possible because we've published downloadable copies of all Wikipedia articles for as long as I remember, through: http://download.wikipedia.org/
More recent efforts
More recently, there have been attempts to build standardized Wikipedia offline distributions that can work on multiple languages and platforms.
See w:Wikipedia:Version 1.0 Editorial Team/FAQs for more information.
1) Storage format:
- There's significant support for the openZIM project as an effort to standardize on a single compressed storage format for offline readers. We've given our backing to the project and will aim to make downloadable copies of Wikipedia available in this format.
- Sugar Labs developed Info Slicer: a way for content to be taken from Wikipedia and put into an offline format in order to be edited, packaged, and distributed as teaching materials.
2) Readers: There are multiple notable open source software projects to create a standardized "reader" software which can read arbitrary content packages:
- Okawix (derived from Kiwix)
- MoulinWiki (merged into Kiwix in 2009)
- WikiBrowse (OLPC)
- Evopedia, which evolved from the Openmoko platform community. Currently it supports Maemo, Ubuntu and SHR (Openmoko). Recent content is available in various languages in Evopedia's own storage format. Evopedia can use geo information in articles to diplay them on a map.
3) Article collections: While openZIM is designed so that anyone may have create a compilation of articles, distinct projects exist in a subset of languages to pull a high quality, typically education-focused subset of articles that would be appropriate for dissemination in education environments. Offline language projects include:
4) Mobile: There are a number of closed source packages for different mobile platforms, which are often sold at a fee (without an agreement with us - this is legal under our license):
5) Print: A partnership with PediaPress.com has made it possible to build your own printed books from Wikipedia articles. This is in production use on our largest Wikipedias and has been used to ship more than 1,000 custom made books already. We've also endorsed a print edition of an abstract summary style version of the German Wikipedia, compiled by Bertelsmann. However, that was more of a one-off project, and we mostly supported it because Wikimedia Germany was excited about doing it and brought it to us pretty late. We have no immediate plans for any officially blessed print copies beyond the PediaPress print-on-demand stuff.
6) Computer software:
- The One Laptop Per Child project has built hand-crafted selections of Wikipedia content, some of which are now shipped in production to thousands of kids. They've also created some multimedia packages from Wikimedia Commons, which have been well-received where they have been tried out.
- eGranary Digital Library delivers millions of educational resources to computer users in developing countries whom lack consistent and/or inexpensive access to Internet. Wikipedia is one such resource.
7) External Devices:
- WikiReader is a standalone device that provides access to Wikipedia offline.
- The Humane Reader is a device that can be connected to a TV and content can be accessed from there. An external keyboard is used to navigate.
- Cooliris released an iPad app called Discover which is essentially a Wikipedia magazine: it re-packages the feature article on a daily basis. Originally it was not available for offline use, but now it allows the user to view offline any article they had previously viewed.
8) Example pilot projects
- The collection of Wikipedia articles labeled Version 0.7 was released at the beginning of 2010, containing around 31,000 articles. Both Kiwix and Okawix versions were made available in a variety of offline formats, and used for example in schools in South Africa.
- Wikimedia Kenya is in the beginning stages of developing a plan to place an offline version of Wikipedia on computers in schools' computer labs across the country. They are planning on using the openZIM/Kiwix platform.
We're very carefully examining opportunities in this space primarily because of our commitment to reach people who aren't online. At the end of the day, we're looking for models that can scale to millions of potential new readers. The steps to this assessment are being documented on the Offline page.
1) The USB stick version you've seen was a proof-of-concept built by Kul Wadhwa together with the WikiPock folks (who are local to the SF Bay Area); we're exploring whether to build a long-term partnership with Kingston or another storage device maker around this, i.e., large-scale distribution of Wikipedia on USB sticks.
2) There are some efforts to build dedicated or at least optimized Wikipedia reader devices; we're in touch with a few groups working in that area, and may lend our branding and support to an effort that seems innovative and likely to succeed.
3) We may want to explore building a strategic relationship with a major telecommunications company to bundle Wikipedia directly with cell phones that are widely used in developing countries.
4) As we find successful offline distribution models, we'll want to determine whether there are ways in which people with at least occasional connectivity can usefully contribute - annotating articles, taking photos, drawing illustrations, translating, creating spoken versions, etc.
5) Finally, another key area to think about is the content selection and assessment. Martin Walker's "Wikipedia 1.0" effort has done a lot of work in this area, and we're exploring the use of technologies like FlaggedRevisions and WikiTrust to ensure that the versions of Wikipedia articles that we distribute offline approximate the best available recent revision of articles, so that readers aren't exposed to vandalism and as little inaccurate information as possible. Adapted WikiTrust code was used successfully in 2010 for putting together the 50,000-article English test release, Version 0.8, which is expected to be available for Kiwix, Okawix at the end of 2010. Releases are planned in Windows, Mac, Linux and mobile formats, available on USB-stick, memory card, or for download directly or via BitTorrent. A Kiwix Sugar version for OLPC use is planned. The first English "official" release, Version 1.0, is being planned for 2011.