Wikimedia Polska, in cooperation with Helion SA, has published the Polish Wikipedia on DVD. The publication strictly follows the GNU FDL license and is available in shops all over Poland for a price of 39 PLN.
The profit from the DVD sales will be distributed among Wikimedia Polska, Wikimedia Foundation and Helion SA.
We believe that the resulting DVD will be a new method of spreading the free knowledge, complementary, but not competitive with the mother Wikipedia. As a side effect, our work should improve the quality of Wikipedia too, as we are going to return the checked articles to the parent project, as long as they don't break the rules of Wikipedia.
Preparation of DVD
DVD content was prepared by a group of 13 paid Wikipedia contributors, supported by about 20 volunteers. Their task was to check all articles on a separate wiki with a Wikipedia dump from 4 June 2006.
We had created a fork of pl-Wikipedia Wiki with an access restricted to registered users only. The 13 people, chosen both by Polish Wikimedia Chapter and Helion, were paid by Helion to clean up the 130 000 articles. We were supposed to complete the job in 100 days, so every paid contributor was expected to check 100 articles per day. Their task was to remove all obvious mistakes, to remove obvious bias, and to correct spelling, grammar, and formatting of the articles.
After 50 days, on 2006-09-03, it turned out that the quota of 100 articles to check per day per person was too large. We were around ten days behind the schedule, but this was partly due to the holiday season and a few problems with the server (see the chart). In autumn 2006 the work was speeded up, and it was finished on December 2006, only several days after the initial deadline.
Polish Wikipedia had 250 000+ articles on 4 June 2006, but 120 000 of them did not require the check-up, as:
- they were bot-generated, or
- the only contributors were people from so called "white-list" of trusted Wikipedia contributors – don't ask how we composed this list, as we had decided to keep it secret for the good of the community.
The work was organised in this way:
- The articles had been divided into 225 packs, sorted by categories
- The contributors were picking the articles they wanted to check by removing them from lists and moving them to their own subpages
- Every day a bot was calculating the number of articles picked and checked, and the statistics were sent to all the contributors.
- We monitored the quality of the contributors' work, and if their work was not good enough, we have sent them (by individual e-mails) suggestions, how their work could be possibly improved.
The final product preparation
The final product contains all the selected and checked articles with scaled down illustrations. It was prepared from the wiki database as a set of HTML files – one file per article + HTML files for image descriptions. It contains 239 000 articles and 59 000 pictures.
The main user interface is written by Olaf Matyja, contracted by Helion. It is a standalone Java application, working as a search engine, which transfers resulting lists of articles and articles themselves to the user's default web browser. Therefore, the DVD works on any operating system on which one can install Java Runtime Environment (JRE) ver. 1.5 or higher. It is also possible to use the DVD without Java – just by browsing the alphabetical list of articles or categories.
The final product preparation was finished on 20 January 2007. The final DVD ISO file has about 8 GB. The date of the first official release of the DVD was set at the end of April 2007, but it was finally released in the end of July 2007 .