Chris Ball, a Mad bio-savvy artisan, and Wade Brainerd all spent part of the past two weeks getting a disk-conserving wikireader onto the XO that supports browsing and simple searching over a 100-fold compressed set of articles.
The result :
- a 100M activity containing most of the Spanish Wikipedia, with illustrations, math fontification, and templates
- scripts that support generating a new version from the latest articles, from heuristics defining the most popular titles, with only a few hours of work
There is also a short blacklist of pages and images that need improvement which will change over time. A whitelist of unpopular but crucial pages will surely build up, and the process will find a way to learn from the subject-specific wikireader efforts to produce smaller uncompressed collections. The same idea and scripts can provide a roughly Britannica-sized collection for every major language; or a multilingual cover of the 200 smallest languages; expect an English one soon for comparison.
While this reader (which has to unzip each page as it is requested) is slower than browsing html, it is still a pleasure to use. The real lack, shared with other readers to date, is that comments and editing don’t yet work…