Friday, March 23, 2007

Citizendium, Wikipedia and Tinfinger

I've set up a CompInt tab in my NetVibes to track blogs of those founders who I see as having similar goals to me. Pride of place goes to Larry Sanger's blog about Citizendium. The more I work on Tinfinger the more I am appreciating how the 5-second elevator pitch I have been using - "Tinfinger will do to the Who's Who what Wikipedia did to the Encyclopedia Britannica" - means that we will have to learn a lot from the Wikipedia experience.

Citizendium has already made many mistakes in trying to break free from the Wikipedia model while still retaining many of its advantages. Larry's latest post, entitled We Aren't Wikipedia, is the starkest illustration yet of what I think is his biggest error: he is continually defining Citizendium in terms of comparisons to Wikipedia. In January he and the CZ community made a fine decision to unfork, which should have been the impetus to delete all references to Wikipedia not only from the content pages, but also in the minds of the contributors. Yet Larry spent many post-unfork blog posts detailing his criticisms of the Essjay scandal, and he never lets an opportunity pass to bag the big W for some ethical slight or another, as if he thinks that is what motivates the CZ base. It may well be what motivates them, but if so, then that's not the basis for a healthy community IMO. You want people who are in your community for the community's sake, not just to spite some other project.

As for Tinfinger, we only share Citizendium's similarities with Wikipedia on items 5, 7 and 8 in Larry's list. We do share more differences, though: all but item 7 fit us to some extent. More details on how all of this works closer to the full launch.

The reason I have got so far into this debate is that Tinfinger is going to adopt Wikipedia's data structure: unique identifier strings for pages, which are cross-linked via category, type and property metatags for which pages are automatically generated. We have downloaded the relevant data from of the roughly 60,000 profiles of people in Wikipedia. Instead of being stored in text-based infoboxes, however, we will store the tags as relational data in rows of MySQL tables to allow greater granularity in searches. However, like Citizendium, we will not republish any Wikipedia articles.

I don't know what that means in terms of licensing - the Wikipedia:Copyrights page doesn't talk about this issue - which is part of why I'm stating this in public. Perhaps there are lawyers well-versed in the GDFL who can tell me whether taking just the titles and metadata of these profile articles, not any of the prose text, means that the GDFL still applies. I want to act in good faith here, so I need some help.


Blogger OllyOnions said...


9:12 am, March 24, 2007  

Post a Comment

<< Home