I want a way to track all my clicks when I read the Wikipedia, and ideally your clicks too. Who looked at what article, when and from where they clicked in and out.
Think of all the cool things that should be possible with that data. Figuring out which articles had strong connections to one another across multiple degrees of separation; from different perspectives based aggregate user types. I’m not sure what some of the results would be but I’d love to be able to find out.
- Show me car related articles viewed by people who have viewed lots of geography related articles. Shows cars with available widely worldwide? Cars people who travel widely are interested in? Something anyway.
- Show me music related articles viewed by people who have read a lot of philosophy articles; or cooking articles; or both; or cooking but not philosophy.
- Show me the most common paths coming and going from an article.
- Suggest articles I might want to read based on users with similar reading habits.
- Show me any pages related to a given set of articles, identify users with similar patterns and make suggestions.
- Identify pages with small but persistent relationships so I can figure out the story.
I tried to make a web based gizmo to get collect data well pulling Wikipedia pages on the client side but phishing and XSS attacks have made it so browsers won’t allow that kind of thing no matter how sneaky I got. XMLRPC doesn’t allow remote server requests and once you go cross server in a frame or iframe even the location attribute is off limits to other windows. It’s technically possible but the browser just won’t let me do it. Annoying.
An extension for Firefox would do the trick but it would need a killer app to motivate people. People don’t like having their browsing history tracked; even if Wikipedia logs it already at least that isn’t public.
Adding the functionality directly to the Wikipedia software itself would be the best way. Track logged in users who opt-in, cookie track anonymous users, then give me the sweet sweet data.