Hardly surprising, but I thought this article [link dead 2024] (see the archived version) by Stewart Meagher summarized one researcher's findings about Facebook's tracking methods and explained the methodology so well that it was worth posting a link to it.

By now everyone and their Grandmother knows that Facebook is building a huge database on its users which will eventually be used to market to them more effectively, but its hardly common knowledge that the Facebook Connect system now embedded in nearly every high traffic web site on the internet provides the means whereby Facebook can keep tabs on what you're surfing even if you never sign up for an account.

This history can even be paired up later with an account if you choose to set one up in the future. You might wonder how this is different, if at all from what Yahoo or Google or Microsoft have been doing for years as you visit their web search portals or for that matter the logs that are kept by the ISP(s) that provide your connection to the internet.

Content creators (e-zines, media and news outlets, etc.) that choose to add the facebook connect APIs to their sites are providing a means for Facebook to build a database even on people that don't even use their service, storing it in the hope they will one day identify that person, and for users of Facebook, adding to that flat statistical data layers of information about a user's social connections and their self-professed likes and dislikes on any and all topics (as well as providing a way to get statistics on how popular those opinions are in their social groups).