Have it installed for a few weeks now, it fails for many domains. If you click a link from example.com to example.org, it will already mark example.org as 'potentially tracking you on example.com' (because it sent the referer of example.com to example.org <i>once</i>).<p>Also it reports nearly every domain I visited in the past weeks to be tracked by lesswrong.com, which is rather impossible.<p>Lastly, it gets cluttered very fast. Not within 5 minutes, but a day later no user could make sense of it anymore. There should also be a clear overview with things like top domains (which would be Facebook for their Like button, Google ads and analytics, and probably more).
There's a very impressive demo at <a href="http://www.mozilla.org/en-US/collusion/demo/" rel="nofollow">http://www.mozilla.org/en-US/collusion/demo/</a> too.<p>One thing I didn't understand from the diagram in the article is that some grey circles are connected to each other : Is that from that that the browser sees, or is it linkages that are known to exist (outside of the raw data)?
The article implies that Mozilla is on our side, and wouldn't sell the user to marketing firms. But if you visit mozilla.org, collusion reveals that Mozilla (potentially) pass your information to webtrendslive.com.