A nice way to take back some control over unintential browser history leakage:

More and more websites use like-buttons from Facebook, Google+ and Twitter. However, these buttons send information to these social networks even if the user doesn’t click them, but even if they are just present on a webpage. This way these networks are able to track which websites users are visiting and are able to build fairly complete browser histories of their users. Because this is neither what a user might expect nor what many website operators that embed like-buttons want, this alternative way of using these social services was developed.

Though one wonders why this sort of feature isn’t built-in to browsers to begin with.