The process is quite simple.
You link an image from Facebook, such as a "like" button. Every time someone views your page, that button is accessed, and your domain is the referrer. Facebook then sees the IP of the person viewing that image, and that referrer, and can potentially correlate that easily with the known IPs of Facebook users. Volla, Facebook knows every site you have been to.
This is not some kind of mythical concept. This was actually used to cause a brief hack scare on a site I used to administer.
That is not even getting into CSS hacks, which are still, effectively, open, and allow any site to access your browser history.
It's also pretty classic you're accusing me of being a tinfoil hat, considering you spent several hours going on about the evils of cookies and how this law is needed to protect your privacy, but that's ok, it's obvious you're a troll. I post this for the benefit of others, as I realize my original post was not totally clear.