One problem is that they collect "shadow" profiles of people who aren't even on the service. If someone who has your contact info is on the service, they'll start a profile of you based on that. Plus, there's many web sites that are connected to Facebook and send info back to it without you even knowing about it, just by having a Facebook share button on the page. It's very difficult to avoid having any of your data on Facebook's servers.
Maybe the real problem here is that people talk about "your data". There's no "your data" in a shadow profile of you. There's data about you. I have a feeling that talking about the data as if it belonged to the person it refers to makes people reach weird conclusions.
I'm having problems with articulating clear examples, but I don't like the very concept of ownership of facts - whether they're algorithms or phone<->name associations.
In my mind, it's one thing if a Web site that I go to learnt about me based on my interactions and a whole another thing when people start correlating data across sites. That's what Facebook is doing with its stupid buttons on nearly every news site.
Amazon knows a great deal about my buying habits but that makes complete sense. Facebook knowing what articles I read from the Globe and Mail and the CBC is something else.
