What data is enough to identify a unique person, and why?

It also shows clearly why it is possible to learn many things about a person (for instance, sexual orientation) from a few "what do you like" questions.

In regard to the case of trying to learn about movie preference associations from crowdsourcing, identifying a person from a data base that lists which movies each person has watched seems to be possible only if you know some set of movies the target person has watched. But I think that if you don't have a way of getting that starting data about a person, it is impossible to identify anyone.