Even in a double-blind review, it is possible to figure out whose paper you are reviewing if you are familiar with their work. The system can also be gamed through collusion or sucking up (approving a paper because they reference a paper of yours or of a colleague).