Understanding user behavior in user generated-contend systems (UGCs) is a fundamental building block for maximizing usefulness of these systems. For example, the importance of users' connections and the degree of iteration is needed in order to build meaningful recommendation systems. In this paper, we evaluate the impact of users' network on marking photos as favorites in Flickr considering various levels of distance between users in the network topology. We have observed that users' first-degree contacts are clearly the biggest contributors to the evaluation of hers/his photos (i.e., marking them as favorites). This result indicates that, in contrast to other UGCs such as Twitter and Digg, contact relations are a good indication of content interests in Flickr.