This paper asks how much the strength of network effects depends on the stability and structure of the underlying social network. I answer this using extensive micro-data on all potential adopters of a firm's internal video-messaging system and their subsequent video-messaging. This firm's New York office had to be relocated due to the terrorist attacks of 2001 which lead to a physical re-organization of teams in that city but not in other comparable cities. I study the consequences of this disruption for adoption of video-messaging and the size of network effects. I find evidence that generally network effects are based on direct social interactions. Potential adopters react to adoption only by people they wish to communicate with: They are not affected by adoption by other people. However, when there is a disruption to the social network and communication patterns become less predictable, users become more responsive to adoption by a broader group of users.