Collaboration can take many forms, for which technology has long provided digital support. But when collaborators are located remotely, to what extent does technology support these activities? In this dissertation, I argue that the success of a telecommunications system does not depend on its capacity to imitate co-located conditions, but in its ability to support the collaborative practices that emerge from the specific characteristics of the technology. I explore this using wall-sized displays as a collaborative technology. I started by observing collaborators perform their daily work at a distance using prototypes. I then conducted experiments and found that people can accurately interpret remote deictic instructions and direct gaze when performed by a remote collaborator through video, even when this video is not placed directly in front of the observer. Based on these findings, I built CamRay, a telecommunication system that uses an array of cameras to capture users' faces as they physically navigate data on a wall-sized display, and presents this video in a remote display on top of existing content. I propose two ways of displaying video: Follow-Local, where the video feed of the remote collaborator follows the local user, and Follow-Remote, where it follows the remote user. I find that Follow-Remote preserves the spatial relations between the remote speaker and the content, supporting pointing gestures, while Follow-Local enables virtual face-to-face conversations, supporting representational gestures. Finally, I summarize these findings to inform the design of future systems for remote collaboration across wall-sized displays.