#8 — option 2

The Kraut et al. paper revealed some interesting characteristics about visual co-presence, particularly as it operates as part of and separate from physical co-presence. Their results that indicated that duplex video resulted in communication no more efficient than audio-only, but still less efficient than FTF, indicates that there is something in physical co-presence that all existing implementations of visual co-presence are missing. Otherwise, the extra linguistic turns in visual co-present media to establish common ground would not be needed. Their own observations also indicate that it may merely be an issue of finding the right implementation, such as systems that indicate data about focus or angle and context, rather than trying to approach the efficiency of FTF by just throwing more video at people, since their experiments and the experiments of others have shown that doesn't work. The problem with their video system (and other video systems) seems to be that people don't actually use visual information directly: they translate visual perceptions into implicit statements (e.g. "So-and-so is looking at the whoosit") which can be much harder to deduce from video feeds than physical co-presence (likely because of issues of depth, resolution, etc.), and so technological communication solutions should try and reproduce the types of data people try and deduce from their visual perceptions more than the visual perceptions themselves.

Since the video system developed in the experiments appears to have not been a successful solution, what might be a more successful design? The conclusion of the paper lists some enticing suggestions and non-video oriented aspects of what might be successful, but how could these be synthesized into a real design? What would be a good balance of raw video data and data distilled from video or other input to present to users? How could such a system be robust enough to be useful, but still usable?