I like bicycles

COMM 450 blog entry #8

This paper discusses two experiments that were run to assess aspects
of technologically mediated communication in goal-oriented tasks. The
variables that were controlled dealt with the shared media of two
participants, one of whom was a naive worker and the other of whom was
an "expert" instruction-giver.

Predictions were, essentially, that measured success at the tasks
would increase along a continuum of shared communication space ranging
from least successful with half-duplex shared audio to most successful
with person-to-person contact in a shared physical space. Between
these two extremes existed full-duplex shared audio and shared audio
and video wherein the expert could see a portion of the field of view
of the worker.

Surprisingly, there was little significant variation in success
between using full-duplex audio and video. Here I suggest a few
reasons for that which were not considered in the article, based
around the actions that an instruction-giver may take in a shared
space that may not be taken otherwise.

The article uses the opportunity to make physical gestures indicating
specific objects in a shared space in order to ground discussion of
tasks to be performed and the lack of an equivalent opportunity in a
remote shared-media environment as one reason that a side-by-side
environment is more effective. The availability of a shared visual
environment in the shared video setup helped by allowing deictic
references to items in the shared visual space. I suggest also that
it is easier to partake in interactive dialogue when the
instruction-giver has more than passive control over their field of
view. The experimenters have suggested that the limited field of view
of the video apparatus may have restricted its usefulness; I suggest
also that the fact that the instruction-giver was merely "along for
the ride" and not able to direct their field of view independently of
the worker was significant in the dialogue between instruction-giver
and worker.

One way of testing this would be to mount the video camera that the
worker wears in such a fashion that its field of view may be directed
remotely by the instruction-giver. This, coupled with the ability of
the worker to see the video image that the instruction-giver sees,
takes some of the burden off of the worker to position themself so
that the instruction-giver can see what they are talking about. It
also would provide some feedback to the worker as to what specifically
the instruction-giver is concerned with at a given moment; this would
permit a part of the shared physical environment circumstances to be
duplicated in the remote shared video environment.

Eye-tracking devices similar to those found in camcorders could also
be employed to focus the instruction-giver's camera (which would be
controlled by what the instruction-giver is looking at) or to "box" on
the instruction-giver's viewscreen what the worker is looking at and
to mark in a similar fashion on the worker's HUD what the
instruction-giver is looking at. This is, indeed, suggested as
important for design of such a system in the future within the paper.

This would introduce to the shared video environment further aspects
of some of the nonverbal, unspoken communication that is available to
those who are in a shared physical environment.

A further issue is that of the familiarity of participants with the
video equipment and its limitations. Greater facility at providing
good video input to the instruction-giver might be attained by giving
workers and instruction-givers a chance to experience the experimental
setup from each others' sides as a part of a warm-up exercise. This
could help to defeat some of the asymmetry inherent in the remote
location of the instruction-giver.

Another suggestion I have would be to set up a similar scenario in a
fully virtual environment. It is common in multi-player first-person
computer games to allow players who are out of play to follow those
who are in play using different over-the-shoulder, through-the-eyes,
or roaming cameras. The issue of whether limitations in the video
feed were at fault for a lesser measure of success in the shared video
environment could be addressed by creating a task for a worker to
carry out in a virtual environment, such as running through a maze and
performing tasks at different points throughout it, with an "expert"
to help. Limiting the points of view which would be available to the
expert could provide further edification as to whether different
vantages on another's activities are more or less conducive to giving
instructions in completing a task.