NOT BEING THERE
Miller Puckette
Center for Research in
Computing and the Arts
University of California
San Diego
The term "Network Performance" covers a wide range of
possible situations, with such differing promises and
difficulties that it is difficult to make generalizations.
Instead, I'll talk about the specific possibilities and
limitations of possible specific approaches. The
following three examples of possible networked artistic
encounters don't cover all the territory but at least serve
as examples:

In scenario number one, two or more concert halls
schedule a simultaneous event, each with its own local
audience and performers. Each concert hall arranges to
project images and sound from the remote performers;
their images show up on one or more screens arranged
behind the local performers, so that the audience sees
one or more remote stages as well as the local one, all
from more or less the usual viewing angle. This is often
more expensive than simply bringing the remote
performers to the local performance spaces, each in turn,
would be, because the additional complexities in the
setup are likely to outweigh the travel costs they offset.
On the positive side, since it's costing a fortune anyway,
it might be worth it to pay for an upgraded network
connection that can guarantee a low latency of
communication. Another problem is the geometry of the
situation, since the local and remote performers can't
easily make eye contact (although at least each can be
given a video monitor showing the other from the
audience's perspective). Finally, if more than three time
zones separate locales in which there will be an
audience, it can get very difficult to schedule an
appropriate time to hold the performance.

In scenario number two, a concert (or any other event)
takes place in a single location, real or not. But the
audience, or at least some of it, is “telepresent”, getting
video and audio feeds from the event, and perhaps can
also send information back to the event. This is a good
way to save money and fuel (most of which is usually
spent getting the audience to the show). Latency is not
much of an issue (since we could probably tolerate a few
tenths of a second of delay in audience feedback, so we
can just let the audience hear everything a fraction of a
second late and nobody will be the wiser). On the other
hand, there's no way to monitor what quality of sound
and picture the remote audience hears and sees over their
home systems. Also, members of the audience don't hear
and feel each other's presence except indirectly; the
psychology of the audience as a group is missing.

Yet a third scenario uses networking only for
experimentation and rehearsal, presumably in
preparation for one or more old-fashioned physical
performances afterward, for which the musicians will
have to travel to get to a fixed locale. Here, since we're
likely to try to contain costs for a rehearsal in a way we
might not for a performance, we will probably not end
up with access to special low-latency networking
infrastructures (that will always be boutique stuff), and
so latencies will be noticeable and may sometimes
change capriciously. On the positive side, we gain
tremendous flexibility in scheduling, since the
performers can presumably work from locations close to
wherever they happen to be. Time zone differences are
less problematic than in Scenario one, since there are
only performers to deal with, and performers are more
able to operate at strange hours than audiences are.

These examples show that there are wide differences in
artistic and practical needs from one networked music
performance to another. No one way of doing it will
satisfy everyone's needs; and careful thought will always
be needed in designing a networked encounter for any
specific situation.

My own early experiences with networked
performances were very much in the first mold above:
performers in different sites interacting with each other
in front of local audiences. In the years 1997-1999,
Rand Steiger, Vibeke Sorensen, Mark Danks and I
formed a group we called Global Visual Music. We did a
series of performances involving music improvisation
and live computer graphics images, one of which was
networked between two sites (Miller Theater in New
York, and Intel headquarters outside Portland, Oregon;
the performers were Steven Schick, Anthony Davis,
Vanessa Tomlinson, and Scott Walton).

We quickly decided that instead of doing
“telepresence” (which had already been well explored by
that date), we would try to do something that actually
explored the idea of the physical distance between the
sites. Instead of sending sounds and images across the
network, we opted to send real-time audio analyses of
the instruments and for each of the four instruments to
create, on the other side, a possibly variable 'ghost' for
the instrument. In the case of the pianos, we used MIDI
interfaces that could either be used to play each other's
physical piano remotely or to play synthetic sounds. For
the percussion we did an instrument-by-instrument
analysis (this is the genesis of Pd's bonk~ object) and